{
  "best_metric": 2.2021772861480713,
  "best_model_checkpoint": "final_models/laft_lug_mpt/checkpoint-4530",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 9060,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006622516556291391,
      "grad_norm": 25.366834045369046,
      "learning_rate": 1.5e-07,
      "loss": 4.8438,
      "step": 1
    },
    {
      "epoch": 0.0013245033112582781,
      "grad_norm": 24.103572996230607,
      "learning_rate": 3e-07,
      "loss": 4.75,
      "step": 2
    },
    {
      "epoch": 0.001986754966887417,
      "grad_norm": 31.403042933752992,
      "learning_rate": 4.5e-07,
      "loss": 4.7812,
      "step": 3
    },
    {
      "epoch": 0.0026490066225165563,
      "grad_norm": 28.349556331713593,
      "learning_rate": 6e-07,
      "loss": 4.8438,
      "step": 4
    },
    {
      "epoch": 0.0033112582781456954,
      "grad_norm": 47.522014401524785,
      "learning_rate": 7.499999999999999e-07,
      "loss": 4.9062,
      "step": 5
    },
    {
      "epoch": 0.003973509933774834,
      "grad_norm": 33.42440788025143,
      "learning_rate": 9e-07,
      "loss": 4.875,
      "step": 6
    },
    {
      "epoch": 0.004635761589403974,
      "grad_norm": 37.37771966002289,
      "learning_rate": 1.05e-06,
      "loss": 4.75,
      "step": 7
    },
    {
      "epoch": 0.005298013245033113,
      "grad_norm": 28.49686603636974,
      "learning_rate": 1.2e-06,
      "loss": 4.625,
      "step": 8
    },
    {
      "epoch": 0.005960264900662252,
      "grad_norm": 38.58555547468358,
      "learning_rate": 1.3499999999999998e-06,
      "loss": 4.625,
      "step": 9
    },
    {
      "epoch": 0.006622516556291391,
      "grad_norm": 15.5788294149647,
      "learning_rate": 1.4999999999999998e-06,
      "loss": 4.3125,
      "step": 10
    },
    {
      "epoch": 0.00728476821192053,
      "grad_norm": 13.539642570093081,
      "learning_rate": 1.6499999999999997e-06,
      "loss": 4.2188,
      "step": 11
    },
    {
      "epoch": 0.007947019867549669,
      "grad_norm": 16.45997685007064,
      "learning_rate": 1.8e-06,
      "loss": 4.125,
      "step": 12
    },
    {
      "epoch": 0.008609271523178808,
      "grad_norm": 10.338870374620004,
      "learning_rate": 1.9499999999999995e-06,
      "loss": 4.0312,
      "step": 13
    },
    {
      "epoch": 0.009271523178807948,
      "grad_norm": 12.531078700007233,
      "learning_rate": 2.1e-06,
      "loss": 4.1562,
      "step": 14
    },
    {
      "epoch": 0.009933774834437087,
      "grad_norm": 12.085342249480263,
      "learning_rate": 2.2499999999999996e-06,
      "loss": 4.3125,
      "step": 15
    },
    {
      "epoch": 0.010596026490066225,
      "grad_norm": 15.697423559496999,
      "learning_rate": 2.4e-06,
      "loss": 4.25,
      "step": 16
    },
    {
      "epoch": 0.011258278145695364,
      "grad_norm": 11.03089343361638,
      "learning_rate": 2.55e-06,
      "loss": 4.0625,
      "step": 17
    },
    {
      "epoch": 0.011920529801324504,
      "grad_norm": 11.783898396824728,
      "learning_rate": 2.6999999999999996e-06,
      "loss": 4.1562,
      "step": 18
    },
    {
      "epoch": 0.012582781456953643,
      "grad_norm": 14.035038701029494,
      "learning_rate": 2.85e-06,
      "loss": 3.875,
      "step": 19
    },
    {
      "epoch": 0.013245033112582781,
      "grad_norm": 181.286768740991,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 4.1562,
      "step": 20
    },
    {
      "epoch": 0.01390728476821192,
      "grad_norm": 28.84619958472304,
      "learning_rate": 3.15e-06,
      "loss": 4.0625,
      "step": 21
    },
    {
      "epoch": 0.01456953642384106,
      "grad_norm": 9.485673104253173,
      "learning_rate": 3.2999999999999993e-06,
      "loss": 4.0625,
      "step": 22
    },
    {
      "epoch": 0.015231788079470199,
      "grad_norm": 9.570221121660918,
      "learning_rate": 3.4499999999999996e-06,
      "loss": 3.9688,
      "step": 23
    },
    {
      "epoch": 0.015894039735099338,
      "grad_norm": 9.685835428032023,
      "learning_rate": 3.6e-06,
      "loss": 4.0,
      "step": 24
    },
    {
      "epoch": 0.016556291390728478,
      "grad_norm": 8.965141745792456,
      "learning_rate": 3.7499999999999997e-06,
      "loss": 4.0625,
      "step": 25
    },
    {
      "epoch": 0.017218543046357615,
      "grad_norm": 8.443756038943226,
      "learning_rate": 3.899999999999999e-06,
      "loss": 4.0312,
      "step": 26
    },
    {
      "epoch": 0.017880794701986755,
      "grad_norm": 9.00900750414927,
      "learning_rate": 4.049999999999999e-06,
      "loss": 3.9531,
      "step": 27
    },
    {
      "epoch": 0.018543046357615896,
      "grad_norm": 21.43724193727691,
      "learning_rate": 4.2e-06,
      "loss": 3.9375,
      "step": 28
    },
    {
      "epoch": 0.019205298013245033,
      "grad_norm": 8.00759282451024,
      "learning_rate": 4.35e-06,
      "loss": 4.0625,
      "step": 29
    },
    {
      "epoch": 0.019867549668874173,
      "grad_norm": 8.335236533716525,
      "learning_rate": 4.499999999999999e-06,
      "loss": 4.0312,
      "step": 30
    },
    {
      "epoch": 0.02052980132450331,
      "grad_norm": 8.523181382739859,
      "learning_rate": 4.6499999999999995e-06,
      "loss": 3.7344,
      "step": 31
    },
    {
      "epoch": 0.02119205298013245,
      "grad_norm": 7.335662695749709,
      "learning_rate": 4.8e-06,
      "loss": 3.9531,
      "step": 32
    },
    {
      "epoch": 0.02185430463576159,
      "grad_norm": 14.642056640956355,
      "learning_rate": 4.95e-06,
      "loss": 3.875,
      "step": 33
    },
    {
      "epoch": 0.022516556291390728,
      "grad_norm": 7.226751530548763,
      "learning_rate": 5.1e-06,
      "loss": 3.625,
      "step": 34
    },
    {
      "epoch": 0.023178807947019868,
      "grad_norm": 6.931934901699227,
      "learning_rate": 5.25e-06,
      "loss": 3.8125,
      "step": 35
    },
    {
      "epoch": 0.02384105960264901,
      "grad_norm": 6.961482163524123,
      "learning_rate": 5.399999999999999e-06,
      "loss": 3.7969,
      "step": 36
    },
    {
      "epoch": 0.024503311258278145,
      "grad_norm": 7.166952548626372,
      "learning_rate": 5.549999999999999e-06,
      "loss": 3.7344,
      "step": 37
    },
    {
      "epoch": 0.025165562913907286,
      "grad_norm": 7.5011039990677855,
      "learning_rate": 5.7e-06,
      "loss": 3.8594,
      "step": 38
    },
    {
      "epoch": 0.025827814569536423,
      "grad_norm": 6.979537062109198,
      "learning_rate": 5.85e-06,
      "loss": 3.625,
      "step": 39
    },
    {
      "epoch": 0.026490066225165563,
      "grad_norm": 7.501463225997184,
      "learning_rate": 5.999999999999999e-06,
      "loss": 3.7969,
      "step": 40
    },
    {
      "epoch": 0.027152317880794703,
      "grad_norm": 7.128116057728067,
      "learning_rate": 6.1499999999999996e-06,
      "loss": 3.875,
      "step": 41
    },
    {
      "epoch": 0.02781456953642384,
      "grad_norm": 7.450186004966472,
      "learning_rate": 6.3e-06,
      "loss": 3.3906,
      "step": 42
    },
    {
      "epoch": 0.02847682119205298,
      "grad_norm": 7.262599008700824,
      "learning_rate": 6.449999999999999e-06,
      "loss": 3.7344,
      "step": 43
    },
    {
      "epoch": 0.02913907284768212,
      "grad_norm": 7.157354371834633,
      "learning_rate": 6.599999999999999e-06,
      "loss": 3.5469,
      "step": 44
    },
    {
      "epoch": 0.029801324503311258,
      "grad_norm": 6.529833935351724,
      "learning_rate": 6.749999999999999e-06,
      "loss": 3.3594,
      "step": 45
    },
    {
      "epoch": 0.030463576158940398,
      "grad_norm": 7.9320289419469185,
      "learning_rate": 6.899999999999999e-06,
      "loss": 3.8594,
      "step": 46
    },
    {
      "epoch": 0.031125827814569535,
      "grad_norm": 7.060377804922717,
      "learning_rate": 7.049999999999999e-06,
      "loss": 3.625,
      "step": 47
    },
    {
      "epoch": 0.031788079470198675,
      "grad_norm": 6.891224904152876,
      "learning_rate": 7.2e-06,
      "loss": 3.7656,
      "step": 48
    },
    {
      "epoch": 0.03245033112582781,
      "grad_norm": 9.294009042390158,
      "learning_rate": 7.35e-06,
      "loss": 3.7344,
      "step": 49
    },
    {
      "epoch": 0.033112582781456956,
      "grad_norm": 7.084400003040845,
      "learning_rate": 7.499999999999999e-06,
      "loss": 3.625,
      "step": 50
    },
    {
      "epoch": 0.03377483443708609,
      "grad_norm": 6.4800219931444945,
      "learning_rate": 7.65e-06,
      "loss": 3.6562,
      "step": 51
    },
    {
      "epoch": 0.03443708609271523,
      "grad_norm": 7.035922127455148,
      "learning_rate": 7.799999999999998e-06,
      "loss": 3.7656,
      "step": 52
    },
    {
      "epoch": 0.035099337748344374,
      "grad_norm": 6.278399860408964,
      "learning_rate": 7.949999999999998e-06,
      "loss": 3.2031,
      "step": 53
    },
    {
      "epoch": 0.03576158940397351,
      "grad_norm": 6.355358791009081,
      "learning_rate": 8.099999999999999e-06,
      "loss": 3.4844,
      "step": 54
    },
    {
      "epoch": 0.03642384105960265,
      "grad_norm": 7.164881703796917,
      "learning_rate": 8.249999999999999e-06,
      "loss": 3.3281,
      "step": 55
    },
    {
      "epoch": 0.03708609271523179,
      "grad_norm": 47.03805644062284,
      "learning_rate": 8.4e-06,
      "loss": 3.5625,
      "step": 56
    },
    {
      "epoch": 0.03774834437086093,
      "grad_norm": 7.518813612641354,
      "learning_rate": 8.55e-06,
      "loss": 3.7188,
      "step": 57
    },
    {
      "epoch": 0.038410596026490065,
      "grad_norm": 6.307641293758115,
      "learning_rate": 8.7e-06,
      "loss": 3.0,
      "step": 58
    },
    {
      "epoch": 0.0390728476821192,
      "grad_norm": 6.101815811381738,
      "learning_rate": 8.849999999999998e-06,
      "loss": 3.5,
      "step": 59
    },
    {
      "epoch": 0.039735099337748346,
      "grad_norm": 6.121500036003077,
      "learning_rate": 8.999999999999999e-06,
      "loss": 3.1406,
      "step": 60
    },
    {
      "epoch": 0.04039735099337748,
      "grad_norm": 6.322691623459317,
      "learning_rate": 9.149999999999999e-06,
      "loss": 3.6562,
      "step": 61
    },
    {
      "epoch": 0.04105960264900662,
      "grad_norm": 8.550403796972356,
      "learning_rate": 9.299999999999999e-06,
      "loss": 3.6094,
      "step": 62
    },
    {
      "epoch": 0.041721854304635764,
      "grad_norm": 5.881021205952493,
      "learning_rate": 9.45e-06,
      "loss": 3.4531,
      "step": 63
    },
    {
      "epoch": 0.0423841059602649,
      "grad_norm": 6.343767768578617,
      "learning_rate": 9.6e-06,
      "loss": 3.4375,
      "step": 64
    },
    {
      "epoch": 0.04304635761589404,
      "grad_norm": 5.468997291403601,
      "learning_rate": 9.75e-06,
      "loss": 3.3594,
      "step": 65
    },
    {
      "epoch": 0.04370860927152318,
      "grad_norm": 13.21834323159589,
      "learning_rate": 9.9e-06,
      "loss": 3.6406,
      "step": 66
    },
    {
      "epoch": 0.04437086092715232,
      "grad_norm": 5.751675099558176,
      "learning_rate": 1.005e-05,
      "loss": 3.6875,
      "step": 67
    },
    {
      "epoch": 0.045033112582781455,
      "grad_norm": 5.789432849160812,
      "learning_rate": 1.02e-05,
      "loss": 3.5,
      "step": 68
    },
    {
      "epoch": 0.0456953642384106,
      "grad_norm": 9.22798294268364,
      "learning_rate": 1.035e-05,
      "loss": 3.5312,
      "step": 69
    },
    {
      "epoch": 0.046357615894039736,
      "grad_norm": 5.81339002256944,
      "learning_rate": 1.05e-05,
      "loss": 3.4062,
      "step": 70
    },
    {
      "epoch": 0.04701986754966887,
      "grad_norm": 6.483865471822502,
      "learning_rate": 1.0649999999999998e-05,
      "loss": 3.5469,
      "step": 71
    },
    {
      "epoch": 0.04768211920529802,
      "grad_norm": 6.681407312460688,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 3.0469,
      "step": 72
    },
    {
      "epoch": 0.048344370860927154,
      "grad_norm": 5.900284275769466,
      "learning_rate": 1.0949999999999998e-05,
      "loss": 3.4062,
      "step": 73
    },
    {
      "epoch": 0.04900662251655629,
      "grad_norm": 6.565697552430453,
      "learning_rate": 1.1099999999999999e-05,
      "loss": 3.4219,
      "step": 74
    },
    {
      "epoch": 0.04966887417218543,
      "grad_norm": 6.479208793249415,
      "learning_rate": 1.1249999999999999e-05,
      "loss": 3.4375,
      "step": 75
    },
    {
      "epoch": 0.05033112582781457,
      "grad_norm": 10.119589497347254,
      "learning_rate": 1.14e-05,
      "loss": 3.4062,
      "step": 76
    },
    {
      "epoch": 0.05099337748344371,
      "grad_norm": 5.657504471843712,
      "learning_rate": 1.155e-05,
      "loss": 3.4375,
      "step": 77
    },
    {
      "epoch": 0.051655629139072845,
      "grad_norm": 5.762975347033508,
      "learning_rate": 1.17e-05,
      "loss": 3.3906,
      "step": 78
    },
    {
      "epoch": 0.05231788079470199,
      "grad_norm": 5.472175426409925,
      "learning_rate": 1.1849999999999998e-05,
      "loss": 3.3281,
      "step": 79
    },
    {
      "epoch": 0.052980132450331126,
      "grad_norm": 5.36080185986848,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 3.3125,
      "step": 80
    },
    {
      "epoch": 0.05364238410596026,
      "grad_norm": 8.6638918869147,
      "learning_rate": 1.2149999999999999e-05,
      "loss": 3.3438,
      "step": 81
    },
    {
      "epoch": 0.054304635761589407,
      "grad_norm": 5.674651496967873,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 3.4062,
      "step": 82
    },
    {
      "epoch": 0.05496688741721854,
      "grad_norm": 5.361382771034597,
      "learning_rate": 1.245e-05,
      "loss": 2.9844,
      "step": 83
    },
    {
      "epoch": 0.05562913907284768,
      "grad_norm": 6.09753258999747,
      "learning_rate": 1.26e-05,
      "loss": 3.0469,
      "step": 84
    },
    {
      "epoch": 0.056291390728476824,
      "grad_norm": 7.251556938686882,
      "learning_rate": 1.275e-05,
      "loss": 3.3906,
      "step": 85
    },
    {
      "epoch": 0.05695364238410596,
      "grad_norm": 8.686673351259,
      "learning_rate": 1.2899999999999998e-05,
      "loss": 3.4062,
      "step": 86
    },
    {
      "epoch": 0.0576158940397351,
      "grad_norm": 5.322342010285953,
      "learning_rate": 1.3049999999999999e-05,
      "loss": 3.2031,
      "step": 87
    },
    {
      "epoch": 0.05827814569536424,
      "grad_norm": 7.568736522500978,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 3.2031,
      "step": 88
    },
    {
      "epoch": 0.05894039735099338,
      "grad_norm": 7.860436148140724,
      "learning_rate": 1.3349999999999998e-05,
      "loss": 3.25,
      "step": 89
    },
    {
      "epoch": 0.059602649006622516,
      "grad_norm": 5.184123992360204,
      "learning_rate": 1.3499999999999998e-05,
      "loss": 3.4219,
      "step": 90
    },
    {
      "epoch": 0.06026490066225165,
      "grad_norm": 5.15407267726703,
      "learning_rate": 1.3649999999999998e-05,
      "loss": 3.2969,
      "step": 91
    },
    {
      "epoch": 0.060927152317880796,
      "grad_norm": 6.113863867178894,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 3.5469,
      "step": 92
    },
    {
      "epoch": 0.06158940397350993,
      "grad_norm": 5.501189495418315,
      "learning_rate": 1.3949999999999999e-05,
      "loss": 3.2031,
      "step": 93
    },
    {
      "epoch": 0.06225165562913907,
      "grad_norm": 5.771850788341689,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 3.3281,
      "step": 94
    },
    {
      "epoch": 0.06291390728476821,
      "grad_norm": 5.807834746409673,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 3.5625,
      "step": 95
    },
    {
      "epoch": 0.06357615894039735,
      "grad_norm": 63.21753084442461,
      "learning_rate": 1.44e-05,
      "loss": 3.4062,
      "step": 96
    },
    {
      "epoch": 0.06423841059602649,
      "grad_norm": 5.462087047107723,
      "learning_rate": 1.455e-05,
      "loss": 3.2969,
      "step": 97
    },
    {
      "epoch": 0.06490066225165562,
      "grad_norm": 6.078939244629908,
      "learning_rate": 1.47e-05,
      "loss": 3.7031,
      "step": 98
    },
    {
      "epoch": 0.06556291390728476,
      "grad_norm": 4.6847295880819075,
      "learning_rate": 1.485e-05,
      "loss": 3.1875,
      "step": 99
    },
    {
      "epoch": 0.06622516556291391,
      "grad_norm": 5.81891908564673,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 2.9219,
      "step": 100
    },
    {
      "epoch": 0.06688741721854305,
      "grad_norm": 5.860334849479316,
      "learning_rate": 1.5149999999999999e-05,
      "loss": 3.4688,
      "step": 101
    },
    {
      "epoch": 0.06754966887417219,
      "grad_norm": 5.771205621871528,
      "learning_rate": 1.53e-05,
      "loss": 3.375,
      "step": 102
    },
    {
      "epoch": 0.06821192052980132,
      "grad_norm": 5.50689544288883,
      "learning_rate": 1.545e-05,
      "loss": 3.4219,
      "step": 103
    },
    {
      "epoch": 0.06887417218543046,
      "grad_norm": 8.700164391791958,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 3.1562,
      "step": 104
    },
    {
      "epoch": 0.0695364238410596,
      "grad_norm": 5.26785768330116,
      "learning_rate": 1.5749999999999997e-05,
      "loss": 3.2656,
      "step": 105
    },
    {
      "epoch": 0.07019867549668875,
      "grad_norm": 5.1421669104430086,
      "learning_rate": 1.5899999999999997e-05,
      "loss": 3.25,
      "step": 106
    },
    {
      "epoch": 0.07086092715231788,
      "grad_norm": 5.3978457374794555,
      "learning_rate": 1.6049999999999997e-05,
      "loss": 3.2188,
      "step": 107
    },
    {
      "epoch": 0.07152317880794702,
      "grad_norm": 5.33549150599327,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 3.3281,
      "step": 108
    },
    {
      "epoch": 0.07218543046357616,
      "grad_norm": 5.173401967196686,
      "learning_rate": 1.6349999999999998e-05,
      "loss": 3.3906,
      "step": 109
    },
    {
      "epoch": 0.0728476821192053,
      "grad_norm": 5.2978481071786145,
      "learning_rate": 1.6499999999999998e-05,
      "loss": 3.2656,
      "step": 110
    },
    {
      "epoch": 0.07350993377483443,
      "grad_norm": 9.97935657801475,
      "learning_rate": 1.6649999999999998e-05,
      "loss": 3.2812,
      "step": 111
    },
    {
      "epoch": 0.07417218543046358,
      "grad_norm": 4.6848286454403105,
      "learning_rate": 1.68e-05,
      "loss": 3.4688,
      "step": 112
    },
    {
      "epoch": 0.07483443708609272,
      "grad_norm": 4.918402389219796,
      "learning_rate": 1.695e-05,
      "loss": 3.3281,
      "step": 113
    },
    {
      "epoch": 0.07549668874172186,
      "grad_norm": 6.630124232917517,
      "learning_rate": 1.71e-05,
      "loss": 3.0312,
      "step": 114
    },
    {
      "epoch": 0.076158940397351,
      "grad_norm": 5.1924776683830105,
      "learning_rate": 1.725e-05,
      "loss": 3.0625,
      "step": 115
    },
    {
      "epoch": 0.07682119205298013,
      "grad_norm": 6.984685865380342,
      "learning_rate": 1.74e-05,
      "loss": 3.5312,
      "step": 116
    },
    {
      "epoch": 0.07748344370860927,
      "grad_norm": 4.67514968669566,
      "learning_rate": 1.755e-05,
      "loss": 3.1875,
      "step": 117
    },
    {
      "epoch": 0.0781456953642384,
      "grad_norm": 4.8054194464335325,
      "learning_rate": 1.7699999999999997e-05,
      "loss": 3.2969,
      "step": 118
    },
    {
      "epoch": 0.07880794701986756,
      "grad_norm": 4.989066147447696,
      "learning_rate": 1.7849999999999997e-05,
      "loss": 2.8438,
      "step": 119
    },
    {
      "epoch": 0.07947019867549669,
      "grad_norm": 4.434139647061637,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 3.3281,
      "step": 120
    },
    {
      "epoch": 0.08013245033112583,
      "grad_norm": 4.848827523568246,
      "learning_rate": 1.8149999999999997e-05,
      "loss": 3.25,
      "step": 121
    },
    {
      "epoch": 0.08079470198675497,
      "grad_norm": 4.819274007507709,
      "learning_rate": 1.8299999999999998e-05,
      "loss": 3.375,
      "step": 122
    },
    {
      "epoch": 0.0814569536423841,
      "grad_norm": 4.6709028641541614,
      "learning_rate": 1.8449999999999998e-05,
      "loss": 3.125,
      "step": 123
    },
    {
      "epoch": 0.08211920529801324,
      "grad_norm": 5.111429794195006,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 3.1094,
      "step": 124
    },
    {
      "epoch": 0.08278145695364239,
      "grad_norm": 5.096698395147178,
      "learning_rate": 1.875e-05,
      "loss": 3.2969,
      "step": 125
    },
    {
      "epoch": 0.08344370860927153,
      "grad_norm": 5.314646436061757,
      "learning_rate": 1.89e-05,
      "loss": 3.25,
      "step": 126
    },
    {
      "epoch": 0.08410596026490066,
      "grad_norm": 4.843553978032765,
      "learning_rate": 1.905e-05,
      "loss": 3.3594,
      "step": 127
    },
    {
      "epoch": 0.0847682119205298,
      "grad_norm": 4.240870302572376,
      "learning_rate": 1.92e-05,
      "loss": 3.2656,
      "step": 128
    },
    {
      "epoch": 0.08543046357615894,
      "grad_norm": 4.791622055588395,
      "learning_rate": 1.935e-05,
      "loss": 3.0156,
      "step": 129
    },
    {
      "epoch": 0.08609271523178808,
      "grad_norm": 4.485961677489531,
      "learning_rate": 1.95e-05,
      "loss": 3.2969,
      "step": 130
    },
    {
      "epoch": 0.08675496688741721,
      "grad_norm": 4.759869683803728,
      "learning_rate": 1.965e-05,
      "loss": 3.4219,
      "step": 131
    },
    {
      "epoch": 0.08741721854304636,
      "grad_norm": 4.391280244843061,
      "learning_rate": 1.98e-05,
      "loss": 3.1094,
      "step": 132
    },
    {
      "epoch": 0.0880794701986755,
      "grad_norm": 4.3469035092862045,
      "learning_rate": 1.995e-05,
      "loss": 3.1094,
      "step": 133
    },
    {
      "epoch": 0.08874172185430464,
      "grad_norm": 4.190341140841672,
      "learning_rate": 2.01e-05,
      "loss": 3.0469,
      "step": 134
    },
    {
      "epoch": 0.08940397350993377,
      "grad_norm": 4.49563004174637,
      "learning_rate": 2.025e-05,
      "loss": 2.9844,
      "step": 135
    },
    {
      "epoch": 0.09006622516556291,
      "grad_norm": 5.209854057537542,
      "learning_rate": 2.04e-05,
      "loss": 3.6094,
      "step": 136
    },
    {
      "epoch": 0.09072847682119205,
      "grad_norm": 4.470904397461312,
      "learning_rate": 2.055e-05,
      "loss": 3.3594,
      "step": 137
    },
    {
      "epoch": 0.0913907284768212,
      "grad_norm": 4.404925028420012,
      "learning_rate": 2.07e-05,
      "loss": 3.3438,
      "step": 138
    },
    {
      "epoch": 0.09205298013245033,
      "grad_norm": 4.235668053153122,
      "learning_rate": 2.085e-05,
      "loss": 3.2031,
      "step": 139
    },
    {
      "epoch": 0.09271523178807947,
      "grad_norm": 5.1489886719249585,
      "learning_rate": 2.1e-05,
      "loss": 3.2656,
      "step": 140
    },
    {
      "epoch": 0.09337748344370861,
      "grad_norm": 4.367981545093435,
      "learning_rate": 2.1149999999999996e-05,
      "loss": 3.1719,
      "step": 141
    },
    {
      "epoch": 0.09403973509933775,
      "grad_norm": 4.2364539730829245,
      "learning_rate": 2.1299999999999996e-05,
      "loss": 3.25,
      "step": 142
    },
    {
      "epoch": 0.09470198675496688,
      "grad_norm": 4.5575893875267015,
      "learning_rate": 2.1449999999999996e-05,
      "loss": 3.2969,
      "step": 143
    },
    {
      "epoch": 0.09536423841059603,
      "grad_norm": 3.949076483181059,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 3.1406,
      "step": 144
    },
    {
      "epoch": 0.09602649006622517,
      "grad_norm": 4.244144790740239,
      "learning_rate": 2.1749999999999997e-05,
      "loss": 3.1094,
      "step": 145
    },
    {
      "epoch": 0.09668874172185431,
      "grad_norm": 4.294050854634997,
      "learning_rate": 2.1899999999999997e-05,
      "loss": 3.1875,
      "step": 146
    },
    {
      "epoch": 0.09735099337748344,
      "grad_norm": 4.239134750651539,
      "learning_rate": 2.2049999999999997e-05,
      "loss": 2.8594,
      "step": 147
    },
    {
      "epoch": 0.09801324503311258,
      "grad_norm": 4.212574580317083,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 3.2344,
      "step": 148
    },
    {
      "epoch": 0.09867549668874172,
      "grad_norm": 4.092774522252905,
      "learning_rate": 2.2349999999999998e-05,
      "loss": 3.1562,
      "step": 149
    },
    {
      "epoch": 0.09933774834437085,
      "grad_norm": 4.153421618460364,
      "learning_rate": 2.2499999999999998e-05,
      "loss": 2.8281,
      "step": 150
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.827904818874844,
      "learning_rate": 2.2649999999999998e-05,
      "loss": 3.3594,
      "step": 151
    },
    {
      "epoch": 0.10066225165562914,
      "grad_norm": 4.652360213901008,
      "learning_rate": 2.28e-05,
      "loss": 3.2188,
      "step": 152
    },
    {
      "epoch": 0.10132450331125828,
      "grad_norm": 4.267631099265926,
      "learning_rate": 2.295e-05,
      "loss": 3.125,
      "step": 153
    },
    {
      "epoch": 0.10198675496688742,
      "grad_norm": 4.052690594441697,
      "learning_rate": 2.31e-05,
      "loss": 3.3906,
      "step": 154
    },
    {
      "epoch": 0.10264900662251655,
      "grad_norm": 4.021559285610502,
      "learning_rate": 2.325e-05,
      "loss": 3.1562,
      "step": 155
    },
    {
      "epoch": 0.10331125827814569,
      "grad_norm": 4.222179144888645,
      "learning_rate": 2.34e-05,
      "loss": 3.1719,
      "step": 156
    },
    {
      "epoch": 0.10397350993377484,
      "grad_norm": 4.1681408751929006,
      "learning_rate": 2.3549999999999996e-05,
      "loss": 3.1719,
      "step": 157
    },
    {
      "epoch": 0.10463576158940398,
      "grad_norm": 4.1208863311935735,
      "learning_rate": 2.3699999999999997e-05,
      "loss": 3.2344,
      "step": 158
    },
    {
      "epoch": 0.10529801324503311,
      "grad_norm": 4.15281753214379,
      "learning_rate": 2.3849999999999997e-05,
      "loss": 3.2031,
      "step": 159
    },
    {
      "epoch": 0.10596026490066225,
      "grad_norm": 3.7206428623442744,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 2.875,
      "step": 160
    },
    {
      "epoch": 0.10662251655629139,
      "grad_norm": 4.043442327454647,
      "learning_rate": 2.4149999999999997e-05,
      "loss": 3.1875,
      "step": 161
    },
    {
      "epoch": 0.10728476821192053,
      "grad_norm": 3.8834759506069174,
      "learning_rate": 2.4299999999999998e-05,
      "loss": 2.7812,
      "step": 162
    },
    {
      "epoch": 0.10794701986754966,
      "grad_norm": 3.6602152758977318,
      "learning_rate": 2.4449999999999998e-05,
      "loss": 3.1875,
      "step": 163
    },
    {
      "epoch": 0.10860927152317881,
      "grad_norm": 4.2287619105009675,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 2.75,
      "step": 164
    },
    {
      "epoch": 0.10927152317880795,
      "grad_norm": 110.11644419668347,
      "learning_rate": 2.475e-05,
      "loss": 3.1875,
      "step": 165
    },
    {
      "epoch": 0.10993377483443709,
      "grad_norm": 4.548875327770721,
      "learning_rate": 2.49e-05,
      "loss": 3.1719,
      "step": 166
    },
    {
      "epoch": 0.11059602649006622,
      "grad_norm": 4.2580577760972425,
      "learning_rate": 2.505e-05,
      "loss": 3.3594,
      "step": 167
    },
    {
      "epoch": 0.11125827814569536,
      "grad_norm": 4.897121920365212,
      "learning_rate": 2.52e-05,
      "loss": 3.0625,
      "step": 168
    },
    {
      "epoch": 0.1119205298013245,
      "grad_norm": 4.201568083604385,
      "learning_rate": 2.535e-05,
      "loss": 3.1875,
      "step": 169
    },
    {
      "epoch": 0.11258278145695365,
      "grad_norm": 4.529239410849345,
      "learning_rate": 2.55e-05,
      "loss": 2.875,
      "step": 170
    },
    {
      "epoch": 0.11324503311258279,
      "grad_norm": 5.123301636622721,
      "learning_rate": 2.565e-05,
      "loss": 3.3906,
      "step": 171
    },
    {
      "epoch": 0.11390728476821192,
      "grad_norm": 4.49654856501852,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 3.1406,
      "step": 172
    },
    {
      "epoch": 0.11456953642384106,
      "grad_norm": 4.492921901944353,
      "learning_rate": 2.5949999999999997e-05,
      "loss": 3.0938,
      "step": 173
    },
    {
      "epoch": 0.1152317880794702,
      "grad_norm": 4.923309698447017,
      "learning_rate": 2.6099999999999997e-05,
      "loss": 3.0,
      "step": 174
    },
    {
      "epoch": 0.11589403973509933,
      "grad_norm": 4.690247596431798,
      "learning_rate": 2.6249999999999998e-05,
      "loss": 3.4688,
      "step": 175
    },
    {
      "epoch": 0.11655629139072848,
      "grad_norm": 4.272596871015295,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 3.2969,
      "step": 176
    },
    {
      "epoch": 0.11721854304635762,
      "grad_norm": 4.732381577853729,
      "learning_rate": 2.6549999999999995e-05,
      "loss": 3.0469,
      "step": 177
    },
    {
      "epoch": 0.11788079470198676,
      "grad_norm": 4.501743902648583,
      "learning_rate": 2.6699999999999995e-05,
      "loss": 3.25,
      "step": 178
    },
    {
      "epoch": 0.1185430463576159,
      "grad_norm": 4.179809817716381,
      "learning_rate": 2.6849999999999995e-05,
      "loss": 3.25,
      "step": 179
    },
    {
      "epoch": 0.11920529801324503,
      "grad_norm": 3.8587769390686555,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 2.9375,
      "step": 180
    },
    {
      "epoch": 0.11986754966887417,
      "grad_norm": 4.092849684724,
      "learning_rate": 2.7149999999999996e-05,
      "loss": 3.1406,
      "step": 181
    },
    {
      "epoch": 0.1205298013245033,
      "grad_norm": 4.053902606102931,
      "learning_rate": 2.7299999999999996e-05,
      "loss": 2.8594,
      "step": 182
    },
    {
      "epoch": 0.12119205298013246,
      "grad_norm": 3.9965128349894683,
      "learning_rate": 2.7449999999999996e-05,
      "loss": 3.2344,
      "step": 183
    },
    {
      "epoch": 0.12185430463576159,
      "grad_norm": 4.196957126428847,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 3.3125,
      "step": 184
    },
    {
      "epoch": 0.12251655629139073,
      "grad_norm": 3.924473440394618,
      "learning_rate": 2.7749999999999997e-05,
      "loss": 3.25,
      "step": 185
    },
    {
      "epoch": 0.12317880794701987,
      "grad_norm": 3.9970385700387805,
      "learning_rate": 2.7899999999999997e-05,
      "loss": 2.9844,
      "step": 186
    },
    {
      "epoch": 0.123841059602649,
      "grad_norm": 3.947088309652612,
      "learning_rate": 2.8049999999999997e-05,
      "loss": 3.1562,
      "step": 187
    },
    {
      "epoch": 0.12450331125827814,
      "grad_norm": 4.382439862482131,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 2.8438,
      "step": 188
    },
    {
      "epoch": 0.1251655629139073,
      "grad_norm": 4.385501978447136,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 2.9062,
      "step": 189
    },
    {
      "epoch": 0.12582781456953643,
      "grad_norm": 3.9039332679765724,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 3.0312,
      "step": 190
    },
    {
      "epoch": 0.12649006622516556,
      "grad_norm": 3.991429810304012,
      "learning_rate": 2.865e-05,
      "loss": 3.25,
      "step": 191
    },
    {
      "epoch": 0.1271523178807947,
      "grad_norm": 3.7001444238201446,
      "learning_rate": 2.88e-05,
      "loss": 3.2031,
      "step": 192
    },
    {
      "epoch": 0.12781456953642384,
      "grad_norm": 3.7322489527420672,
      "learning_rate": 2.895e-05,
      "loss": 3.3906,
      "step": 193
    },
    {
      "epoch": 0.12847682119205298,
      "grad_norm": 3.623251477088876,
      "learning_rate": 2.91e-05,
      "loss": 3.3281,
      "step": 194
    },
    {
      "epoch": 0.1291390728476821,
      "grad_norm": 3.867124928209203,
      "learning_rate": 2.925e-05,
      "loss": 3.2188,
      "step": 195
    },
    {
      "epoch": 0.12980132450331125,
      "grad_norm": 3.8359874276884285,
      "learning_rate": 2.94e-05,
      "loss": 3.0938,
      "step": 196
    },
    {
      "epoch": 0.1304635761589404,
      "grad_norm": 4.080413503782875,
      "learning_rate": 2.955e-05,
      "loss": 2.6875,
      "step": 197
    },
    {
      "epoch": 0.13112582781456952,
      "grad_norm": 4.601332797904551,
      "learning_rate": 2.97e-05,
      "loss": 3.2344,
      "step": 198
    },
    {
      "epoch": 0.1317880794701987,
      "grad_norm": 4.127875712856974,
      "learning_rate": 2.985e-05,
      "loss": 3.0938,
      "step": 199
    },
    {
      "epoch": 0.13245033112582782,
      "grad_norm": 3.675504261209941,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 3.0938,
      "step": 200
    },
    {
      "epoch": 0.13311258278145696,
      "grad_norm": 3.84472757645993,
      "learning_rate": 3.0149999999999998e-05,
      "loss": 3.2656,
      "step": 201
    },
    {
      "epoch": 0.1337748344370861,
      "grad_norm": 3.6825027293286814,
      "learning_rate": 3.0299999999999998e-05,
      "loss": 3.0469,
      "step": 202
    },
    {
      "epoch": 0.13443708609271524,
      "grad_norm": 3.503359203782447,
      "learning_rate": 3.0449999999999998e-05,
      "loss": 3.1562,
      "step": 203
    },
    {
      "epoch": 0.13509933774834437,
      "grad_norm": 4.067763833084508,
      "learning_rate": 3.06e-05,
      "loss": 3.3281,
      "step": 204
    },
    {
      "epoch": 0.1357615894039735,
      "grad_norm": 3.5888029350269224,
      "learning_rate": 3.0749999999999995e-05,
      "loss": 3.0625,
      "step": 205
    },
    {
      "epoch": 0.13642384105960265,
      "grad_norm": 4.0400209174875155,
      "learning_rate": 3.09e-05,
      "loss": 3.3281,
      "step": 206
    },
    {
      "epoch": 0.13708609271523178,
      "grad_norm": 3.5904414498861272,
      "learning_rate": 3.1049999999999996e-05,
      "loss": 3.0938,
      "step": 207
    },
    {
      "epoch": 0.13774834437086092,
      "grad_norm": 3.625287833426859,
      "learning_rate": 3.119999999999999e-05,
      "loss": 3.0312,
      "step": 208
    },
    {
      "epoch": 0.13841059602649006,
      "grad_norm": 3.608411288215931,
      "learning_rate": 3.1349999999999996e-05,
      "loss": 2.9531,
      "step": 209
    },
    {
      "epoch": 0.1390728476821192,
      "grad_norm": 3.5996876944737,
      "learning_rate": 3.149999999999999e-05,
      "loss": 3.25,
      "step": 210
    },
    {
      "epoch": 0.13973509933774833,
      "grad_norm": 3.3429836589835347,
      "learning_rate": 3.165e-05,
      "loss": 2.8594,
      "step": 211
    },
    {
      "epoch": 0.1403973509933775,
      "grad_norm": 3.6784230937164337,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 2.75,
      "step": 212
    },
    {
      "epoch": 0.14105960264900663,
      "grad_norm": 3.6646729961201854,
      "learning_rate": 3.195e-05,
      "loss": 2.9844,
      "step": 213
    },
    {
      "epoch": 0.14172185430463577,
      "grad_norm": 3.61158194251105,
      "learning_rate": 3.2099999999999994e-05,
      "loss": 3.2031,
      "step": 214
    },
    {
      "epoch": 0.1423841059602649,
      "grad_norm": 3.732622006010356,
      "learning_rate": 3.225e-05,
      "loss": 2.9219,
      "step": 215
    },
    {
      "epoch": 0.14304635761589404,
      "grad_norm": 3.5848477173929734,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 2.8594,
      "step": 216
    },
    {
      "epoch": 0.14370860927152318,
      "grad_norm": 3.511184340284042,
      "learning_rate": 3.255e-05,
      "loss": 2.9688,
      "step": 217
    },
    {
      "epoch": 0.14437086092715232,
      "grad_norm": 4.194066170987906,
      "learning_rate": 3.2699999999999995e-05,
      "loss": 3.0781,
      "step": 218
    },
    {
      "epoch": 0.14503311258278145,
      "grad_norm": 3.280374157945727,
      "learning_rate": 3.285e-05,
      "loss": 2.9062,
      "step": 219
    },
    {
      "epoch": 0.1456953642384106,
      "grad_norm": 3.3743497592004257,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 2.9219,
      "step": 220
    },
    {
      "epoch": 0.14635761589403973,
      "grad_norm": 3.362357268675812,
      "learning_rate": 3.315e-05,
      "loss": 3.375,
      "step": 221
    },
    {
      "epoch": 0.14701986754966886,
      "grad_norm": 3.5982062729311957,
      "learning_rate": 3.3299999999999996e-05,
      "loss": 3.2344,
      "step": 222
    },
    {
      "epoch": 0.147682119205298,
      "grad_norm": 3.4681319551500676,
      "learning_rate": 3.345e-05,
      "loss": 3.0,
      "step": 223
    },
    {
      "epoch": 0.14834437086092717,
      "grad_norm": 3.3635239042832974,
      "learning_rate": 3.36e-05,
      "loss": 3.2969,
      "step": 224
    },
    {
      "epoch": 0.1490066225165563,
      "grad_norm": 3.2019600675360316,
      "learning_rate": 3.375e-05,
      "loss": 3.0469,
      "step": 225
    },
    {
      "epoch": 0.14966887417218544,
      "grad_norm": 3.494575607343542,
      "learning_rate": 3.39e-05,
      "loss": 2.9375,
      "step": 226
    },
    {
      "epoch": 0.15033112582781458,
      "grad_norm": 3.4057386810355155,
      "learning_rate": 3.405e-05,
      "loss": 3.3594,
      "step": 227
    },
    {
      "epoch": 0.1509933774834437,
      "grad_norm": 3.3601993218846906,
      "learning_rate": 3.42e-05,
      "loss": 3.1406,
      "step": 228
    },
    {
      "epoch": 0.15165562913907285,
      "grad_norm": 3.3991399692466477,
      "learning_rate": 3.435e-05,
      "loss": 2.9062,
      "step": 229
    },
    {
      "epoch": 0.152317880794702,
      "grad_norm": 3.294297373515897,
      "learning_rate": 3.45e-05,
      "loss": 2.6719,
      "step": 230
    },
    {
      "epoch": 0.15298013245033112,
      "grad_norm": 3.987609355582827,
      "learning_rate": 3.465e-05,
      "loss": 3.0781,
      "step": 231
    },
    {
      "epoch": 0.15364238410596026,
      "grad_norm": 3.9084259587624213,
      "learning_rate": 3.48e-05,
      "loss": 3.1406,
      "step": 232
    },
    {
      "epoch": 0.1543046357615894,
      "grad_norm": 3.3596587106716904,
      "learning_rate": 3.4949999999999996e-05,
      "loss": 3.0469,
      "step": 233
    },
    {
      "epoch": 0.15496688741721854,
      "grad_norm": 3.459787228106157,
      "learning_rate": 3.51e-05,
      "loss": 3.1406,
      "step": 234
    },
    {
      "epoch": 0.15562913907284767,
      "grad_norm": 3.2562410822583665,
      "learning_rate": 3.5249999999999996e-05,
      "loss": 3.0312,
      "step": 235
    },
    {
      "epoch": 0.1562913907284768,
      "grad_norm": 3.474251339129339,
      "learning_rate": 3.539999999999999e-05,
      "loss": 3.1094,
      "step": 236
    },
    {
      "epoch": 0.15695364238410597,
      "grad_norm": 3.2986637399227536,
      "learning_rate": 3.555e-05,
      "loss": 3.0625,
      "step": 237
    },
    {
      "epoch": 0.1576158940397351,
      "grad_norm": 3.2746009385243386,
      "learning_rate": 3.5699999999999994e-05,
      "loss": 3.25,
      "step": 238
    },
    {
      "epoch": 0.15827814569536425,
      "grad_norm": 3.0918839052299756,
      "learning_rate": 3.585e-05,
      "loss": 3.0781,
      "step": 239
    },
    {
      "epoch": 0.15894039735099338,
      "grad_norm": 3.2220388664461024,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 2.6875,
      "step": 240
    },
    {
      "epoch": 0.15960264900662252,
      "grad_norm": 3.1368033793353365,
      "learning_rate": 3.615e-05,
      "loss": 2.9219,
      "step": 241
    },
    {
      "epoch": 0.16026490066225166,
      "grad_norm": 3.33429128151937,
      "learning_rate": 3.6299999999999995e-05,
      "loss": 3.2188,
      "step": 242
    },
    {
      "epoch": 0.1609271523178808,
      "grad_norm": 3.127316735595342,
      "learning_rate": 3.645e-05,
      "loss": 3.1562,
      "step": 243
    },
    {
      "epoch": 0.16158940397350993,
      "grad_norm": 3.2386107224953395,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 3.1719,
      "step": 244
    },
    {
      "epoch": 0.16225165562913907,
      "grad_norm": 3.058043859797535,
      "learning_rate": 3.675e-05,
      "loss": 3.0625,
      "step": 245
    },
    {
      "epoch": 0.1629139072847682,
      "grad_norm": 3.157439443754408,
      "learning_rate": 3.6899999999999996e-05,
      "loss": 2.875,
      "step": 246
    },
    {
      "epoch": 0.16357615894039734,
      "grad_norm": 3.2561897128261603,
      "learning_rate": 3.705e-05,
      "loss": 3.25,
      "step": 247
    },
    {
      "epoch": 0.16423841059602648,
      "grad_norm": 3.179349634953305,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 3.1406,
      "step": 248
    },
    {
      "epoch": 0.16490066225165562,
      "grad_norm": 3.242858743296943,
      "learning_rate": 3.735e-05,
      "loss": 3.1562,
      "step": 249
    },
    {
      "epoch": 0.16556291390728478,
      "grad_norm": 2.8791647506080795,
      "learning_rate": 3.75e-05,
      "loss": 2.9219,
      "step": 250
    },
    {
      "epoch": 0.16622516556291392,
      "grad_norm": 3.076401994057578,
      "learning_rate": 3.7649999999999994e-05,
      "loss": 2.8594,
      "step": 251
    },
    {
      "epoch": 0.16688741721854305,
      "grad_norm": 3.0618051560972988,
      "learning_rate": 3.78e-05,
      "loss": 3.2344,
      "step": 252
    },
    {
      "epoch": 0.1675496688741722,
      "grad_norm": 3.0401093762867117,
      "learning_rate": 3.7949999999999994e-05,
      "loss": 3.0312,
      "step": 253
    },
    {
      "epoch": 0.16821192052980133,
      "grad_norm": 3.0317580316936565,
      "learning_rate": 3.81e-05,
      "loss": 2.5938,
      "step": 254
    },
    {
      "epoch": 0.16887417218543047,
      "grad_norm": 3.4131359537704227,
      "learning_rate": 3.8249999999999995e-05,
      "loss": 2.9531,
      "step": 255
    },
    {
      "epoch": 0.1695364238410596,
      "grad_norm": 2.824527994440723,
      "learning_rate": 3.84e-05,
      "loss": 2.8906,
      "step": 256
    },
    {
      "epoch": 0.17019867549668874,
      "grad_norm": 2.830675833417707,
      "learning_rate": 3.8549999999999995e-05,
      "loss": 2.5781,
      "step": 257
    },
    {
      "epoch": 0.17086092715231788,
      "grad_norm": 3.041340070819647,
      "learning_rate": 3.87e-05,
      "loss": 2.9688,
      "step": 258
    },
    {
      "epoch": 0.171523178807947,
      "grad_norm": 3.035707324733149,
      "learning_rate": 3.8849999999999996e-05,
      "loss": 3.0,
      "step": 259
    },
    {
      "epoch": 0.17218543046357615,
      "grad_norm": 3.297057110848455,
      "learning_rate": 3.9e-05,
      "loss": 3.0781,
      "step": 260
    },
    {
      "epoch": 0.1728476821192053,
      "grad_norm": 3.0752860411867613,
      "learning_rate": 3.9149999999999996e-05,
      "loss": 3.125,
      "step": 261
    },
    {
      "epoch": 0.17350993377483442,
      "grad_norm": 2.9046723573594737,
      "learning_rate": 3.93e-05,
      "loss": 2.8438,
      "step": 262
    },
    {
      "epoch": 0.1741721854304636,
      "grad_norm": 3.171530410999741,
      "learning_rate": 3.945e-05,
      "loss": 3.4062,
      "step": 263
    },
    {
      "epoch": 0.17483443708609273,
      "grad_norm": 2.8562672189292933,
      "learning_rate": 3.96e-05,
      "loss": 3.0312,
      "step": 264
    },
    {
      "epoch": 0.17549668874172186,
      "grad_norm": 3.0497161696094186,
      "learning_rate": 3.975e-05,
      "loss": 3.1562,
      "step": 265
    },
    {
      "epoch": 0.176158940397351,
      "grad_norm": 3.0277077489573587,
      "learning_rate": 3.99e-05,
      "loss": 3.1562,
      "step": 266
    },
    {
      "epoch": 0.17682119205298014,
      "grad_norm": 3.023428348791391,
      "learning_rate": 4.005e-05,
      "loss": 3.0781,
      "step": 267
    },
    {
      "epoch": 0.17748344370860927,
      "grad_norm": 2.7763290247551726,
      "learning_rate": 4.02e-05,
      "loss": 3.0469,
      "step": 268
    },
    {
      "epoch": 0.1781456953642384,
      "grad_norm": 2.933416617657179,
      "learning_rate": 4.035e-05,
      "loss": 2.7188,
      "step": 269
    },
    {
      "epoch": 0.17880794701986755,
      "grad_norm": 2.9765295399280896,
      "learning_rate": 4.05e-05,
      "loss": 2.9844,
      "step": 270
    },
    {
      "epoch": 0.17947019867549668,
      "grad_norm": 3.0030915229628175,
      "learning_rate": 4.065e-05,
      "loss": 3.0156,
      "step": 271
    },
    {
      "epoch": 0.18013245033112582,
      "grad_norm": 2.979687167041924,
      "learning_rate": 4.08e-05,
      "loss": 3.1094,
      "step": 272
    },
    {
      "epoch": 0.18079470198675496,
      "grad_norm": 2.9898594996365406,
      "learning_rate": 4.095e-05,
      "loss": 2.8594,
      "step": 273
    },
    {
      "epoch": 0.1814569536423841,
      "grad_norm": 3.063881338434763,
      "learning_rate": 4.11e-05,
      "loss": 3.0781,
      "step": 274
    },
    {
      "epoch": 0.18211920529801323,
      "grad_norm": 2.827998356797006,
      "learning_rate": 4.125e-05,
      "loss": 3.0,
      "step": 275
    },
    {
      "epoch": 0.1827814569536424,
      "grad_norm": 2.830631906167569,
      "learning_rate": 4.14e-05,
      "loss": 2.9844,
      "step": 276
    },
    {
      "epoch": 0.18344370860927153,
      "grad_norm": 2.671075701611255,
      "learning_rate": 4.155e-05,
      "loss": 3.1562,
      "step": 277
    },
    {
      "epoch": 0.18410596026490067,
      "grad_norm": 2.9432234846518672,
      "learning_rate": 4.17e-05,
      "loss": 2.875,
      "step": 278
    },
    {
      "epoch": 0.1847682119205298,
      "grad_norm": 2.8007775083308384,
      "learning_rate": 4.185e-05,
      "loss": 2.9219,
      "step": 279
    },
    {
      "epoch": 0.18543046357615894,
      "grad_norm": 2.827153897323655,
      "learning_rate": 4.2e-05,
      "loss": 2.8594,
      "step": 280
    },
    {
      "epoch": 0.18609271523178808,
      "grad_norm": 3.092860587637192,
      "learning_rate": 4.215e-05,
      "loss": 2.625,
      "step": 281
    },
    {
      "epoch": 0.18675496688741722,
      "grad_norm": 3.3331355753296332,
      "learning_rate": 4.229999999999999e-05,
      "loss": 3.1094,
      "step": 282
    },
    {
      "epoch": 0.18741721854304635,
      "grad_norm": 2.8022509072008037,
      "learning_rate": 4.2449999999999995e-05,
      "loss": 2.8281,
      "step": 283
    },
    {
      "epoch": 0.1880794701986755,
      "grad_norm": 2.8557193406310604,
      "learning_rate": 4.259999999999999e-05,
      "loss": 2.9688,
      "step": 284
    },
    {
      "epoch": 0.18874172185430463,
      "grad_norm": 3.149206840898262,
      "learning_rate": 4.2749999999999996e-05,
      "loss": 3.0,
      "step": 285
    },
    {
      "epoch": 0.18940397350993377,
      "grad_norm": 2.9594738453379015,
      "learning_rate": 4.289999999999999e-05,
      "loss": 3.2031,
      "step": 286
    },
    {
      "epoch": 0.1900662251655629,
      "grad_norm": 2.7987124337866636,
      "learning_rate": 4.3049999999999996e-05,
      "loss": 2.9375,
      "step": 287
    },
    {
      "epoch": 0.19072847682119207,
      "grad_norm": 2.9720094900470366,
      "learning_rate": 4.319999999999999e-05,
      "loss": 3.1875,
      "step": 288
    },
    {
      "epoch": 0.1913907284768212,
      "grad_norm": 2.871756222471837,
      "learning_rate": 4.334999999999999e-05,
      "loss": 2.9219,
      "step": 289
    },
    {
      "epoch": 0.19205298013245034,
      "grad_norm": 3.0726921131145146,
      "learning_rate": 4.3499999999999993e-05,
      "loss": 3.2031,
      "step": 290
    },
    {
      "epoch": 0.19271523178807948,
      "grad_norm": 2.796357850006255,
      "learning_rate": 4.364999999999999e-05,
      "loss": 2.75,
      "step": 291
    },
    {
      "epoch": 0.19337748344370861,
      "grad_norm": 2.6140692779608314,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 2.7812,
      "step": 292
    },
    {
      "epoch": 0.19403973509933775,
      "grad_norm": 2.784207431419087,
      "learning_rate": 4.394999999999999e-05,
      "loss": 2.9375,
      "step": 293
    },
    {
      "epoch": 0.1947019867549669,
      "grad_norm": 2.8355379111399164,
      "learning_rate": 4.4099999999999995e-05,
      "loss": 3.0,
      "step": 294
    },
    {
      "epoch": 0.19536423841059603,
      "grad_norm": 3.2318235743742516,
      "learning_rate": 4.424999999999999e-05,
      "loss": 3.0156,
      "step": 295
    },
    {
      "epoch": 0.19602649006622516,
      "grad_norm": 2.9764608799525303,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 2.8906,
      "step": 296
    },
    {
      "epoch": 0.1966887417218543,
      "grad_norm": 3.050950613384662,
      "learning_rate": 4.454999999999999e-05,
      "loss": 3.0938,
      "step": 297
    },
    {
      "epoch": 0.19735099337748344,
      "grad_norm": 2.8236597243290387,
      "learning_rate": 4.4699999999999996e-05,
      "loss": 3.0625,
      "step": 298
    },
    {
      "epoch": 0.19801324503311257,
      "grad_norm": 3.07416263735879,
      "learning_rate": 4.484999999999999e-05,
      "loss": 3.0,
      "step": 299
    },
    {
      "epoch": 0.1986754966887417,
      "grad_norm": 2.8291727163734564,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 2.8281,
      "step": 300
    },
    {
      "epoch": 0.19933774834437087,
      "grad_norm": 2.7160819305649793,
      "learning_rate": 4.514999999999999e-05,
      "loss": 2.9062,
      "step": 301
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.8935347954748423,
      "learning_rate": 4.5299999999999997e-05,
      "loss": 2.9531,
      "step": 302
    },
    {
      "epoch": 0.20066225165562915,
      "grad_norm": 2.746357879541038,
      "learning_rate": 4.5449999999999993e-05,
      "loss": 2.625,
      "step": 303
    },
    {
      "epoch": 0.20132450331125828,
      "grad_norm": 2.864849257272957,
      "learning_rate": 4.56e-05,
      "loss": 2.4531,
      "step": 304
    },
    {
      "epoch": 0.20198675496688742,
      "grad_norm": 3.4970751732130183,
      "learning_rate": 4.5749999999999994e-05,
      "loss": 2.9219,
      "step": 305
    },
    {
      "epoch": 0.20264900662251656,
      "grad_norm": 2.8068190998074867,
      "learning_rate": 4.59e-05,
      "loss": 2.8438,
      "step": 306
    },
    {
      "epoch": 0.2033112582781457,
      "grad_norm": 2.7541039381240644,
      "learning_rate": 4.6049999999999994e-05,
      "loss": 2.9844,
      "step": 307
    },
    {
      "epoch": 0.20397350993377483,
      "grad_norm": 2.8593958986526617,
      "learning_rate": 4.62e-05,
      "loss": 2.5938,
      "step": 308
    },
    {
      "epoch": 0.20463576158940397,
      "grad_norm": 2.5503599862388855,
      "learning_rate": 4.6349999999999995e-05,
      "loss": 2.9219,
      "step": 309
    },
    {
      "epoch": 0.2052980132450331,
      "grad_norm": 2.805329586203324,
      "learning_rate": 4.65e-05,
      "loss": 2.9688,
      "step": 310
    },
    {
      "epoch": 0.20596026490066224,
      "grad_norm": 2.6603779276962194,
      "learning_rate": 4.6649999999999996e-05,
      "loss": 2.9375,
      "step": 311
    },
    {
      "epoch": 0.20662251655629138,
      "grad_norm": 2.7850339145149885,
      "learning_rate": 4.68e-05,
      "loss": 2.8438,
      "step": 312
    },
    {
      "epoch": 0.20728476821192052,
      "grad_norm": 2.6835506268112455,
      "learning_rate": 4.6949999999999996e-05,
      "loss": 3.1094,
      "step": 313
    },
    {
      "epoch": 0.20794701986754968,
      "grad_norm": 2.8575973590020025,
      "learning_rate": 4.709999999999999e-05,
      "loss": 3.2344,
      "step": 314
    },
    {
      "epoch": 0.20860927152317882,
      "grad_norm": 2.8124172645280714,
      "learning_rate": 4.7249999999999997e-05,
      "loss": 3.0469,
      "step": 315
    },
    {
      "epoch": 0.20927152317880796,
      "grad_norm": 2.6673390249864735,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 3.0156,
      "step": 316
    },
    {
      "epoch": 0.2099337748344371,
      "grad_norm": 2.864915607284846,
      "learning_rate": 4.755e-05,
      "loss": 3.0312,
      "step": 317
    },
    {
      "epoch": 0.21059602649006623,
      "grad_norm": 2.8974326959697847,
      "learning_rate": 4.7699999999999994e-05,
      "loss": 2.375,
      "step": 318
    },
    {
      "epoch": 0.21125827814569537,
      "grad_norm": 2.486815622189848,
      "learning_rate": 4.785e-05,
      "loss": 2.9531,
      "step": 319
    },
    {
      "epoch": 0.2119205298013245,
      "grad_norm": 3.1389257973535845,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 3.0781,
      "step": 320
    },
    {
      "epoch": 0.21258278145695364,
      "grad_norm": 2.855097913494004,
      "learning_rate": 4.815e-05,
      "loss": 3.2812,
      "step": 321
    },
    {
      "epoch": 0.21324503311258278,
      "grad_norm": 2.5855596268711416,
      "learning_rate": 4.8299999999999995e-05,
      "loss": 2.8594,
      "step": 322
    },
    {
      "epoch": 0.2139072847682119,
      "grad_norm": 2.7674153641588086,
      "learning_rate": 4.845e-05,
      "loss": 2.9062,
      "step": 323
    },
    {
      "epoch": 0.21456953642384105,
      "grad_norm": 2.796227189614921,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 3.0938,
      "step": 324
    },
    {
      "epoch": 0.2152317880794702,
      "grad_norm": 2.4121401508696283,
      "learning_rate": 4.875e-05,
      "loss": 2.9219,
      "step": 325
    },
    {
      "epoch": 0.21589403973509932,
      "grad_norm": 2.6208351101059097,
      "learning_rate": 4.8899999999999996e-05,
      "loss": 2.4062,
      "step": 326
    },
    {
      "epoch": 0.2165562913907285,
      "grad_norm": 2.90840734668696,
      "learning_rate": 4.905e-05,
      "loss": 2.9375,
      "step": 327
    },
    {
      "epoch": 0.21721854304635763,
      "grad_norm": 2.9364709830554445,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 3.1406,
      "step": 328
    },
    {
      "epoch": 0.21788079470198676,
      "grad_norm": 2.5824857389931966,
      "learning_rate": 4.935e-05,
      "loss": 2.9844,
      "step": 329
    },
    {
      "epoch": 0.2185430463576159,
      "grad_norm": 2.6891028958819483,
      "learning_rate": 4.95e-05,
      "loss": 3.0312,
      "step": 330
    },
    {
      "epoch": 0.21920529801324504,
      "grad_norm": 2.7301777533336073,
      "learning_rate": 4.965e-05,
      "loss": 3.0781,
      "step": 331
    },
    {
      "epoch": 0.21986754966887417,
      "grad_norm": 2.6687642863091665,
      "learning_rate": 4.98e-05,
      "loss": 3.1875,
      "step": 332
    },
    {
      "epoch": 0.2205298013245033,
      "grad_norm": 2.7919695483649307,
      "learning_rate": 4.995e-05,
      "loss": 2.9531,
      "step": 333
    },
    {
      "epoch": 0.22119205298013245,
      "grad_norm": 2.7075124944529314,
      "learning_rate": 5.01e-05,
      "loss": 3.0625,
      "step": 334
    },
    {
      "epoch": 0.22185430463576158,
      "grad_norm": 2.670072571378032,
      "learning_rate": 5.025e-05,
      "loss": 2.5781,
      "step": 335
    },
    {
      "epoch": 0.22251655629139072,
      "grad_norm": 2.9293439496870524,
      "learning_rate": 5.04e-05,
      "loss": 2.8594,
      "step": 336
    },
    {
      "epoch": 0.22317880794701986,
      "grad_norm": 2.5723702112330353,
      "learning_rate": 5.055e-05,
      "loss": 2.8594,
      "step": 337
    },
    {
      "epoch": 0.223841059602649,
      "grad_norm": 2.683002407769579,
      "learning_rate": 5.07e-05,
      "loss": 2.9219,
      "step": 338
    },
    {
      "epoch": 0.22450331125827813,
      "grad_norm": 2.520196330168102,
      "learning_rate": 5.0849999999999996e-05,
      "loss": 2.7656,
      "step": 339
    },
    {
      "epoch": 0.2251655629139073,
      "grad_norm": 2.617867096499689,
      "learning_rate": 5.1e-05,
      "loss": 2.9844,
      "step": 340
    },
    {
      "epoch": 0.22582781456953643,
      "grad_norm": 2.878837165143681,
      "learning_rate": 5.1149999999999996e-05,
      "loss": 3.2188,
      "step": 341
    },
    {
      "epoch": 0.22649006622516557,
      "grad_norm": 2.4772843115116374,
      "learning_rate": 5.13e-05,
      "loss": 2.7812,
      "step": 342
    },
    {
      "epoch": 0.2271523178807947,
      "grad_norm": 2.744876601078272,
      "learning_rate": 5.145e-05,
      "loss": 2.5469,
      "step": 343
    },
    {
      "epoch": 0.22781456953642384,
      "grad_norm": 2.7275454126235656,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 2.8594,
      "step": 344
    },
    {
      "epoch": 0.22847682119205298,
      "grad_norm": 2.6875250538505524,
      "learning_rate": 5.174999999999999e-05,
      "loss": 2.8281,
      "step": 345
    },
    {
      "epoch": 0.22913907284768212,
      "grad_norm": 2.9809323072875547,
      "learning_rate": 5.1899999999999994e-05,
      "loss": 3.0938,
      "step": 346
    },
    {
      "epoch": 0.22980132450331126,
      "grad_norm": 2.810630277845135,
      "learning_rate": 5.204999999999999e-05,
      "loss": 3.2969,
      "step": 347
    },
    {
      "epoch": 0.2304635761589404,
      "grad_norm": 2.7421934741225105,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 3.1406,
      "step": 348
    },
    {
      "epoch": 0.23112582781456953,
      "grad_norm": 2.651253979198602,
      "learning_rate": 5.234999999999999e-05,
      "loss": 2.8594,
      "step": 349
    },
    {
      "epoch": 0.23178807947019867,
      "grad_norm": 2.695943182177301,
      "learning_rate": 5.2499999999999995e-05,
      "loss": 2.9375,
      "step": 350
    },
    {
      "epoch": 0.2324503311258278,
      "grad_norm": 3.0932176133525915,
      "learning_rate": 5.264999999999999e-05,
      "loss": 2.9844,
      "step": 351
    },
    {
      "epoch": 0.23311258278145697,
      "grad_norm": 2.8776369927511847,
      "learning_rate": 5.279999999999999e-05,
      "loss": 3.0938,
      "step": 352
    },
    {
      "epoch": 0.2337748344370861,
      "grad_norm": 2.9391023537205987,
      "learning_rate": 5.294999999999999e-05,
      "loss": 3.0781,
      "step": 353
    },
    {
      "epoch": 0.23443708609271524,
      "grad_norm": 2.8236525368121788,
      "learning_rate": 5.309999999999999e-05,
      "loss": 3.0625,
      "step": 354
    },
    {
      "epoch": 0.23509933774834438,
      "grad_norm": 2.5022637128814464,
      "learning_rate": 5.324999999999999e-05,
      "loss": 2.9688,
      "step": 355
    },
    {
      "epoch": 0.23576158940397351,
      "grad_norm": 2.595956447908909,
      "learning_rate": 5.339999999999999e-05,
      "loss": 2.75,
      "step": 356
    },
    {
      "epoch": 0.23642384105960265,
      "grad_norm": 2.5911712210683966,
      "learning_rate": 5.3549999999999994e-05,
      "loss": 2.9375,
      "step": 357
    },
    {
      "epoch": 0.2370860927152318,
      "grad_norm": 2.5410609072424797,
      "learning_rate": 5.369999999999999e-05,
      "loss": 2.75,
      "step": 358
    },
    {
      "epoch": 0.23774834437086093,
      "grad_norm": 2.5520508356259692,
      "learning_rate": 5.3849999999999994e-05,
      "loss": 2.9375,
      "step": 359
    },
    {
      "epoch": 0.23841059602649006,
      "grad_norm": 2.81119890515144,
      "learning_rate": 5.399999999999999e-05,
      "loss": 3.0312,
      "step": 360
    },
    {
      "epoch": 0.2390728476821192,
      "grad_norm": 2.3977756283972234,
      "learning_rate": 5.4149999999999995e-05,
      "loss": 2.5781,
      "step": 361
    },
    {
      "epoch": 0.23973509933774834,
      "grad_norm": 2.439623785981216,
      "learning_rate": 5.429999999999999e-05,
      "loss": 2.8594,
      "step": 362
    },
    {
      "epoch": 0.24039735099337747,
      "grad_norm": 2.5215218942203017,
      "learning_rate": 5.4449999999999995e-05,
      "loss": 2.8281,
      "step": 363
    },
    {
      "epoch": 0.2410596026490066,
      "grad_norm": 2.4832011076478246,
      "learning_rate": 5.459999999999999e-05,
      "loss": 2.6719,
      "step": 364
    },
    {
      "epoch": 0.24172185430463577,
      "grad_norm": 2.6187938721609467,
      "learning_rate": 5.4749999999999996e-05,
      "loss": 2.9688,
      "step": 365
    },
    {
      "epoch": 0.2423841059602649,
      "grad_norm": 2.440141000241782,
      "learning_rate": 5.489999999999999e-05,
      "loss": 2.9062,
      "step": 366
    },
    {
      "epoch": 0.24304635761589405,
      "grad_norm": 2.3704319515535897,
      "learning_rate": 5.5049999999999996e-05,
      "loss": 2.8594,
      "step": 367
    },
    {
      "epoch": 0.24370860927152319,
      "grad_norm": 2.4838728753813903,
      "learning_rate": 5.519999999999999e-05,
      "loss": 2.9844,
      "step": 368
    },
    {
      "epoch": 0.24437086092715232,
      "grad_norm": 2.629678501417108,
      "learning_rate": 5.535e-05,
      "loss": 2.9375,
      "step": 369
    },
    {
      "epoch": 0.24503311258278146,
      "grad_norm": 2.3583989808916725,
      "learning_rate": 5.5499999999999994e-05,
      "loss": 2.8125,
      "step": 370
    },
    {
      "epoch": 0.2456953642384106,
      "grad_norm": 2.7176064872795704,
      "learning_rate": 5.565e-05,
      "loss": 2.9688,
      "step": 371
    },
    {
      "epoch": 0.24635761589403973,
      "grad_norm": 2.515644935973342,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 3.0781,
      "step": 372
    },
    {
      "epoch": 0.24701986754966887,
      "grad_norm": 2.785493564497169,
      "learning_rate": 5.595e-05,
      "loss": 2.9531,
      "step": 373
    },
    {
      "epoch": 0.247682119205298,
      "grad_norm": 2.5610668919549835,
      "learning_rate": 5.6099999999999995e-05,
      "loss": 2.875,
      "step": 374
    },
    {
      "epoch": 0.24834437086092714,
      "grad_norm": 2.3132216863861528,
      "learning_rate": 5.625e-05,
      "loss": 2.5469,
      "step": 375
    },
    {
      "epoch": 0.24900662251655628,
      "grad_norm": 2.490089384614338,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 2.625,
      "step": 376
    },
    {
      "epoch": 0.24966887417218542,
      "grad_norm": 2.732437840492482,
      "learning_rate": 5.654999999999999e-05,
      "loss": 3.1406,
      "step": 377
    },
    {
      "epoch": 0.2503311258278146,
      "grad_norm": 2.4087218496183747,
      "learning_rate": 5.6699999999999996e-05,
      "loss": 2.8438,
      "step": 378
    },
    {
      "epoch": 0.2509933774834437,
      "grad_norm": 2.643332835501332,
      "learning_rate": 5.684999999999999e-05,
      "loss": 3.0156,
      "step": 379
    },
    {
      "epoch": 0.25165562913907286,
      "grad_norm": 2.555866937201364,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 3.0781,
      "step": 380
    },
    {
      "epoch": 0.25231788079470197,
      "grad_norm": 2.505409022852811,
      "learning_rate": 5.714999999999999e-05,
      "loss": 2.8906,
      "step": 381
    },
    {
      "epoch": 0.25298013245033113,
      "grad_norm": 2.5153381688711134,
      "learning_rate": 5.73e-05,
      "loss": 2.75,
      "step": 382
    },
    {
      "epoch": 0.25364238410596024,
      "grad_norm": 2.4545178566050176,
      "learning_rate": 5.7449999999999994e-05,
      "loss": 2.8594,
      "step": 383
    },
    {
      "epoch": 0.2543046357615894,
      "grad_norm": 2.7056541046271767,
      "learning_rate": 5.76e-05,
      "loss": 2.6875,
      "step": 384
    },
    {
      "epoch": 0.25496688741721857,
      "grad_norm": 2.404529635277738,
      "learning_rate": 5.7749999999999994e-05,
      "loss": 2.875,
      "step": 385
    },
    {
      "epoch": 0.2556291390728477,
      "grad_norm": 2.583284337641058,
      "learning_rate": 5.79e-05,
      "loss": 2.8906,
      "step": 386
    },
    {
      "epoch": 0.25629139072847684,
      "grad_norm": 2.708736300949351,
      "learning_rate": 5.8049999999999995e-05,
      "loss": 3.0312,
      "step": 387
    },
    {
      "epoch": 0.25695364238410595,
      "grad_norm": 2.6099765912735973,
      "learning_rate": 5.82e-05,
      "loss": 2.4844,
      "step": 388
    },
    {
      "epoch": 0.2576158940397351,
      "grad_norm": 2.36515435616245,
      "learning_rate": 5.8349999999999995e-05,
      "loss": 2.7656,
      "step": 389
    },
    {
      "epoch": 0.2582781456953642,
      "grad_norm": 2.509318371395626,
      "learning_rate": 5.85e-05,
      "loss": 2.9219,
      "step": 390
    },
    {
      "epoch": 0.2589403973509934,
      "grad_norm": 2.6817505049098,
      "learning_rate": 5.8649999999999996e-05,
      "loss": 2.8438,
      "step": 391
    },
    {
      "epoch": 0.2596026490066225,
      "grad_norm": 2.3377568467031247,
      "learning_rate": 5.88e-05,
      "loss": 2.8281,
      "step": 392
    },
    {
      "epoch": 0.26026490066225166,
      "grad_norm": 2.668249930337885,
      "learning_rate": 5.8949999999999996e-05,
      "loss": 3.125,
      "step": 393
    },
    {
      "epoch": 0.2609271523178808,
      "grad_norm": 2.514411634505482,
      "learning_rate": 5.91e-05,
      "loss": 2.9375,
      "step": 394
    },
    {
      "epoch": 0.26158940397350994,
      "grad_norm": 2.359331343947323,
      "learning_rate": 5.925e-05,
      "loss": 2.9375,
      "step": 395
    },
    {
      "epoch": 0.26225165562913905,
      "grad_norm": 2.487143418586103,
      "learning_rate": 5.94e-05,
      "loss": 2.875,
      "step": 396
    },
    {
      "epoch": 0.2629139072847682,
      "grad_norm": 2.407214953122946,
      "learning_rate": 5.955e-05,
      "loss": 2.9531,
      "step": 397
    },
    {
      "epoch": 0.2635761589403974,
      "grad_norm": 2.5547282744195883,
      "learning_rate": 5.97e-05,
      "loss": 3.0156,
      "step": 398
    },
    {
      "epoch": 0.2642384105960265,
      "grad_norm": 2.3445270528639246,
      "learning_rate": 5.985e-05,
      "loss": 2.8281,
      "step": 399
    },
    {
      "epoch": 0.26490066225165565,
      "grad_norm": 2.2435710612762785,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 2.8438,
      "step": 400
    },
    {
      "epoch": 0.26556291390728476,
      "grad_norm": 2.411873316003341,
      "learning_rate": 6.015e-05,
      "loss": 2.7969,
      "step": 401
    },
    {
      "epoch": 0.2662251655629139,
      "grad_norm": 2.262658715287052,
      "learning_rate": 6.0299999999999995e-05,
      "loss": 3.0,
      "step": 402
    },
    {
      "epoch": 0.26688741721854303,
      "grad_norm": 2.320253164823349,
      "learning_rate": 6.045e-05,
      "loss": 2.9219,
      "step": 403
    },
    {
      "epoch": 0.2675496688741722,
      "grad_norm": 2.42375862691307,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 2.9844,
      "step": 404
    },
    {
      "epoch": 0.2682119205298013,
      "grad_norm": 2.403873454275729,
      "learning_rate": 6.075e-05,
      "loss": 3.0312,
      "step": 405
    },
    {
      "epoch": 0.26887417218543047,
      "grad_norm": 2.463958278863841,
      "learning_rate": 6.0899999999999996e-05,
      "loss": 2.5,
      "step": 406
    },
    {
      "epoch": 0.2695364238410596,
      "grad_norm": 2.5748882636255037,
      "learning_rate": 6.104999999999999e-05,
      "loss": 2.7812,
      "step": 407
    },
    {
      "epoch": 0.27019867549668874,
      "grad_norm": 2.847270850293623,
      "learning_rate": 6.12e-05,
      "loss": 3.125,
      "step": 408
    },
    {
      "epoch": 0.27086092715231785,
      "grad_norm": 2.424187088690875,
      "learning_rate": 6.134999999999999e-05,
      "loss": 2.5938,
      "step": 409
    },
    {
      "epoch": 0.271523178807947,
      "grad_norm": 2.3012265403934054,
      "learning_rate": 6.149999999999999e-05,
      "loss": 3.0156,
      "step": 410
    },
    {
      "epoch": 0.2721854304635762,
      "grad_norm": 2.4615639529933193,
      "learning_rate": 6.165e-05,
      "loss": 2.7969,
      "step": 411
    },
    {
      "epoch": 0.2728476821192053,
      "grad_norm": 2.288145587908721,
      "learning_rate": 6.18e-05,
      "loss": 2.875,
      "step": 412
    },
    {
      "epoch": 0.27350993377483446,
      "grad_norm": 2.3886795870975153,
      "learning_rate": 6.194999999999999e-05,
      "loss": 2.8438,
      "step": 413
    },
    {
      "epoch": 0.27417218543046357,
      "grad_norm": 2.2443562936132815,
      "learning_rate": 6.209999999999999e-05,
      "loss": 2.4219,
      "step": 414
    },
    {
      "epoch": 0.27483443708609273,
      "grad_norm": 2.332050495769141,
      "learning_rate": 6.225e-05,
      "loss": 2.9688,
      "step": 415
    },
    {
      "epoch": 0.27549668874172184,
      "grad_norm": 2.40828791389515,
      "learning_rate": 6.239999999999999e-05,
      "loss": 2.4375,
      "step": 416
    },
    {
      "epoch": 0.276158940397351,
      "grad_norm": 2.3452239662126595,
      "learning_rate": 6.254999999999999e-05,
      "loss": 3.0,
      "step": 417
    },
    {
      "epoch": 0.2768211920529801,
      "grad_norm": 2.3501313210712977,
      "learning_rate": 6.269999999999999e-05,
      "loss": 2.9375,
      "step": 418
    },
    {
      "epoch": 0.2774834437086093,
      "grad_norm": 2.4994458958900267,
      "learning_rate": 6.285e-05,
      "loss": 2.9219,
      "step": 419
    },
    {
      "epoch": 0.2781456953642384,
      "grad_norm": 2.3765244843110973,
      "learning_rate": 6.299999999999999e-05,
      "loss": 2.9219,
      "step": 420
    },
    {
      "epoch": 0.27880794701986755,
      "grad_norm": 2.551917170430421,
      "learning_rate": 6.314999999999999e-05,
      "loss": 2.7344,
      "step": 421
    },
    {
      "epoch": 0.27947019867549666,
      "grad_norm": 2.3140198073766283,
      "learning_rate": 6.33e-05,
      "loss": 2.9531,
      "step": 422
    },
    {
      "epoch": 0.2801324503311258,
      "grad_norm": 2.327598517584333,
      "learning_rate": 6.345e-05,
      "loss": 2.6094,
      "step": 423
    },
    {
      "epoch": 0.280794701986755,
      "grad_norm": 2.447317426935946,
      "learning_rate": 6.359999999999999e-05,
      "loss": 2.8281,
      "step": 424
    },
    {
      "epoch": 0.2814569536423841,
      "grad_norm": 2.1877708252048818,
      "learning_rate": 6.374999999999999e-05,
      "loss": 2.8906,
      "step": 425
    },
    {
      "epoch": 0.28211920529801326,
      "grad_norm": 2.3467691452449744,
      "learning_rate": 6.39e-05,
      "loss": 2.7031,
      "step": 426
    },
    {
      "epoch": 0.2827814569536424,
      "grad_norm": 2.3848096996238035,
      "learning_rate": 6.405e-05,
      "loss": 3.0625,
      "step": 427
    },
    {
      "epoch": 0.28344370860927154,
      "grad_norm": 2.5789295552069023,
      "learning_rate": 6.419999999999999e-05,
      "loss": 3.0469,
      "step": 428
    },
    {
      "epoch": 0.28410596026490065,
      "grad_norm": 2.4515640818058957,
      "learning_rate": 6.434999999999999e-05,
      "loss": 2.9844,
      "step": 429
    },
    {
      "epoch": 0.2847682119205298,
      "grad_norm": 2.394402167936939,
      "learning_rate": 6.45e-05,
      "loss": 2.8281,
      "step": 430
    },
    {
      "epoch": 0.2854304635761589,
      "grad_norm": 2.194167494650524,
      "learning_rate": 6.465e-05,
      "loss": 2.9375,
      "step": 431
    },
    {
      "epoch": 0.2860927152317881,
      "grad_norm": 2.359277627825071,
      "learning_rate": 6.479999999999999e-05,
      "loss": 2.9375,
      "step": 432
    },
    {
      "epoch": 0.2867549668874172,
      "grad_norm": 2.247107450667477,
      "learning_rate": 6.494999999999999e-05,
      "loss": 3.0,
      "step": 433
    },
    {
      "epoch": 0.28741721854304636,
      "grad_norm": 2.2533850982167114,
      "learning_rate": 6.51e-05,
      "loss": 2.6719,
      "step": 434
    },
    {
      "epoch": 0.28807947019867547,
      "grad_norm": 2.6216883376256006,
      "learning_rate": 6.525e-05,
      "loss": 2.9531,
      "step": 435
    },
    {
      "epoch": 0.28874172185430463,
      "grad_norm": 2.3376819223729175,
      "learning_rate": 6.539999999999999e-05,
      "loss": 2.8281,
      "step": 436
    },
    {
      "epoch": 0.2894039735099338,
      "grad_norm": 2.518200368497074,
      "learning_rate": 6.555e-05,
      "loss": 2.9062,
      "step": 437
    },
    {
      "epoch": 0.2900662251655629,
      "grad_norm": 2.4745855317949434,
      "learning_rate": 6.57e-05,
      "loss": 2.9062,
      "step": 438
    },
    {
      "epoch": 0.29072847682119207,
      "grad_norm": 2.4856170019322077,
      "learning_rate": 6.584999999999999e-05,
      "loss": 2.875,
      "step": 439
    },
    {
      "epoch": 0.2913907284768212,
      "grad_norm": 2.4507140192218078,
      "learning_rate": 6.599999999999999e-05,
      "loss": 3.1875,
      "step": 440
    },
    {
      "epoch": 0.29205298013245035,
      "grad_norm": 2.241419553933538,
      "learning_rate": 6.615e-05,
      "loss": 2.6875,
      "step": 441
    },
    {
      "epoch": 0.29271523178807946,
      "grad_norm": 2.194376983134961,
      "learning_rate": 6.63e-05,
      "loss": 2.7812,
      "step": 442
    },
    {
      "epoch": 0.2933774834437086,
      "grad_norm": 2.1216090492917696,
      "learning_rate": 6.644999999999999e-05,
      "loss": 2.8438,
      "step": 443
    },
    {
      "epoch": 0.29403973509933773,
      "grad_norm": 2.411637276719395,
      "learning_rate": 6.659999999999999e-05,
      "loss": 2.5469,
      "step": 444
    },
    {
      "epoch": 0.2947019867549669,
      "grad_norm": 2.178699410808115,
      "learning_rate": 6.675e-05,
      "loss": 2.8594,
      "step": 445
    },
    {
      "epoch": 0.295364238410596,
      "grad_norm": 2.295636232653236,
      "learning_rate": 6.69e-05,
      "loss": 2.9375,
      "step": 446
    },
    {
      "epoch": 0.29602649006622517,
      "grad_norm": 2.3053502822067045,
      "learning_rate": 6.704999999999999e-05,
      "loss": 2.9844,
      "step": 447
    },
    {
      "epoch": 0.29668874172185433,
      "grad_norm": 2.3678044953235395,
      "learning_rate": 6.72e-05,
      "loss": 2.9062,
      "step": 448
    },
    {
      "epoch": 0.29735099337748344,
      "grad_norm": 2.2617717075948303,
      "learning_rate": 6.735e-05,
      "loss": 2.4219,
      "step": 449
    },
    {
      "epoch": 0.2980132450331126,
      "grad_norm": 2.269047766376487,
      "learning_rate": 6.75e-05,
      "loss": 3.0156,
      "step": 450
    },
    {
      "epoch": 0.2986754966887417,
      "grad_norm": 2.2524481369608935,
      "learning_rate": 6.764999999999999e-05,
      "loss": 2.875,
      "step": 451
    },
    {
      "epoch": 0.2993377483443709,
      "grad_norm": 2.4434987542996986,
      "learning_rate": 6.78e-05,
      "loss": 2.9688,
      "step": 452
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3192386902318263,
      "learning_rate": 6.795e-05,
      "loss": 2.8594,
      "step": 453
    },
    {
      "epoch": 0.30066225165562915,
      "grad_norm": 2.48875337721921,
      "learning_rate": 6.81e-05,
      "loss": 2.7969,
      "step": 454
    },
    {
      "epoch": 0.30132450331125826,
      "grad_norm": 2.362226242785205,
      "learning_rate": 6.824999999999999e-05,
      "loss": 2.9844,
      "step": 455
    },
    {
      "epoch": 0.3019867549668874,
      "grad_norm": 2.2670149552987824,
      "learning_rate": 6.84e-05,
      "loss": 2.7188,
      "step": 456
    },
    {
      "epoch": 0.30264900662251654,
      "grad_norm": 2.8193079158488255,
      "learning_rate": 6.855e-05,
      "loss": 3.1406,
      "step": 457
    },
    {
      "epoch": 0.3033112582781457,
      "grad_norm": 2.2646496189624985,
      "learning_rate": 6.87e-05,
      "loss": 2.5156,
      "step": 458
    },
    {
      "epoch": 0.3039735099337748,
      "grad_norm": 2.179387864370677,
      "learning_rate": 6.884999999999999e-05,
      "loss": 2.5156,
      "step": 459
    },
    {
      "epoch": 0.304635761589404,
      "grad_norm": 2.165880982658605,
      "learning_rate": 6.9e-05,
      "loss": 2.75,
      "step": 460
    },
    {
      "epoch": 0.30529801324503314,
      "grad_norm": 2.332414974818671,
      "learning_rate": 6.915e-05,
      "loss": 2.9375,
      "step": 461
    },
    {
      "epoch": 0.30596026490066225,
      "grad_norm": 2.1893904582741524,
      "learning_rate": 6.93e-05,
      "loss": 2.4688,
      "step": 462
    },
    {
      "epoch": 0.3066225165562914,
      "grad_norm": 2.221098659049963,
      "learning_rate": 6.945e-05,
      "loss": 2.8438,
      "step": 463
    },
    {
      "epoch": 0.3072847682119205,
      "grad_norm": 2.407181623611898,
      "learning_rate": 6.96e-05,
      "loss": 2.9531,
      "step": 464
    },
    {
      "epoch": 0.3079470198675497,
      "grad_norm": 2.5102658872763457,
      "learning_rate": 6.975e-05,
      "loss": 2.8281,
      "step": 465
    },
    {
      "epoch": 0.3086092715231788,
      "grad_norm": 2.169804910319373,
      "learning_rate": 6.989999999999999e-05,
      "loss": 2.7812,
      "step": 466
    },
    {
      "epoch": 0.30927152317880796,
      "grad_norm": 2.2954365732232294,
      "learning_rate": 7.005e-05,
      "loss": 2.7188,
      "step": 467
    },
    {
      "epoch": 0.30993377483443707,
      "grad_norm": 2.1603254104763354,
      "learning_rate": 7.02e-05,
      "loss": 2.625,
      "step": 468
    },
    {
      "epoch": 0.31059602649006623,
      "grad_norm": 2.132158855452012,
      "learning_rate": 7.034999999999999e-05,
      "loss": 2.3125,
      "step": 469
    },
    {
      "epoch": 0.31125827814569534,
      "grad_norm": 2.36494053243161,
      "learning_rate": 7.049999999999999e-05,
      "loss": 2.6719,
      "step": 470
    },
    {
      "epoch": 0.3119205298013245,
      "grad_norm": 2.6763288687775164,
      "learning_rate": 7.065e-05,
      "loss": 2.9844,
      "step": 471
    },
    {
      "epoch": 0.3125827814569536,
      "grad_norm": 2.0562046255180033,
      "learning_rate": 7.079999999999999e-05,
      "loss": 2.3125,
      "step": 472
    },
    {
      "epoch": 0.3132450331125828,
      "grad_norm": 2.597697997536415,
      "learning_rate": 7.094999999999999e-05,
      "loss": 3.0625,
      "step": 473
    },
    {
      "epoch": 0.31390728476821195,
      "grad_norm": 2.6390175110906844,
      "learning_rate": 7.11e-05,
      "loss": 2.9219,
      "step": 474
    },
    {
      "epoch": 0.31456953642384106,
      "grad_norm": 2.2129669633402806,
      "learning_rate": 7.125e-05,
      "loss": 2.875,
      "step": 475
    },
    {
      "epoch": 0.3152317880794702,
      "grad_norm": 2.1135558621950414,
      "learning_rate": 7.139999999999999e-05,
      "loss": 2.375,
      "step": 476
    },
    {
      "epoch": 0.31589403973509933,
      "grad_norm": 2.482459306148282,
      "learning_rate": 7.154999999999999e-05,
      "loss": 2.8906,
      "step": 477
    },
    {
      "epoch": 0.3165562913907285,
      "grad_norm": 2.336344546182186,
      "learning_rate": 7.17e-05,
      "loss": 2.8906,
      "step": 478
    },
    {
      "epoch": 0.3172185430463576,
      "grad_norm": 2.3879979269272202,
      "learning_rate": 7.184999999999998e-05,
      "loss": 3.0938,
      "step": 479
    },
    {
      "epoch": 0.31788079470198677,
      "grad_norm": 2.4547318137649614,
      "learning_rate": 7.199999999999999e-05,
      "loss": 2.9062,
      "step": 480
    },
    {
      "epoch": 0.3185430463576159,
      "grad_norm": 2.084216959471573,
      "learning_rate": 7.214999999999999e-05,
      "loss": 2.8125,
      "step": 481
    },
    {
      "epoch": 0.31920529801324504,
      "grad_norm": 2.0090281660434,
      "learning_rate": 7.23e-05,
      "loss": 2.4844,
      "step": 482
    },
    {
      "epoch": 0.31986754966887415,
      "grad_norm": 2.367565101489103,
      "learning_rate": 7.244999999999999e-05,
      "loss": 3.0781,
      "step": 483
    },
    {
      "epoch": 0.3205298013245033,
      "grad_norm": 2.0099732200872413,
      "learning_rate": 7.259999999999999e-05,
      "loss": 2.8281,
      "step": 484
    },
    {
      "epoch": 0.3211920529801324,
      "grad_norm": 2.0290961181524034,
      "learning_rate": 7.274999999999999e-05,
      "loss": 2.7969,
      "step": 485
    },
    {
      "epoch": 0.3218543046357616,
      "grad_norm": 2.136064762662133,
      "learning_rate": 7.29e-05,
      "loss": 2.6875,
      "step": 486
    },
    {
      "epoch": 0.32251655629139075,
      "grad_norm": 2.139831585913877,
      "learning_rate": 7.304999999999999e-05,
      "loss": 2.75,
      "step": 487
    },
    {
      "epoch": 0.32317880794701986,
      "grad_norm": 2.1142817397937264,
      "learning_rate": 7.319999999999999e-05,
      "loss": 2.7188,
      "step": 488
    },
    {
      "epoch": 0.32384105960264903,
      "grad_norm": 2.2499956028241455,
      "learning_rate": 7.335e-05,
      "loss": 2.8594,
      "step": 489
    },
    {
      "epoch": 0.32450331125827814,
      "grad_norm": 2.3426591385386315,
      "learning_rate": 7.35e-05,
      "loss": 2.8281,
      "step": 490
    },
    {
      "epoch": 0.3251655629139073,
      "grad_norm": 2.0769202460708986,
      "learning_rate": 7.364999999999999e-05,
      "loss": 2.7969,
      "step": 491
    },
    {
      "epoch": 0.3258278145695364,
      "grad_norm": 2.418156136483668,
      "learning_rate": 7.379999999999999e-05,
      "loss": 3.125,
      "step": 492
    },
    {
      "epoch": 0.3264900662251656,
      "grad_norm": 2.1335328743175723,
      "learning_rate": 7.395e-05,
      "loss": 2.25,
      "step": 493
    },
    {
      "epoch": 0.3271523178807947,
      "grad_norm": 2.0070473888666807,
      "learning_rate": 7.41e-05,
      "loss": 2.7812,
      "step": 494
    },
    {
      "epoch": 0.32781456953642385,
      "grad_norm": 2.359416477826231,
      "learning_rate": 7.424999999999999e-05,
      "loss": 2.8906,
      "step": 495
    },
    {
      "epoch": 0.32847682119205296,
      "grad_norm": 2.21706296665862,
      "learning_rate": 7.439999999999999e-05,
      "loss": 2.8125,
      "step": 496
    },
    {
      "epoch": 0.3291390728476821,
      "grad_norm": 2.4261006835004166,
      "learning_rate": 7.455e-05,
      "loss": 2.8906,
      "step": 497
    },
    {
      "epoch": 0.32980132450331123,
      "grad_norm": 1.9729049797261617,
      "learning_rate": 7.47e-05,
      "loss": 2.5312,
      "step": 498
    },
    {
      "epoch": 0.3304635761589404,
      "grad_norm": 2.150616973668717,
      "learning_rate": 7.484999999999999e-05,
      "loss": 2.8594,
      "step": 499
    },
    {
      "epoch": 0.33112582781456956,
      "grad_norm": 2.0215825279015225,
      "learning_rate": 7.5e-05,
      "loss": 2.4062,
      "step": 500
    },
    {
      "epoch": 0.33178807947019867,
      "grad_norm": 2.144181930845469,
      "learning_rate": 7.515e-05,
      "loss": 2.8125,
      "step": 501
    },
    {
      "epoch": 0.33245033112582784,
      "grad_norm": 2.2225352152596427,
      "learning_rate": 7.529999999999999e-05,
      "loss": 2.7812,
      "step": 502
    },
    {
      "epoch": 0.33311258278145695,
      "grad_norm": 2.1853496502619825,
      "learning_rate": 7.544999999999999e-05,
      "loss": 2.9062,
      "step": 503
    },
    {
      "epoch": 0.3337748344370861,
      "grad_norm": 2.057238568866444,
      "learning_rate": 7.56e-05,
      "loss": 2.875,
      "step": 504
    },
    {
      "epoch": 0.3344370860927152,
      "grad_norm": 2.3800183087172115,
      "learning_rate": 7.575e-05,
      "loss": 2.7344,
      "step": 505
    },
    {
      "epoch": 0.3350993377483444,
      "grad_norm": 2.1238206723295345,
      "learning_rate": 7.589999999999999e-05,
      "loss": 3.0781,
      "step": 506
    },
    {
      "epoch": 0.3357615894039735,
      "grad_norm": 2.4227058139679745,
      "learning_rate": 7.604999999999999e-05,
      "loss": 2.4844,
      "step": 507
    },
    {
      "epoch": 0.33642384105960266,
      "grad_norm": 2.4387307938851692,
      "learning_rate": 7.62e-05,
      "loss": 2.9531,
      "step": 508
    },
    {
      "epoch": 0.33708609271523177,
      "grad_norm": 2.1937530823153413,
      "learning_rate": 7.635e-05,
      "loss": 2.7031,
      "step": 509
    },
    {
      "epoch": 0.33774834437086093,
      "grad_norm": 2.4708036750827462,
      "learning_rate": 7.649999999999999e-05,
      "loss": 3.1094,
      "step": 510
    },
    {
      "epoch": 0.33841059602649004,
      "grad_norm": 2.290066860117554,
      "learning_rate": 7.664999999999999e-05,
      "loss": 3.0312,
      "step": 511
    },
    {
      "epoch": 0.3390728476821192,
      "grad_norm": 2.188103534587859,
      "learning_rate": 7.68e-05,
      "loss": 2.75,
      "step": 512
    },
    {
      "epoch": 0.33973509933774837,
      "grad_norm": 2.025423983041444,
      "learning_rate": 7.695e-05,
      "loss": 2.6562,
      "step": 513
    },
    {
      "epoch": 0.3403973509933775,
      "grad_norm": 2.0366991940625905,
      "learning_rate": 7.709999999999999e-05,
      "loss": 2.9219,
      "step": 514
    },
    {
      "epoch": 0.34105960264900664,
      "grad_norm": 2.2736736955011247,
      "learning_rate": 7.725e-05,
      "loss": 2.9062,
      "step": 515
    },
    {
      "epoch": 0.34172185430463575,
      "grad_norm": 2.052252999434754,
      "learning_rate": 7.74e-05,
      "loss": 2.75,
      "step": 516
    },
    {
      "epoch": 0.3423841059602649,
      "grad_norm": 2.1958190937817452,
      "learning_rate": 7.755e-05,
      "loss": 2.25,
      "step": 517
    },
    {
      "epoch": 0.343046357615894,
      "grad_norm": 2.40519867353714,
      "learning_rate": 7.769999999999999e-05,
      "loss": 2.8906,
      "step": 518
    },
    {
      "epoch": 0.3437086092715232,
      "grad_norm": 2.1762621592660647,
      "learning_rate": 7.785e-05,
      "loss": 2.75,
      "step": 519
    },
    {
      "epoch": 0.3443708609271523,
      "grad_norm": 2.118278409709848,
      "learning_rate": 7.8e-05,
      "loss": 2.4688,
      "step": 520
    },
    {
      "epoch": 0.34503311258278146,
      "grad_norm": 2.546749299367953,
      "learning_rate": 7.815e-05,
      "loss": 2.9844,
      "step": 521
    },
    {
      "epoch": 0.3456953642384106,
      "grad_norm": 1.938743046755281,
      "learning_rate": 7.829999999999999e-05,
      "loss": 2.7031,
      "step": 522
    },
    {
      "epoch": 0.34635761589403974,
      "grad_norm": 2.42018310611329,
      "learning_rate": 7.845e-05,
      "loss": 2.9062,
      "step": 523
    },
    {
      "epoch": 0.34701986754966885,
      "grad_norm": 2.1081235927722166,
      "learning_rate": 7.86e-05,
      "loss": 2.9375,
      "step": 524
    },
    {
      "epoch": 0.347682119205298,
      "grad_norm": 2.228552809018641,
      "learning_rate": 7.874999999999999e-05,
      "loss": 2.9844,
      "step": 525
    },
    {
      "epoch": 0.3483443708609272,
      "grad_norm": 2.1421400778159914,
      "learning_rate": 7.89e-05,
      "loss": 2.4375,
      "step": 526
    },
    {
      "epoch": 0.3490066225165563,
      "grad_norm": 2.0156998292333856,
      "learning_rate": 7.905e-05,
      "loss": 2.2812,
      "step": 527
    },
    {
      "epoch": 0.34966887417218545,
      "grad_norm": 2.033857679663199,
      "learning_rate": 7.92e-05,
      "loss": 2.625,
      "step": 528
    },
    {
      "epoch": 0.35033112582781456,
      "grad_norm": 2.3222847795850883,
      "learning_rate": 7.934999999999999e-05,
      "loss": 2.9688,
      "step": 529
    },
    {
      "epoch": 0.3509933774834437,
      "grad_norm": 2.2444800210312166,
      "learning_rate": 7.95e-05,
      "loss": 2.9531,
      "step": 530
    },
    {
      "epoch": 0.35165562913907283,
      "grad_norm": 2.098809012772908,
      "learning_rate": 7.965e-05,
      "loss": 2.9062,
      "step": 531
    },
    {
      "epoch": 0.352317880794702,
      "grad_norm": 2.286347529024447,
      "learning_rate": 7.98e-05,
      "loss": 3.1875,
      "step": 532
    },
    {
      "epoch": 0.3529801324503311,
      "grad_norm": 2.1173668810192936,
      "learning_rate": 7.994999999999999e-05,
      "loss": 2.5781,
      "step": 533
    },
    {
      "epoch": 0.3536423841059603,
      "grad_norm": 2.0782603257194676,
      "learning_rate": 8.01e-05,
      "loss": 2.3438,
      "step": 534
    },
    {
      "epoch": 0.3543046357615894,
      "grad_norm": 2.201978072412506,
      "learning_rate": 8.025e-05,
      "loss": 2.8281,
      "step": 535
    },
    {
      "epoch": 0.35496688741721855,
      "grad_norm": 2.3946762770655514,
      "learning_rate": 8.04e-05,
      "loss": 2.7812,
      "step": 536
    },
    {
      "epoch": 0.35562913907284766,
      "grad_norm": 2.001994300905327,
      "learning_rate": 8.054999999999999e-05,
      "loss": 2.7656,
      "step": 537
    },
    {
      "epoch": 0.3562913907284768,
      "grad_norm": 2.1168591409651856,
      "learning_rate": 8.07e-05,
      "loss": 2.9531,
      "step": 538
    },
    {
      "epoch": 0.356953642384106,
      "grad_norm": 2.0907962944096132,
      "learning_rate": 8.085e-05,
      "loss": 2.9375,
      "step": 539
    },
    {
      "epoch": 0.3576158940397351,
      "grad_norm": 2.177488970869635,
      "learning_rate": 8.1e-05,
      "loss": 2.7031,
      "step": 540
    },
    {
      "epoch": 0.35827814569536426,
      "grad_norm": 2.0479111615480243,
      "learning_rate": 8.115e-05,
      "loss": 2.7812,
      "step": 541
    },
    {
      "epoch": 0.35894039735099337,
      "grad_norm": 2.1091162204223797,
      "learning_rate": 8.13e-05,
      "loss": 2.8906,
      "step": 542
    },
    {
      "epoch": 0.35960264900662253,
      "grad_norm": 2.2231684884671528,
      "learning_rate": 8.145e-05,
      "loss": 2.8594,
      "step": 543
    },
    {
      "epoch": 0.36026490066225164,
      "grad_norm": 2.0836665707454096,
      "learning_rate": 8.16e-05,
      "loss": 2.4688,
      "step": 544
    },
    {
      "epoch": 0.3609271523178808,
      "grad_norm": 2.0626081607228595,
      "learning_rate": 8.175e-05,
      "loss": 2.5938,
      "step": 545
    },
    {
      "epoch": 0.3615894039735099,
      "grad_norm": 2.4295458049423004,
      "learning_rate": 8.19e-05,
      "loss": 3.0625,
      "step": 546
    },
    {
      "epoch": 0.3622516556291391,
      "grad_norm": 2.1222122289891088,
      "learning_rate": 8.205e-05,
      "loss": 2.4531,
      "step": 547
    },
    {
      "epoch": 0.3629139072847682,
      "grad_norm": 2.307170306861792,
      "learning_rate": 8.22e-05,
      "loss": 2.8906,
      "step": 548
    },
    {
      "epoch": 0.36357615894039735,
      "grad_norm": 1.9782636664950333,
      "learning_rate": 8.235e-05,
      "loss": 2.8281,
      "step": 549
    },
    {
      "epoch": 0.36423841059602646,
      "grad_norm": 2.1683571917876088,
      "learning_rate": 8.25e-05,
      "loss": 2.375,
      "step": 550
    },
    {
      "epoch": 0.3649006622516556,
      "grad_norm": 1.9704901363214848,
      "learning_rate": 8.265e-05,
      "loss": 2.7188,
      "step": 551
    },
    {
      "epoch": 0.3655629139072848,
      "grad_norm": 2.040234027674569,
      "learning_rate": 8.28e-05,
      "loss": 2.7344,
      "step": 552
    },
    {
      "epoch": 0.3662251655629139,
      "grad_norm": 1.998458962114073,
      "learning_rate": 8.295e-05,
      "loss": 2.8594,
      "step": 553
    },
    {
      "epoch": 0.36688741721854307,
      "grad_norm": 2.168751581349317,
      "learning_rate": 8.31e-05,
      "loss": 2.7969,
      "step": 554
    },
    {
      "epoch": 0.3675496688741722,
      "grad_norm": 2.1855271225047814,
      "learning_rate": 8.325e-05,
      "loss": 2.9219,
      "step": 555
    },
    {
      "epoch": 0.36821192052980134,
      "grad_norm": 1.9517021263998031,
      "learning_rate": 8.34e-05,
      "loss": 2.3594,
      "step": 556
    },
    {
      "epoch": 0.36887417218543045,
      "grad_norm": 2.215857471172879,
      "learning_rate": 8.355e-05,
      "loss": 2.7812,
      "step": 557
    },
    {
      "epoch": 0.3695364238410596,
      "grad_norm": 2.0802274038028807,
      "learning_rate": 8.37e-05,
      "loss": 2.2656,
      "step": 558
    },
    {
      "epoch": 0.3701986754966887,
      "grad_norm": 2.360471645672153,
      "learning_rate": 8.385e-05,
      "loss": 2.8281,
      "step": 559
    },
    {
      "epoch": 0.3708609271523179,
      "grad_norm": 2.129907727372624,
      "learning_rate": 8.4e-05,
      "loss": 2.3906,
      "step": 560
    },
    {
      "epoch": 0.371523178807947,
      "grad_norm": 1.9548570896462385,
      "learning_rate": 8.415e-05,
      "loss": 2.6875,
      "step": 561
    },
    {
      "epoch": 0.37218543046357616,
      "grad_norm": 1.9443511712863943,
      "learning_rate": 8.43e-05,
      "loss": 2.7031,
      "step": 562
    },
    {
      "epoch": 0.37284768211920527,
      "grad_norm": 1.9921331008283054,
      "learning_rate": 8.444999999999998e-05,
      "loss": 2.625,
      "step": 563
    },
    {
      "epoch": 0.37350993377483444,
      "grad_norm": 2.2038111441940798,
      "learning_rate": 8.459999999999998e-05,
      "loss": 2.8281,
      "step": 564
    },
    {
      "epoch": 0.3741721854304636,
      "grad_norm": 2.061644948976771,
      "learning_rate": 8.474999999999999e-05,
      "loss": 2.8125,
      "step": 565
    },
    {
      "epoch": 0.3748344370860927,
      "grad_norm": 2.035210471472948,
      "learning_rate": 8.489999999999999e-05,
      "loss": 2.7031,
      "step": 566
    },
    {
      "epoch": 0.3754966887417219,
      "grad_norm": 2.1134537403824862,
      "learning_rate": 8.504999999999998e-05,
      "loss": 2.7812,
      "step": 567
    },
    {
      "epoch": 0.376158940397351,
      "grad_norm": 1.8678257670327116,
      "learning_rate": 8.519999999999998e-05,
      "loss": 2.375,
      "step": 568
    },
    {
      "epoch": 0.37682119205298015,
      "grad_norm": 2.114658535914786,
      "learning_rate": 8.534999999999999e-05,
      "loss": 2.8438,
      "step": 569
    },
    {
      "epoch": 0.37748344370860926,
      "grad_norm": 2.159830785368374,
      "learning_rate": 8.549999999999999e-05,
      "loss": 2.5938,
      "step": 570
    },
    {
      "epoch": 0.3781456953642384,
      "grad_norm": 2.034177069544218,
      "learning_rate": 8.564999999999998e-05,
      "loss": 2.7031,
      "step": 571
    },
    {
      "epoch": 0.37880794701986753,
      "grad_norm": 2.073626405444586,
      "learning_rate": 8.579999999999998e-05,
      "loss": 2.7656,
      "step": 572
    },
    {
      "epoch": 0.3794701986754967,
      "grad_norm": 2.1168575441688104,
      "learning_rate": 8.594999999999999e-05,
      "loss": 2.5781,
      "step": 573
    },
    {
      "epoch": 0.3801324503311258,
      "grad_norm": 2.169006369823301,
      "learning_rate": 8.609999999999999e-05,
      "loss": 3.0,
      "step": 574
    },
    {
      "epoch": 0.38079470198675497,
      "grad_norm": 2.3854472275404786,
      "learning_rate": 8.624999999999998e-05,
      "loss": 2.7656,
      "step": 575
    },
    {
      "epoch": 0.38145695364238413,
      "grad_norm": 2.0804248612595466,
      "learning_rate": 8.639999999999999e-05,
      "loss": 2.7344,
      "step": 576
    },
    {
      "epoch": 0.38211920529801324,
      "grad_norm": 2.152225440715124,
      "learning_rate": 8.654999999999999e-05,
      "loss": 2.6719,
      "step": 577
    },
    {
      "epoch": 0.3827814569536424,
      "grad_norm": 1.9819913756203338,
      "learning_rate": 8.669999999999998e-05,
      "loss": 2.7188,
      "step": 578
    },
    {
      "epoch": 0.3834437086092715,
      "grad_norm": 2.139813687539947,
      "learning_rate": 8.684999999999998e-05,
      "loss": 3.125,
      "step": 579
    },
    {
      "epoch": 0.3841059602649007,
      "grad_norm": 2.166576351014731,
      "learning_rate": 8.699999999999999e-05,
      "loss": 2.75,
      "step": 580
    },
    {
      "epoch": 0.3847682119205298,
      "grad_norm": 4.829014486210198,
      "learning_rate": 8.714999999999999e-05,
      "loss": 3.0469,
      "step": 581
    },
    {
      "epoch": 0.38543046357615895,
      "grad_norm": 2.2544202346196096,
      "learning_rate": 8.729999999999998e-05,
      "loss": 2.6406,
      "step": 582
    },
    {
      "epoch": 0.38609271523178806,
      "grad_norm": 2.446261154077621,
      "learning_rate": 8.744999999999998e-05,
      "loss": 2.9688,
      "step": 583
    },
    {
      "epoch": 0.38675496688741723,
      "grad_norm": 1.9489873398139181,
      "learning_rate": 8.759999999999999e-05,
      "loss": 2.5938,
      "step": 584
    },
    {
      "epoch": 0.38741721854304634,
      "grad_norm": 2.024691193681043,
      "learning_rate": 8.774999999999999e-05,
      "loss": 2.875,
      "step": 585
    },
    {
      "epoch": 0.3880794701986755,
      "grad_norm": 2.389301092223556,
      "learning_rate": 8.789999999999998e-05,
      "loss": 2.8125,
      "step": 586
    },
    {
      "epoch": 0.3887417218543046,
      "grad_norm": 2.049220232738323,
      "learning_rate": 8.804999999999999e-05,
      "loss": 3.0,
      "step": 587
    },
    {
      "epoch": 0.3894039735099338,
      "grad_norm": 2.0842758833328094,
      "learning_rate": 8.819999999999999e-05,
      "loss": 2.6719,
      "step": 588
    },
    {
      "epoch": 0.39006622516556294,
      "grad_norm": 2.054243717019012,
      "learning_rate": 8.834999999999999e-05,
      "loss": 2.375,
      "step": 589
    },
    {
      "epoch": 0.39072847682119205,
      "grad_norm": 2.5472372555582523,
      "learning_rate": 8.849999999999998e-05,
      "loss": 2.9219,
      "step": 590
    },
    {
      "epoch": 0.3913907284768212,
      "grad_norm": 2.0970125602736798,
      "learning_rate": 8.864999999999999e-05,
      "loss": 2.9844,
      "step": 591
    },
    {
      "epoch": 0.3920529801324503,
      "grad_norm": 2.208078333809169,
      "learning_rate": 8.879999999999999e-05,
      "loss": 2.8281,
      "step": 592
    },
    {
      "epoch": 0.3927152317880795,
      "grad_norm": 2.1727502142386035,
      "learning_rate": 8.895e-05,
      "loss": 2.7344,
      "step": 593
    },
    {
      "epoch": 0.3933774834437086,
      "grad_norm": 2.092224867746833,
      "learning_rate": 8.909999999999998e-05,
      "loss": 2.875,
      "step": 594
    },
    {
      "epoch": 0.39403973509933776,
      "grad_norm": 2.0422559691902524,
      "learning_rate": 8.924999999999999e-05,
      "loss": 2.7656,
      "step": 595
    },
    {
      "epoch": 0.39470198675496687,
      "grad_norm": 2.0152291133598395,
      "learning_rate": 8.939999999999999e-05,
      "loss": 2.8594,
      "step": 596
    },
    {
      "epoch": 0.39536423841059604,
      "grad_norm": 2.145609256213379,
      "learning_rate": 8.955e-05,
      "loss": 2.9375,
      "step": 597
    },
    {
      "epoch": 0.39602649006622515,
      "grad_norm": 2.0574829298359547,
      "learning_rate": 8.969999999999998e-05,
      "loss": 3.0,
      "step": 598
    },
    {
      "epoch": 0.3966887417218543,
      "grad_norm": 2.263714036168474,
      "learning_rate": 8.984999999999999e-05,
      "loss": 2.8125,
      "step": 599
    },
    {
      "epoch": 0.3973509933774834,
      "grad_norm": 1.9919958102962712,
      "learning_rate": 8.999999999999999e-05,
      "loss": 2.6875,
      "step": 600
    },
    {
      "epoch": 0.3980132450331126,
      "grad_norm": 2.2260974867140146,
      "learning_rate": 9.014999999999998e-05,
      "loss": 2.6719,
      "step": 601
    },
    {
      "epoch": 0.39867549668874175,
      "grad_norm": 2.18945850177853,
      "learning_rate": 9.029999999999999e-05,
      "loss": 2.7656,
      "step": 602
    },
    {
      "epoch": 0.39933774834437086,
      "grad_norm": 2.069689174843609,
      "learning_rate": 9.044999999999999e-05,
      "loss": 3.0156,
      "step": 603
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9069673794761737,
      "learning_rate": 9.059999999999999e-05,
      "loss": 2.625,
      "step": 604
    },
    {
      "epoch": 0.40066225165562913,
      "grad_norm": 1.9436850056906902,
      "learning_rate": 9.074999999999998e-05,
      "loss": 2.8281,
      "step": 605
    },
    {
      "epoch": 0.4013245033112583,
      "grad_norm": 2.358587861874932,
      "learning_rate": 9.089999999999999e-05,
      "loss": 3.0938,
      "step": 606
    },
    {
      "epoch": 0.4019867549668874,
      "grad_norm": 2.1850120974104845,
      "learning_rate": 9.104999999999999e-05,
      "loss": 2.7656,
      "step": 607
    },
    {
      "epoch": 0.40264900662251657,
      "grad_norm": 2.0374017039032486,
      "learning_rate": 9.12e-05,
      "loss": 2.8281,
      "step": 608
    },
    {
      "epoch": 0.4033112582781457,
      "grad_norm": 1.9800040999475284,
      "learning_rate": 9.134999999999998e-05,
      "loss": 2.7969,
      "step": 609
    },
    {
      "epoch": 0.40397350993377484,
      "grad_norm": 2.0900735210016017,
      "learning_rate": 9.149999999999999e-05,
      "loss": 2.9062,
      "step": 610
    },
    {
      "epoch": 0.40463576158940395,
      "grad_norm": 2.2748193817236704,
      "learning_rate": 9.164999999999999e-05,
      "loss": 2.7812,
      "step": 611
    },
    {
      "epoch": 0.4052980132450331,
      "grad_norm": 2.132043060040571,
      "learning_rate": 9.18e-05,
      "loss": 2.6094,
      "step": 612
    },
    {
      "epoch": 0.4059602649006622,
      "grad_norm": 2.130604006164947,
      "learning_rate": 9.194999999999999e-05,
      "loss": 2.4531,
      "step": 613
    },
    {
      "epoch": 0.4066225165562914,
      "grad_norm": 2.019423383713658,
      "learning_rate": 9.209999999999999e-05,
      "loss": 2.875,
      "step": 614
    },
    {
      "epoch": 0.40728476821192056,
      "grad_norm": 2.4919837549111192,
      "learning_rate": 9.224999999999999e-05,
      "loss": 2.7344,
      "step": 615
    },
    {
      "epoch": 0.40794701986754967,
      "grad_norm": 2.2673189963571776,
      "learning_rate": 9.24e-05,
      "loss": 2.8438,
      "step": 616
    },
    {
      "epoch": 0.40860927152317883,
      "grad_norm": 1.975442301045306,
      "learning_rate": 9.254999999999999e-05,
      "loss": 2.2031,
      "step": 617
    },
    {
      "epoch": 0.40927152317880794,
      "grad_norm": 2.052400141176238,
      "learning_rate": 9.269999999999999e-05,
      "loss": 2.7188,
      "step": 618
    },
    {
      "epoch": 0.4099337748344371,
      "grad_norm": 1.945795507673163,
      "learning_rate": 9.285e-05,
      "loss": 2.7188,
      "step": 619
    },
    {
      "epoch": 0.4105960264900662,
      "grad_norm": 2.0151199729887894,
      "learning_rate": 9.3e-05,
      "loss": 2.6719,
      "step": 620
    },
    {
      "epoch": 0.4112582781456954,
      "grad_norm": 2.4567728770974457,
      "learning_rate": 9.314999999999999e-05,
      "loss": 2.8906,
      "step": 621
    },
    {
      "epoch": 0.4119205298013245,
      "grad_norm": 2.342716049898175,
      "learning_rate": 9.329999999999999e-05,
      "loss": 3.0156,
      "step": 622
    },
    {
      "epoch": 0.41258278145695365,
      "grad_norm": 2.2088217534027397,
      "learning_rate": 9.345e-05,
      "loss": 2.4219,
      "step": 623
    },
    {
      "epoch": 0.41324503311258276,
      "grad_norm": 2.1911966560071474,
      "learning_rate": 9.36e-05,
      "loss": 2.7969,
      "step": 624
    },
    {
      "epoch": 0.4139072847682119,
      "grad_norm": 2.03171666106166,
      "learning_rate": 9.374999999999999e-05,
      "loss": 3.0312,
      "step": 625
    },
    {
      "epoch": 0.41456953642384103,
      "grad_norm": 1.8861248268857203,
      "learning_rate": 9.389999999999999e-05,
      "loss": 2.7188,
      "step": 626
    },
    {
      "epoch": 0.4152317880794702,
      "grad_norm": 1.8014354436451963,
      "learning_rate": 9.405e-05,
      "loss": 2.2812,
      "step": 627
    },
    {
      "epoch": 0.41589403973509936,
      "grad_norm": 2.0901665087319063,
      "learning_rate": 9.419999999999999e-05,
      "loss": 2.9844,
      "step": 628
    },
    {
      "epoch": 0.4165562913907285,
      "grad_norm": 1.9522955103133746,
      "learning_rate": 9.434999999999999e-05,
      "loss": 2.9219,
      "step": 629
    },
    {
      "epoch": 0.41721854304635764,
      "grad_norm": 2.1130816889847677,
      "learning_rate": 9.449999999999999e-05,
      "loss": 2.9219,
      "step": 630
    },
    {
      "epoch": 0.41788079470198675,
      "grad_norm": 2.010240010726036,
      "learning_rate": 9.465e-05,
      "loss": 2.7344,
      "step": 631
    },
    {
      "epoch": 0.4185430463576159,
      "grad_norm": 2.3610163372817152,
      "learning_rate": 9.479999999999999e-05,
      "loss": 2.7656,
      "step": 632
    },
    {
      "epoch": 0.419205298013245,
      "grad_norm": 2.315609038462489,
      "learning_rate": 9.494999999999999e-05,
      "loss": 3.0469,
      "step": 633
    },
    {
      "epoch": 0.4198675496688742,
      "grad_norm": 1.9516267011309307,
      "learning_rate": 9.51e-05,
      "loss": 2.7969,
      "step": 634
    },
    {
      "epoch": 0.4205298013245033,
      "grad_norm": 1.936401379573993,
      "learning_rate": 9.525e-05,
      "loss": 2.5312,
      "step": 635
    },
    {
      "epoch": 0.42119205298013246,
      "grad_norm": 2.164936526837924,
      "learning_rate": 9.539999999999999e-05,
      "loss": 2.8438,
      "step": 636
    },
    {
      "epoch": 0.42185430463576157,
      "grad_norm": 1.811260752899351,
      "learning_rate": 9.554999999999999e-05,
      "loss": 2.6562,
      "step": 637
    },
    {
      "epoch": 0.42251655629139073,
      "grad_norm": 1.9666396458689808,
      "learning_rate": 9.57e-05,
      "loss": 2.7031,
      "step": 638
    },
    {
      "epoch": 0.42317880794701984,
      "grad_norm": 2.0875182774650343,
      "learning_rate": 9.585e-05,
      "loss": 2.9375,
      "step": 639
    },
    {
      "epoch": 0.423841059602649,
      "grad_norm": 2.0909587175752393,
      "learning_rate": 9.599999999999999e-05,
      "loss": 2.75,
      "step": 640
    },
    {
      "epoch": 0.42450331125827817,
      "grad_norm": 2.0494336173868803,
      "learning_rate": 9.614999999999999e-05,
      "loss": 2.4375,
      "step": 641
    },
    {
      "epoch": 0.4251655629139073,
      "grad_norm": 2.2941128104491284,
      "learning_rate": 9.63e-05,
      "loss": 2.6406,
      "step": 642
    },
    {
      "epoch": 0.42582781456953644,
      "grad_norm": 1.8406419411542303,
      "learning_rate": 9.645e-05,
      "loss": 2.7188,
      "step": 643
    },
    {
      "epoch": 0.42649006622516555,
      "grad_norm": 3.072637208363104,
      "learning_rate": 9.659999999999999e-05,
      "loss": 2.2656,
      "step": 644
    },
    {
      "epoch": 0.4271523178807947,
      "grad_norm": 1.9011114934831905,
      "learning_rate": 9.675e-05,
      "loss": 2.5625,
      "step": 645
    },
    {
      "epoch": 0.4278145695364238,
      "grad_norm": 2.162374522265662,
      "learning_rate": 9.69e-05,
      "loss": 2.8125,
      "step": 646
    },
    {
      "epoch": 0.428476821192053,
      "grad_norm": 2.293962730094481,
      "learning_rate": 9.705e-05,
      "loss": 3.0938,
      "step": 647
    },
    {
      "epoch": 0.4291390728476821,
      "grad_norm": 1.980727209921109,
      "learning_rate": 9.719999999999999e-05,
      "loss": 2.6406,
      "step": 648
    },
    {
      "epoch": 0.42980132450331127,
      "grad_norm": 2.0854019802407695,
      "learning_rate": 9.735e-05,
      "loss": 2.875,
      "step": 649
    },
    {
      "epoch": 0.4304635761589404,
      "grad_norm": 2.019699534933072,
      "learning_rate": 9.75e-05,
      "loss": 2.7344,
      "step": 650
    },
    {
      "epoch": 0.43112582781456954,
      "grad_norm": 2.027642675309942,
      "learning_rate": 9.764999999999999e-05,
      "loss": 2.6562,
      "step": 651
    },
    {
      "epoch": 0.43178807947019865,
      "grad_norm": 2.0938822188759825,
      "learning_rate": 9.779999999999999e-05,
      "loss": 2.5312,
      "step": 652
    },
    {
      "epoch": 0.4324503311258278,
      "grad_norm": 1.935918085413364,
      "learning_rate": 9.795e-05,
      "loss": 2.5312,
      "step": 653
    },
    {
      "epoch": 0.433112582781457,
      "grad_norm": 1.9155078370573648,
      "learning_rate": 9.81e-05,
      "loss": 2.5938,
      "step": 654
    },
    {
      "epoch": 0.4337748344370861,
      "grad_norm": 2.026061939320236,
      "learning_rate": 9.824999999999999e-05,
      "loss": 2.5156,
      "step": 655
    },
    {
      "epoch": 0.43443708609271525,
      "grad_norm": 1.8698782681785027,
      "learning_rate": 9.839999999999999e-05,
      "loss": 2.7031,
      "step": 656
    },
    {
      "epoch": 0.43509933774834436,
      "grad_norm": 2.025533528818808,
      "learning_rate": 9.855e-05,
      "loss": 2.8438,
      "step": 657
    },
    {
      "epoch": 0.4357615894039735,
      "grad_norm": 1.9182580412716894,
      "learning_rate": 9.87e-05,
      "loss": 2.8281,
      "step": 658
    },
    {
      "epoch": 0.43642384105960264,
      "grad_norm": 1.7999357845154618,
      "learning_rate": 9.884999999999999e-05,
      "loss": 2.6562,
      "step": 659
    },
    {
      "epoch": 0.4370860927152318,
      "grad_norm": 1.8457615345099856,
      "learning_rate": 9.9e-05,
      "loss": 2.6406,
      "step": 660
    },
    {
      "epoch": 0.4377483443708609,
      "grad_norm": 2.07931221344651,
      "learning_rate": 9.915e-05,
      "loss": 3.0469,
      "step": 661
    },
    {
      "epoch": 0.4384105960264901,
      "grad_norm": 1.8892852393077035,
      "learning_rate": 9.93e-05,
      "loss": 2.5312,
      "step": 662
    },
    {
      "epoch": 0.4390728476821192,
      "grad_norm": 1.8394517245732003,
      "learning_rate": 9.944999999999999e-05,
      "loss": 2.4062,
      "step": 663
    },
    {
      "epoch": 0.43973509933774835,
      "grad_norm": 2.0161302979519653,
      "learning_rate": 9.96e-05,
      "loss": 2.7656,
      "step": 664
    },
    {
      "epoch": 0.44039735099337746,
      "grad_norm": 1.919299333070575,
      "learning_rate": 9.975e-05,
      "loss": 2.7969,
      "step": 665
    },
    {
      "epoch": 0.4410596026490066,
      "grad_norm": 2.0292755557135096,
      "learning_rate": 9.99e-05,
      "loss": 2.8438,
      "step": 666
    },
    {
      "epoch": 0.4417218543046358,
      "grad_norm": 2.1344458842194016,
      "learning_rate": 0.00010004999999999999,
      "loss": 2.8281,
      "step": 667
    },
    {
      "epoch": 0.4423841059602649,
      "grad_norm": 2.117217670341517,
      "learning_rate": 0.0001002,
      "loss": 2.8125,
      "step": 668
    },
    {
      "epoch": 0.44304635761589406,
      "grad_norm": 1.8164444075974442,
      "learning_rate": 0.00010035,
      "loss": 2.5156,
      "step": 669
    },
    {
      "epoch": 0.44370860927152317,
      "grad_norm": 2.0998039410610176,
      "learning_rate": 0.0001005,
      "loss": 2.7656,
      "step": 670
    },
    {
      "epoch": 0.44437086092715233,
      "grad_norm": 2.0638105437895673,
      "learning_rate": 0.00010065,
      "loss": 2.8125,
      "step": 671
    },
    {
      "epoch": 0.44503311258278144,
      "grad_norm": 2.3107447633676332,
      "learning_rate": 0.0001008,
      "loss": 2.3438,
      "step": 672
    },
    {
      "epoch": 0.4456953642384106,
      "grad_norm": 2.1500168990952657,
      "learning_rate": 0.00010095,
      "loss": 2.9531,
      "step": 673
    },
    {
      "epoch": 0.4463576158940397,
      "grad_norm": 2.1030641234995624,
      "learning_rate": 0.0001011,
      "loss": 2.8594,
      "step": 674
    },
    {
      "epoch": 0.4470198675496689,
      "grad_norm": 2.0355334057389696,
      "learning_rate": 0.00010125,
      "loss": 2.7812,
      "step": 675
    },
    {
      "epoch": 0.447682119205298,
      "grad_norm": 1.984890442020896,
      "learning_rate": 0.0001014,
      "loss": 2.7812,
      "step": 676
    },
    {
      "epoch": 0.44834437086092715,
      "grad_norm": 1.9466000003504182,
      "learning_rate": 0.00010155,
      "loss": 2.9062,
      "step": 677
    },
    {
      "epoch": 0.44900662251655626,
      "grad_norm": 2.0935410302089568,
      "learning_rate": 0.00010169999999999999,
      "loss": 2.6406,
      "step": 678
    },
    {
      "epoch": 0.44966887417218543,
      "grad_norm": 1.8879317401494364,
      "learning_rate": 0.00010185,
      "loss": 2.8594,
      "step": 679
    },
    {
      "epoch": 0.4503311258278146,
      "grad_norm": 2.055970696021107,
      "learning_rate": 0.000102,
      "loss": 2.7188,
      "step": 680
    },
    {
      "epoch": 0.4509933774834437,
      "grad_norm": 2.185238399914902,
      "learning_rate": 0.00010215,
      "loss": 2.875,
      "step": 681
    },
    {
      "epoch": 0.45165562913907287,
      "grad_norm": 2.028180092120401,
      "learning_rate": 0.00010229999999999999,
      "loss": 2.5781,
      "step": 682
    },
    {
      "epoch": 0.452317880794702,
      "grad_norm": 1.9564392740966563,
      "learning_rate": 0.00010245,
      "loss": 2.7812,
      "step": 683
    },
    {
      "epoch": 0.45298013245033114,
      "grad_norm": 1.8469818602858226,
      "learning_rate": 0.0001026,
      "loss": 2.8125,
      "step": 684
    },
    {
      "epoch": 0.45364238410596025,
      "grad_norm": 1.7789251265103483,
      "learning_rate": 0.00010275,
      "loss": 2.7969,
      "step": 685
    },
    {
      "epoch": 0.4543046357615894,
      "grad_norm": 1.9462696541428774,
      "learning_rate": 0.0001029,
      "loss": 2.7969,
      "step": 686
    },
    {
      "epoch": 0.4549668874172185,
      "grad_norm": 1.9375468631358366,
      "learning_rate": 0.00010305,
      "loss": 2.2344,
      "step": 687
    },
    {
      "epoch": 0.4556291390728477,
      "grad_norm": 1.8878963337194594,
      "learning_rate": 0.00010319999999999999,
      "loss": 2.8281,
      "step": 688
    },
    {
      "epoch": 0.4562913907284768,
      "grad_norm": 1.856681928092609,
      "learning_rate": 0.00010334999999999998,
      "loss": 2.7656,
      "step": 689
    },
    {
      "epoch": 0.45695364238410596,
      "grad_norm": 1.921732450139847,
      "learning_rate": 0.00010349999999999998,
      "loss": 2.7969,
      "step": 690
    },
    {
      "epoch": 0.45761589403973507,
      "grad_norm": 1.8701944887818123,
      "learning_rate": 0.00010364999999999999,
      "loss": 2.7188,
      "step": 691
    },
    {
      "epoch": 0.45827814569536424,
      "grad_norm": 4.169335389273231,
      "learning_rate": 0.00010379999999999999,
      "loss": 2.9688,
      "step": 692
    },
    {
      "epoch": 0.4589403973509934,
      "grad_norm": 2.167018530169718,
      "learning_rate": 0.00010394999999999998,
      "loss": 2.9062,
      "step": 693
    },
    {
      "epoch": 0.4596026490066225,
      "grad_norm": 1.960256407129528,
      "learning_rate": 0.00010409999999999998,
      "loss": 2.75,
      "step": 694
    },
    {
      "epoch": 0.4602649006622517,
      "grad_norm": 2.1755736510939907,
      "learning_rate": 0.00010424999999999999,
      "loss": 2.7969,
      "step": 695
    },
    {
      "epoch": 0.4609271523178808,
      "grad_norm": 2.0297946354927285,
      "learning_rate": 0.00010439999999999999,
      "loss": 2.8594,
      "step": 696
    },
    {
      "epoch": 0.46158940397350995,
      "grad_norm": 2.0238840685003683,
      "learning_rate": 0.00010454999999999998,
      "loss": 2.9062,
      "step": 697
    },
    {
      "epoch": 0.46225165562913906,
      "grad_norm": 1.9731908343869506,
      "learning_rate": 0.00010469999999999998,
      "loss": 2.5,
      "step": 698
    },
    {
      "epoch": 0.4629139072847682,
      "grad_norm": 1.871549365036509,
      "learning_rate": 0.00010484999999999999,
      "loss": 2.8125,
      "step": 699
    },
    {
      "epoch": 0.46357615894039733,
      "grad_norm": 1.9245222022391868,
      "learning_rate": 0.00010499999999999999,
      "loss": 2.7656,
      "step": 700
    },
    {
      "epoch": 0.4642384105960265,
      "grad_norm": 1.9720270702151717,
      "learning_rate": 0.00010514999999999998,
      "loss": 2.7031,
      "step": 701
    },
    {
      "epoch": 0.4649006622516556,
      "grad_norm": 1.9171855121684747,
      "learning_rate": 0.00010529999999999998,
      "loss": 2.75,
      "step": 702
    },
    {
      "epoch": 0.46556291390728477,
      "grad_norm": 2.124760553049768,
      "learning_rate": 0.00010544999999999999,
      "loss": 2.7031,
      "step": 703
    },
    {
      "epoch": 0.46622516556291393,
      "grad_norm": 2.005358457356863,
      "learning_rate": 0.00010559999999999998,
      "loss": 2.8594,
      "step": 704
    },
    {
      "epoch": 0.46688741721854304,
      "grad_norm": 2.0586727803774836,
      "learning_rate": 0.00010574999999999998,
      "loss": 2.6719,
      "step": 705
    },
    {
      "epoch": 0.4675496688741722,
      "grad_norm": 2.3268192813796182,
      "learning_rate": 0.00010589999999999999,
      "loss": 2.7812,
      "step": 706
    },
    {
      "epoch": 0.4682119205298013,
      "grad_norm": 1.7245416728998577,
      "learning_rate": 0.00010604999999999999,
      "loss": 2.6094,
      "step": 707
    },
    {
      "epoch": 0.4688741721854305,
      "grad_norm": 1.8464381653716269,
      "learning_rate": 0.00010619999999999998,
      "loss": 2.5625,
      "step": 708
    },
    {
      "epoch": 0.4695364238410596,
      "grad_norm": 1.8839914065997356,
      "learning_rate": 0.00010634999999999998,
      "loss": 2.6719,
      "step": 709
    },
    {
      "epoch": 0.47019867549668876,
      "grad_norm": 2.0549100826173086,
      "learning_rate": 0.00010649999999999999,
      "loss": 2.7188,
      "step": 710
    },
    {
      "epoch": 0.47086092715231787,
      "grad_norm": 2.1669033195048932,
      "learning_rate": 0.00010664999999999999,
      "loss": 2.9062,
      "step": 711
    },
    {
      "epoch": 0.47152317880794703,
      "grad_norm": 4.242307772503889,
      "learning_rate": 0.00010679999999999998,
      "loss": 2.3438,
      "step": 712
    },
    {
      "epoch": 0.47218543046357614,
      "grad_norm": 2.230049474588384,
      "learning_rate": 0.00010694999999999998,
      "loss": 2.5,
      "step": 713
    },
    {
      "epoch": 0.4728476821192053,
      "grad_norm": 2.5508420933245186,
      "learning_rate": 0.00010709999999999999,
      "loss": 2.9375,
      "step": 714
    },
    {
      "epoch": 0.4735099337748344,
      "grad_norm": 1.9773917686858677,
      "learning_rate": 0.00010724999999999999,
      "loss": 2.625,
      "step": 715
    },
    {
      "epoch": 0.4741721854304636,
      "grad_norm": 2.3073193812785617,
      "learning_rate": 0.00010739999999999998,
      "loss": 2.5781,
      "step": 716
    },
    {
      "epoch": 0.47483443708609274,
      "grad_norm": 2.1683524322665675,
      "learning_rate": 0.00010754999999999999,
      "loss": 3.0,
      "step": 717
    },
    {
      "epoch": 0.47549668874172185,
      "grad_norm": 1.7891675793673438,
      "learning_rate": 0.00010769999999999999,
      "loss": 2.6875,
      "step": 718
    },
    {
      "epoch": 0.476158940397351,
      "grad_norm": 1.963404117289492,
      "learning_rate": 0.00010784999999999999,
      "loss": 2.6094,
      "step": 719
    },
    {
      "epoch": 0.4768211920529801,
      "grad_norm": 2.102721907639093,
      "learning_rate": 0.00010799999999999998,
      "loss": 2.7969,
      "step": 720
    },
    {
      "epoch": 0.4774834437086093,
      "grad_norm": 2.1921804967485627,
      "learning_rate": 0.00010814999999999999,
      "loss": 2.6875,
      "step": 721
    },
    {
      "epoch": 0.4781456953642384,
      "grad_norm": 2.1426187353732526,
      "learning_rate": 0.00010829999999999999,
      "loss": 2.7188,
      "step": 722
    },
    {
      "epoch": 0.47880794701986756,
      "grad_norm": 2.0710641618379264,
      "learning_rate": 0.00010845,
      "loss": 2.8125,
      "step": 723
    },
    {
      "epoch": 0.4794701986754967,
      "grad_norm": 2.0973632106241973,
      "learning_rate": 0.00010859999999999998,
      "loss": 2.8594,
      "step": 724
    },
    {
      "epoch": 0.48013245033112584,
      "grad_norm": 2.0027135007023915,
      "learning_rate": 0.00010874999999999999,
      "loss": 2.5625,
      "step": 725
    },
    {
      "epoch": 0.48079470198675495,
      "grad_norm": 1.7345598374122004,
      "learning_rate": 0.00010889999999999999,
      "loss": 2.5312,
      "step": 726
    },
    {
      "epoch": 0.4814569536423841,
      "grad_norm": 1.8985617274201236,
      "learning_rate": 0.00010904999999999998,
      "loss": 2.6094,
      "step": 727
    },
    {
      "epoch": 0.4821192052980132,
      "grad_norm": 1.8312647294044948,
      "learning_rate": 0.00010919999999999998,
      "loss": 2.6406,
      "step": 728
    },
    {
      "epoch": 0.4827814569536424,
      "grad_norm": 1.9805082205584479,
      "learning_rate": 0.00010934999999999999,
      "loss": 2.5781,
      "step": 729
    },
    {
      "epoch": 0.48344370860927155,
      "grad_norm": 2.071072549754882,
      "learning_rate": 0.00010949999999999999,
      "loss": 2.6719,
      "step": 730
    },
    {
      "epoch": 0.48410596026490066,
      "grad_norm": 1.9270290991948522,
      "learning_rate": 0.00010964999999999998,
      "loss": 2.75,
      "step": 731
    },
    {
      "epoch": 0.4847682119205298,
      "grad_norm": 1.956164832463769,
      "learning_rate": 0.00010979999999999999,
      "loss": 2.625,
      "step": 732
    },
    {
      "epoch": 0.48543046357615893,
      "grad_norm": 2.219261601014454,
      "learning_rate": 0.00010994999999999999,
      "loss": 3.0312,
      "step": 733
    },
    {
      "epoch": 0.4860927152317881,
      "grad_norm": 2.048459710802969,
      "learning_rate": 0.00011009999999999999,
      "loss": 2.8125,
      "step": 734
    },
    {
      "epoch": 0.4867549668874172,
      "grad_norm": 2.0887789379854698,
      "learning_rate": 0.00011024999999999998,
      "loss": 3.0625,
      "step": 735
    },
    {
      "epoch": 0.48741721854304637,
      "grad_norm": 1.821516856922936,
      "learning_rate": 0.00011039999999999999,
      "loss": 2.6406,
      "step": 736
    },
    {
      "epoch": 0.4880794701986755,
      "grad_norm": 1.8905215179124095,
      "learning_rate": 0.00011054999999999999,
      "loss": 2.9688,
      "step": 737
    },
    {
      "epoch": 0.48874172185430464,
      "grad_norm": 2.0583695629368086,
      "learning_rate": 0.0001107,
      "loss": 2.8594,
      "step": 738
    },
    {
      "epoch": 0.48940397350993375,
      "grad_norm": 1.9024595576268803,
      "learning_rate": 0.00011084999999999998,
      "loss": 2.6562,
      "step": 739
    },
    {
      "epoch": 0.4900662251655629,
      "grad_norm": 2.0054043650657287,
      "learning_rate": 0.00011099999999999999,
      "loss": 2.9375,
      "step": 740
    },
    {
      "epoch": 0.49072847682119203,
      "grad_norm": 2.0531208576121274,
      "learning_rate": 0.00011114999999999999,
      "loss": 2.5625,
      "step": 741
    },
    {
      "epoch": 0.4913907284768212,
      "grad_norm": 2.066240447010065,
      "learning_rate": 0.0001113,
      "loss": 2.7969,
      "step": 742
    },
    {
      "epoch": 0.49205298013245036,
      "grad_norm": 1.8133406532223322,
      "learning_rate": 0.00011144999999999998,
      "loss": 2.7969,
      "step": 743
    },
    {
      "epoch": 0.49271523178807947,
      "grad_norm": 2.0337928729066523,
      "learning_rate": 0.00011159999999999999,
      "loss": 2.625,
      "step": 744
    },
    {
      "epoch": 0.49337748344370863,
      "grad_norm": 1.92101661813358,
      "learning_rate": 0.00011174999999999999,
      "loss": 2.8594,
      "step": 745
    },
    {
      "epoch": 0.49403973509933774,
      "grad_norm": 1.7178335519352124,
      "learning_rate": 0.0001119,
      "loss": 2.75,
      "step": 746
    },
    {
      "epoch": 0.4947019867549669,
      "grad_norm": 1.9742838470927657,
      "learning_rate": 0.00011204999999999999,
      "loss": 3.0156,
      "step": 747
    },
    {
      "epoch": 0.495364238410596,
      "grad_norm": 2.0930678039048245,
      "learning_rate": 0.00011219999999999999,
      "loss": 2.9219,
      "step": 748
    },
    {
      "epoch": 0.4960264900662252,
      "grad_norm": 2.1000336000050717,
      "learning_rate": 0.00011235,
      "loss": 2.4531,
      "step": 749
    },
    {
      "epoch": 0.4966887417218543,
      "grad_norm": 1.689868266249042,
      "learning_rate": 0.0001125,
      "loss": 2.4688,
      "step": 750
    },
    {
      "epoch": 0.49735099337748345,
      "grad_norm": 2.3162031291766674,
      "learning_rate": 0.00011264999999999999,
      "loss": 3.1094,
      "step": 751
    },
    {
      "epoch": 0.49801324503311256,
      "grad_norm": 2.0474483134251655,
      "learning_rate": 0.00011279999999999999,
      "loss": 2.8438,
      "step": 752
    },
    {
      "epoch": 0.4986754966887417,
      "grad_norm": 1.6925971215396884,
      "learning_rate": 0.00011295,
      "loss": 2.8281,
      "step": 753
    },
    {
      "epoch": 0.49933774834437084,
      "grad_norm": 1.8550156080538551,
      "learning_rate": 0.00011309999999999998,
      "loss": 2.625,
      "step": 754
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8886038363874273,
      "learning_rate": 0.00011324999999999999,
      "loss": 2.6719,
      "step": 755
    },
    {
      "epoch": 0.5006622516556292,
      "grad_norm": 1.7207522598131173,
      "learning_rate": 0.00011339999999999999,
      "loss": 2.7812,
      "step": 756
    },
    {
      "epoch": 0.5013245033112583,
      "grad_norm": 1.7412690874170982,
      "learning_rate": 0.00011355,
      "loss": 2.6406,
      "step": 757
    },
    {
      "epoch": 0.5019867549668874,
      "grad_norm": 1.830895304100342,
      "learning_rate": 0.00011369999999999999,
      "loss": 2.6875,
      "step": 758
    },
    {
      "epoch": 0.5026490066225165,
      "grad_norm": 1.861511388202317,
      "learning_rate": 0.00011384999999999999,
      "loss": 3.0938,
      "step": 759
    },
    {
      "epoch": 0.5033112582781457,
      "grad_norm": 1.86325234077852,
      "learning_rate": 0.00011399999999999999,
      "loss": 2.2812,
      "step": 760
    },
    {
      "epoch": 0.5039735099337749,
      "grad_norm": 1.8623911294127775,
      "learning_rate": 0.00011415,
      "loss": 2.3906,
      "step": 761
    },
    {
      "epoch": 0.5046357615894039,
      "grad_norm": 2.1326876062244486,
      "learning_rate": 0.00011429999999999999,
      "loss": 3.0312,
      "step": 762
    },
    {
      "epoch": 0.5052980132450331,
      "grad_norm": 1.720494118454427,
      "learning_rate": 0.00011444999999999999,
      "loss": 2.5312,
      "step": 763
    },
    {
      "epoch": 0.5059602649006623,
      "grad_norm": 1.9522265564763535,
      "learning_rate": 0.0001146,
      "loss": 2.7031,
      "step": 764
    },
    {
      "epoch": 0.5066225165562914,
      "grad_norm": 2.2010704199527096,
      "learning_rate": 0.00011475,
      "loss": 2.5781,
      "step": 765
    },
    {
      "epoch": 0.5072847682119205,
      "grad_norm": 1.7752474504916527,
      "learning_rate": 0.00011489999999999999,
      "loss": 2.75,
      "step": 766
    },
    {
      "epoch": 0.5079470198675496,
      "grad_norm": 1.9573557173601654,
      "learning_rate": 0.00011504999999999999,
      "loss": 2.3906,
      "step": 767
    },
    {
      "epoch": 0.5086092715231788,
      "grad_norm": 2.32855193882003,
      "learning_rate": 0.0001152,
      "loss": 3.0469,
      "step": 768
    },
    {
      "epoch": 0.509271523178808,
      "grad_norm": 1.922631377589676,
      "learning_rate": 0.00011535,
      "loss": 2.7969,
      "step": 769
    },
    {
      "epoch": 0.5099337748344371,
      "grad_norm": 1.9175037190709405,
      "learning_rate": 0.00011549999999999999,
      "loss": 2.3125,
      "step": 770
    },
    {
      "epoch": 0.5105960264900662,
      "grad_norm": 2.6069228766491324,
      "learning_rate": 0.00011564999999999999,
      "loss": 2.7344,
      "step": 771
    },
    {
      "epoch": 0.5112582781456954,
      "grad_norm": 2.20275784898294,
      "learning_rate": 0.0001158,
      "loss": 2.7812,
      "step": 772
    },
    {
      "epoch": 0.5119205298013245,
      "grad_norm": 2.0780138867352433,
      "learning_rate": 0.00011595,
      "loss": 2.6719,
      "step": 773
    },
    {
      "epoch": 0.5125827814569537,
      "grad_norm": 1.9779476186374148,
      "learning_rate": 0.00011609999999999999,
      "loss": 2.2969,
      "step": 774
    },
    {
      "epoch": 0.5132450331125827,
      "grad_norm": 2.0382052240868065,
      "learning_rate": 0.00011624999999999999,
      "loss": 2.8594,
      "step": 775
    },
    {
      "epoch": 0.5139072847682119,
      "grad_norm": 1.6902369875968617,
      "learning_rate": 0.0001164,
      "loss": 2.625,
      "step": 776
    },
    {
      "epoch": 0.5145695364238411,
      "grad_norm": 1.9211441968645673,
      "learning_rate": 0.00011654999999999999,
      "loss": 2.7656,
      "step": 777
    },
    {
      "epoch": 0.5152317880794702,
      "grad_norm": 1.8897116323932281,
      "learning_rate": 0.00011669999999999999,
      "loss": 2.75,
      "step": 778
    },
    {
      "epoch": 0.5158940397350993,
      "grad_norm": 1.7821653101284112,
      "learning_rate": 0.00011685,
      "loss": 2.6719,
      "step": 779
    },
    {
      "epoch": 0.5165562913907285,
      "grad_norm": 1.8477325585046043,
      "learning_rate": 0.000117,
      "loss": 2.5,
      "step": 780
    },
    {
      "epoch": 0.5172185430463576,
      "grad_norm": 2.0495959384389857,
      "learning_rate": 0.00011714999999999999,
      "loss": 2.5469,
      "step": 781
    },
    {
      "epoch": 0.5178807947019868,
      "grad_norm": 1.7864365744361108,
      "learning_rate": 0.00011729999999999999,
      "loss": 2.6875,
      "step": 782
    },
    {
      "epoch": 0.5185430463576159,
      "grad_norm": 1.8740829841389737,
      "learning_rate": 0.00011745,
      "loss": 2.7031,
      "step": 783
    },
    {
      "epoch": 0.519205298013245,
      "grad_norm": 1.859576335191902,
      "learning_rate": 0.0001176,
      "loss": 2.7969,
      "step": 784
    },
    {
      "epoch": 0.5198675496688742,
      "grad_norm": 1.7466716734440257,
      "learning_rate": 0.00011774999999999999,
      "loss": 2.0938,
      "step": 785
    },
    {
      "epoch": 0.5205298013245033,
      "grad_norm": 2.170867807668343,
      "learning_rate": 0.00011789999999999999,
      "loss": 2.5469,
      "step": 786
    },
    {
      "epoch": 0.5211920529801325,
      "grad_norm": 13.077275299264398,
      "learning_rate": 0.00011805,
      "loss": 2.3438,
      "step": 787
    },
    {
      "epoch": 0.5218543046357615,
      "grad_norm": 2.0245919838606157,
      "learning_rate": 0.0001182,
      "loss": 2.7656,
      "step": 788
    },
    {
      "epoch": 0.5225165562913907,
      "grad_norm": 1.7673280376866838,
      "learning_rate": 0.00011834999999999999,
      "loss": 2.9219,
      "step": 789
    },
    {
      "epoch": 0.5231788079470199,
      "grad_norm": 2.1047909435557806,
      "learning_rate": 0.0001185,
      "loss": 2.7656,
      "step": 790
    },
    {
      "epoch": 0.523841059602649,
      "grad_norm": 1.878513072489257,
      "learning_rate": 0.00011865,
      "loss": 2.7812,
      "step": 791
    },
    {
      "epoch": 0.5245033112582781,
      "grad_norm": 1.8641099710480125,
      "learning_rate": 0.0001188,
      "loss": 2.7812,
      "step": 792
    },
    {
      "epoch": 0.5251655629139073,
      "grad_norm": 2.06959887252658,
      "learning_rate": 0.00011894999999999999,
      "loss": 2.8594,
      "step": 793
    },
    {
      "epoch": 0.5258278145695364,
      "grad_norm": 1.7496189165860319,
      "learning_rate": 0.0001191,
      "loss": 2.6875,
      "step": 794
    },
    {
      "epoch": 0.5264900662251656,
      "grad_norm": 1.9912100526135885,
      "learning_rate": 0.00011925,
      "loss": 2.875,
      "step": 795
    },
    {
      "epoch": 0.5271523178807948,
      "grad_norm": 1.8402023511717387,
      "learning_rate": 0.0001194,
      "loss": 2.4219,
      "step": 796
    },
    {
      "epoch": 0.5278145695364238,
      "grad_norm": 1.9057031816263097,
      "learning_rate": 0.00011954999999999999,
      "loss": 2.3594,
      "step": 797
    },
    {
      "epoch": 0.528476821192053,
      "grad_norm": 1.7956758706384683,
      "learning_rate": 0.0001197,
      "loss": 2.75,
      "step": 798
    },
    {
      "epoch": 0.5291390728476821,
      "grad_norm": 1.8087190772255215,
      "learning_rate": 0.00011985,
      "loss": 2.2031,
      "step": 799
    },
    {
      "epoch": 0.5298013245033113,
      "grad_norm": 1.6472549287780949,
      "learning_rate": 0.00011999999999999999,
      "loss": 2.5938,
      "step": 800
    },
    {
      "epoch": 0.5304635761589404,
      "grad_norm": 6.530946947138037,
      "learning_rate": 0.00012014999999999999,
      "loss": 3.0469,
      "step": 801
    },
    {
      "epoch": 0.5311258278145695,
      "grad_norm": 1.9019258907496284,
      "learning_rate": 0.0001203,
      "loss": 2.8125,
      "step": 802
    },
    {
      "epoch": 0.5317880794701987,
      "grad_norm": 1.7844438310074837,
      "learning_rate": 0.00012045,
      "loss": 2.9062,
      "step": 803
    },
    {
      "epoch": 0.5324503311258278,
      "grad_norm": 1.7783963177802107,
      "learning_rate": 0.00012059999999999999,
      "loss": 2.875,
      "step": 804
    },
    {
      "epoch": 0.5331125827814569,
      "grad_norm": 1.937865201568551,
      "learning_rate": 0.00012075,
      "loss": 2.8125,
      "step": 805
    },
    {
      "epoch": 0.5337748344370861,
      "grad_norm": 1.6858975743511662,
      "learning_rate": 0.0001209,
      "loss": 2.6094,
      "step": 806
    },
    {
      "epoch": 0.5344370860927152,
      "grad_norm": 1.9154947485714664,
      "learning_rate": 0.00012105,
      "loss": 3.0469,
      "step": 807
    },
    {
      "epoch": 0.5350993377483444,
      "grad_norm": 2.150974937389423,
      "learning_rate": 0.00012119999999999999,
      "loss": 2.5938,
      "step": 808
    },
    {
      "epoch": 0.5357615894039736,
      "grad_norm": 1.7799830052191583,
      "learning_rate": 0.00012135,
      "loss": 2.625,
      "step": 809
    },
    {
      "epoch": 0.5364238410596026,
      "grad_norm": 2.081073355218276,
      "learning_rate": 0.0001215,
      "loss": 2.7812,
      "step": 810
    },
    {
      "epoch": 0.5370860927152318,
      "grad_norm": 1.8018274757001718,
      "learning_rate": 0.00012165,
      "loss": 2.8281,
      "step": 811
    },
    {
      "epoch": 0.5377483443708609,
      "grad_norm": 1.8423803656335467,
      "learning_rate": 0.00012179999999999999,
      "loss": 2.6406,
      "step": 812
    },
    {
      "epoch": 0.5384105960264901,
      "grad_norm": 2.026709485612334,
      "learning_rate": 0.00012194999999999998,
      "loss": 3.0625,
      "step": 813
    },
    {
      "epoch": 0.5390728476821192,
      "grad_norm": 1.8936927142722246,
      "learning_rate": 0.00012209999999999999,
      "loss": 2.7188,
      "step": 814
    },
    {
      "epoch": 0.5397350993377483,
      "grad_norm": 1.8765357563989975,
      "learning_rate": 0.00012225,
      "loss": 2.2344,
      "step": 815
    },
    {
      "epoch": 0.5403973509933775,
      "grad_norm": 1.9195032529778533,
      "learning_rate": 0.0001224,
      "loss": 2.7344,
      "step": 816
    },
    {
      "epoch": 0.5410596026490067,
      "grad_norm": 1.8918723376543463,
      "learning_rate": 0.00012254999999999997,
      "loss": 2.7344,
      "step": 817
    },
    {
      "epoch": 0.5417218543046357,
      "grad_norm": 1.8326477133298356,
      "learning_rate": 0.00012269999999999997,
      "loss": 2.6094,
      "step": 818
    },
    {
      "epoch": 0.5423841059602649,
      "grad_norm": 1.7975830196144638,
      "learning_rate": 0.00012284999999999998,
      "loss": 2.6406,
      "step": 819
    },
    {
      "epoch": 0.543046357615894,
      "grad_norm": 1.8278078032739093,
      "learning_rate": 0.00012299999999999998,
      "loss": 2.625,
      "step": 820
    },
    {
      "epoch": 0.5437086092715232,
      "grad_norm": 1.8222155606884072,
      "learning_rate": 0.00012314999999999998,
      "loss": 2.6875,
      "step": 821
    },
    {
      "epoch": 0.5443708609271524,
      "grad_norm": 1.7723078578833016,
      "learning_rate": 0.0001233,
      "loss": 2.7812,
      "step": 822
    },
    {
      "epoch": 0.5450331125827814,
      "grad_norm": 1.7667910066878394,
      "learning_rate": 0.00012345,
      "loss": 2.4375,
      "step": 823
    },
    {
      "epoch": 0.5456953642384106,
      "grad_norm": 1.7770662078586374,
      "learning_rate": 0.0001236,
      "loss": 2.8281,
      "step": 824
    },
    {
      "epoch": 0.5463576158940397,
      "grad_norm": 2.0470842973375807,
      "learning_rate": 0.00012374999999999997,
      "loss": 2.7969,
      "step": 825
    },
    {
      "epoch": 0.5470198675496689,
      "grad_norm": 2.032816439716145,
      "learning_rate": 0.00012389999999999998,
      "loss": 2.6562,
      "step": 826
    },
    {
      "epoch": 0.547682119205298,
      "grad_norm": 1.9518305930955175,
      "learning_rate": 0.00012404999999999998,
      "loss": 2.8594,
      "step": 827
    },
    {
      "epoch": 0.5483443708609271,
      "grad_norm": 2.0702529032951333,
      "learning_rate": 0.00012419999999999998,
      "loss": 2.7812,
      "step": 828
    },
    {
      "epoch": 0.5490066225165563,
      "grad_norm": 1.7810457757515485,
      "learning_rate": 0.00012435,
      "loss": 2.7812,
      "step": 829
    },
    {
      "epoch": 0.5496688741721855,
      "grad_norm": 1.874339582643248,
      "learning_rate": 0.0001245,
      "loss": 2.9062,
      "step": 830
    },
    {
      "epoch": 0.5503311258278145,
      "grad_norm": 1.9115618687119742,
      "learning_rate": 0.00012465,
      "loss": 2.75,
      "step": 831
    },
    {
      "epoch": 0.5509933774834437,
      "grad_norm": 1.6226128753757272,
      "learning_rate": 0.00012479999999999997,
      "loss": 2.625,
      "step": 832
    },
    {
      "epoch": 0.5516556291390728,
      "grad_norm": 1.8403839566351927,
      "learning_rate": 0.00012494999999999997,
      "loss": 2.8281,
      "step": 833
    },
    {
      "epoch": 0.552317880794702,
      "grad_norm": 1.7786123114107988,
      "learning_rate": 0.00012509999999999998,
      "loss": 2.8438,
      "step": 834
    },
    {
      "epoch": 0.5529801324503312,
      "grad_norm": 1.9010352888493076,
      "learning_rate": 0.00012524999999999998,
      "loss": 2.1719,
      "step": 835
    },
    {
      "epoch": 0.5536423841059602,
      "grad_norm": 1.8555955810398956,
      "learning_rate": 0.00012539999999999999,
      "loss": 2.75,
      "step": 836
    },
    {
      "epoch": 0.5543046357615894,
      "grad_norm": 1.7684581460775344,
      "learning_rate": 0.00012555,
      "loss": 2.7969,
      "step": 837
    },
    {
      "epoch": 0.5549668874172186,
      "grad_norm": 1.8618762906510204,
      "learning_rate": 0.0001257,
      "loss": 2.2188,
      "step": 838
    },
    {
      "epoch": 0.5556291390728477,
      "grad_norm": 1.9331369675858678,
      "learning_rate": 0.00012585,
      "loss": 2.6875,
      "step": 839
    },
    {
      "epoch": 0.5562913907284768,
      "grad_norm": 1.82406671427144,
      "learning_rate": 0.00012599999999999997,
      "loss": 2.6875,
      "step": 840
    },
    {
      "epoch": 0.5569536423841059,
      "grad_norm": 1.772148670243457,
      "learning_rate": 0.00012614999999999998,
      "loss": 2.7031,
      "step": 841
    },
    {
      "epoch": 0.5576158940397351,
      "grad_norm": 1.7488091792948688,
      "learning_rate": 0.00012629999999999998,
      "loss": 2.8594,
      "step": 842
    },
    {
      "epoch": 0.5582781456953643,
      "grad_norm": 1.8065000955581838,
      "learning_rate": 0.00012644999999999998,
      "loss": 2.2188,
      "step": 843
    },
    {
      "epoch": 0.5589403973509933,
      "grad_norm": 1.7098073194862848,
      "learning_rate": 0.0001266,
      "loss": 2.6875,
      "step": 844
    },
    {
      "epoch": 0.5596026490066225,
      "grad_norm": 1.9531921753484562,
      "learning_rate": 0.00012675,
      "loss": 3.0,
      "step": 845
    },
    {
      "epoch": 0.5602649006622517,
      "grad_norm": 1.963000496497309,
      "learning_rate": 0.0001269,
      "loss": 2.9844,
      "step": 846
    },
    {
      "epoch": 0.5609271523178808,
      "grad_norm": 1.710875948917655,
      "learning_rate": 0.00012705,
      "loss": 2.7344,
      "step": 847
    },
    {
      "epoch": 0.56158940397351,
      "grad_norm": 2.4151996667056617,
      "learning_rate": 0.00012719999999999997,
      "loss": 2.7344,
      "step": 848
    },
    {
      "epoch": 0.562251655629139,
      "grad_norm": 1.8125472612640734,
      "learning_rate": 0.00012734999999999998,
      "loss": 2.8281,
      "step": 849
    },
    {
      "epoch": 0.5629139072847682,
      "grad_norm": 1.8140943095423947,
      "learning_rate": 0.00012749999999999998,
      "loss": 2.8594,
      "step": 850
    },
    {
      "epoch": 0.5635761589403974,
      "grad_norm": 1.7342284250214337,
      "learning_rate": 0.00012764999999999999,
      "loss": 2.2344,
      "step": 851
    },
    {
      "epoch": 0.5642384105960265,
      "grad_norm": 1.942351624840455,
      "learning_rate": 0.0001278,
      "loss": 3.0312,
      "step": 852
    },
    {
      "epoch": 0.5649006622516556,
      "grad_norm": 1.7930118269153643,
      "learning_rate": 0.00012795,
      "loss": 2.8906,
      "step": 853
    },
    {
      "epoch": 0.5655629139072847,
      "grad_norm": 1.9409660419617345,
      "learning_rate": 0.0001281,
      "loss": 2.9219,
      "step": 854
    },
    {
      "epoch": 0.5662251655629139,
      "grad_norm": 1.7411916959780829,
      "learning_rate": 0.00012824999999999997,
      "loss": 2.5938,
      "step": 855
    },
    {
      "epoch": 0.5668874172185431,
      "grad_norm": 1.7628585189329957,
      "learning_rate": 0.00012839999999999998,
      "loss": 2.8594,
      "step": 856
    },
    {
      "epoch": 0.5675496688741721,
      "grad_norm": 1.9807303407821348,
      "learning_rate": 0.00012854999999999998,
      "loss": 2.75,
      "step": 857
    },
    {
      "epoch": 0.5682119205298013,
      "grad_norm": 1.8718308918108135,
      "learning_rate": 0.00012869999999999998,
      "loss": 2.4062,
      "step": 858
    },
    {
      "epoch": 0.5688741721854305,
      "grad_norm": 1.9789335376073256,
      "learning_rate": 0.00012885,
      "loss": 2.7188,
      "step": 859
    },
    {
      "epoch": 0.5695364238410596,
      "grad_norm": 1.7599091552080293,
      "learning_rate": 0.000129,
      "loss": 2.7031,
      "step": 860
    },
    {
      "epoch": 0.5701986754966888,
      "grad_norm": 1.8880914188513047,
      "learning_rate": 0.00012915,
      "loss": 2.8594,
      "step": 861
    },
    {
      "epoch": 0.5708609271523178,
      "grad_norm": 1.7033803644464172,
      "learning_rate": 0.0001293,
      "loss": 2.8125,
      "step": 862
    },
    {
      "epoch": 0.571523178807947,
      "grad_norm": 1.6279207101353814,
      "learning_rate": 0.00012944999999999998,
      "loss": 2.2969,
      "step": 863
    },
    {
      "epoch": 0.5721854304635762,
      "grad_norm": 1.7806742967800837,
      "learning_rate": 0.00012959999999999998,
      "loss": 2.625,
      "step": 864
    },
    {
      "epoch": 0.5728476821192053,
      "grad_norm": 1.6403256333752165,
      "learning_rate": 0.00012974999999999998,
      "loss": 2.6562,
      "step": 865
    },
    {
      "epoch": 0.5735099337748344,
      "grad_norm": 1.7393439532601982,
      "learning_rate": 0.00012989999999999999,
      "loss": 2.9688,
      "step": 866
    },
    {
      "epoch": 0.5741721854304636,
      "grad_norm": 1.84486249966741,
      "learning_rate": 0.00013005,
      "loss": 2.9375,
      "step": 867
    },
    {
      "epoch": 0.5748344370860927,
      "grad_norm": 1.6615087785848357,
      "learning_rate": 0.0001302,
      "loss": 2.2656,
      "step": 868
    },
    {
      "epoch": 0.5754966887417219,
      "grad_norm": 1.7348453249349052,
      "learning_rate": 0.00013035,
      "loss": 2.75,
      "step": 869
    },
    {
      "epoch": 0.5761589403973509,
      "grad_norm": 1.8505684990104836,
      "learning_rate": 0.0001305,
      "loss": 2.6719,
      "step": 870
    },
    {
      "epoch": 0.5768211920529801,
      "grad_norm": 1.7487787552640692,
      "learning_rate": 0.00013064999999999998,
      "loss": 2.9219,
      "step": 871
    },
    {
      "epoch": 0.5774834437086093,
      "grad_norm": 1.6390476132277572,
      "learning_rate": 0.00013079999999999998,
      "loss": 2.4531,
      "step": 872
    },
    {
      "epoch": 0.5781456953642384,
      "grad_norm": 6.920371927179222,
      "learning_rate": 0.00013094999999999998,
      "loss": 2.875,
      "step": 873
    },
    {
      "epoch": 0.5788079470198676,
      "grad_norm": 1.6653294520602393,
      "learning_rate": 0.0001311,
      "loss": 2.7656,
      "step": 874
    },
    {
      "epoch": 0.5794701986754967,
      "grad_norm": 1.9043293860322916,
      "learning_rate": 0.00013125,
      "loss": 2.625,
      "step": 875
    },
    {
      "epoch": 0.5801324503311258,
      "grad_norm": 1.7400297709953005,
      "learning_rate": 0.0001314,
      "loss": 2.7812,
      "step": 876
    },
    {
      "epoch": 0.580794701986755,
      "grad_norm": 2.127468788413566,
      "learning_rate": 0.00013155,
      "loss": 2.3125,
      "step": 877
    },
    {
      "epoch": 0.5814569536423841,
      "grad_norm": 3.3141446300449053,
      "learning_rate": 0.00013169999999999998,
      "loss": 2.75,
      "step": 878
    },
    {
      "epoch": 0.5821192052980132,
      "grad_norm": 1.6401374573515344,
      "learning_rate": 0.00013184999999999998,
      "loss": 2.4844,
      "step": 879
    },
    {
      "epoch": 0.5827814569536424,
      "grad_norm": 2.092500690803508,
      "learning_rate": 0.00013199999999999998,
      "loss": 2.4844,
      "step": 880
    },
    {
      "epoch": 0.5834437086092715,
      "grad_norm": 1.7528797960299052,
      "learning_rate": 0.00013215,
      "loss": 2.75,
      "step": 881
    },
    {
      "epoch": 0.5841059602649007,
      "grad_norm": 1.63393423635461,
      "learning_rate": 0.0001323,
      "loss": 2.6094,
      "step": 882
    },
    {
      "epoch": 0.5847682119205299,
      "grad_norm": 1.9253028232849625,
      "learning_rate": 0.00013245,
      "loss": 2.75,
      "step": 883
    },
    {
      "epoch": 0.5854304635761589,
      "grad_norm": 1.715249792015313,
      "learning_rate": 0.0001326,
      "loss": 2.7812,
      "step": 884
    },
    {
      "epoch": 0.5860927152317881,
      "grad_norm": 1.6981377614464783,
      "learning_rate": 0.00013275,
      "loss": 2.6562,
      "step": 885
    },
    {
      "epoch": 0.5867549668874172,
      "grad_norm": 1.8802317299290754,
      "learning_rate": 0.00013289999999999998,
      "loss": 2.9531,
      "step": 886
    },
    {
      "epoch": 0.5874172185430464,
      "grad_norm": 1.739780615891697,
      "learning_rate": 0.00013304999999999998,
      "loss": 2.2812,
      "step": 887
    },
    {
      "epoch": 0.5880794701986755,
      "grad_norm": 1.7960771387390673,
      "learning_rate": 0.00013319999999999999,
      "loss": 2.7969,
      "step": 888
    },
    {
      "epoch": 0.5887417218543046,
      "grad_norm": 1.5937602429617455,
      "learning_rate": 0.00013335,
      "loss": 2.3906,
      "step": 889
    },
    {
      "epoch": 0.5894039735099338,
      "grad_norm": 1.792822056451431,
      "learning_rate": 0.0001335,
      "loss": 2.6719,
      "step": 890
    },
    {
      "epoch": 0.590066225165563,
      "grad_norm": 1.6338513392343388,
      "learning_rate": 0.00013365,
      "loss": 2.5938,
      "step": 891
    },
    {
      "epoch": 0.590728476821192,
      "grad_norm": 11.809064105651226,
      "learning_rate": 0.0001338,
      "loss": 2.75,
      "step": 892
    },
    {
      "epoch": 0.5913907284768212,
      "grad_norm": 2.052321374588075,
      "learning_rate": 0.00013395,
      "loss": 2.7031,
      "step": 893
    },
    {
      "epoch": 0.5920529801324503,
      "grad_norm": 1.6701304241121868,
      "learning_rate": 0.00013409999999999998,
      "loss": 2.2031,
      "step": 894
    },
    {
      "epoch": 0.5927152317880795,
      "grad_norm": 1.9535092787535708,
      "learning_rate": 0.00013424999999999998,
      "loss": 2.625,
      "step": 895
    },
    {
      "epoch": 0.5933774834437087,
      "grad_norm": 2.0016210945510418,
      "learning_rate": 0.0001344,
      "loss": 2.625,
      "step": 896
    },
    {
      "epoch": 0.5940397350993377,
      "grad_norm": 1.8708541582692386,
      "learning_rate": 0.00013455,
      "loss": 2.625,
      "step": 897
    },
    {
      "epoch": 0.5947019867549669,
      "grad_norm": 13.725045749176015,
      "learning_rate": 0.0001347,
      "loss": 2.7031,
      "step": 898
    },
    {
      "epoch": 0.595364238410596,
      "grad_norm": 1.8149991284796483,
      "learning_rate": 0.00013485,
      "loss": 2.6094,
      "step": 899
    },
    {
      "epoch": 0.5960264900662252,
      "grad_norm": 1.866814167931646,
      "learning_rate": 0.000135,
      "loss": 2.75,
      "step": 900
    },
    {
      "epoch": 0.5966887417218543,
      "grad_norm": 1.5959615345966283,
      "learning_rate": 0.00013514999999999998,
      "loss": 2.9062,
      "step": 901
    },
    {
      "epoch": 0.5973509933774834,
      "grad_norm": 1.808896172507155,
      "learning_rate": 0.00013529999999999998,
      "loss": 2.7188,
      "step": 902
    },
    {
      "epoch": 0.5980132450331126,
      "grad_norm": 2.043905475817425,
      "learning_rate": 0.00013544999999999999,
      "loss": 2.7188,
      "step": 903
    },
    {
      "epoch": 0.5986754966887418,
      "grad_norm": 1.6476851986779213,
      "learning_rate": 0.0001356,
      "loss": 2.9688,
      "step": 904
    },
    {
      "epoch": 0.5993377483443708,
      "grad_norm": 1.6072455464870483,
      "learning_rate": 0.00013575,
      "loss": 2.7344,
      "step": 905
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8522763141275878,
      "learning_rate": 0.0001359,
      "loss": 2.6875,
      "step": 906
    },
    {
      "epoch": 0.6006622516556291,
      "grad_norm": 1.695948375368472,
      "learning_rate": 0.00013605,
      "loss": 2.5625,
      "step": 907
    },
    {
      "epoch": 0.6013245033112583,
      "grad_norm": 1.6639403183630144,
      "learning_rate": 0.0001362,
      "loss": 2.6562,
      "step": 908
    },
    {
      "epoch": 0.6019867549668875,
      "grad_norm": 1.8018234664253343,
      "learning_rate": 0.00013634999999999998,
      "loss": 2.7188,
      "step": 909
    },
    {
      "epoch": 0.6026490066225165,
      "grad_norm": 1.677039196367644,
      "learning_rate": 0.00013649999999999998,
      "loss": 2.4688,
      "step": 910
    },
    {
      "epoch": 0.6033112582781457,
      "grad_norm": 1.5578843517473313,
      "learning_rate": 0.00013665,
      "loss": 2.3125,
      "step": 911
    },
    {
      "epoch": 0.6039735099337749,
      "grad_norm": 1.7357545254241622,
      "learning_rate": 0.0001368,
      "loss": 2.5625,
      "step": 912
    },
    {
      "epoch": 0.604635761589404,
      "grad_norm": 1.9213122814043826,
      "learning_rate": 0.00013695,
      "loss": 2.8438,
      "step": 913
    },
    {
      "epoch": 0.6052980132450331,
      "grad_norm": 1.649019879446452,
      "learning_rate": 0.0001371,
      "loss": 2.5312,
      "step": 914
    },
    {
      "epoch": 0.6059602649006622,
      "grad_norm": 1.686970394985107,
      "learning_rate": 0.00013725,
      "loss": 2.8594,
      "step": 915
    },
    {
      "epoch": 0.6066225165562914,
      "grad_norm": 1.7640827305030788,
      "learning_rate": 0.0001374,
      "loss": 2.3125,
      "step": 916
    },
    {
      "epoch": 0.6072847682119206,
      "grad_norm": 1.6961250615744448,
      "learning_rate": 0.00013754999999999998,
      "loss": 2.6094,
      "step": 917
    },
    {
      "epoch": 0.6079470198675496,
      "grad_norm": 2.016597384861727,
      "learning_rate": 0.00013769999999999999,
      "loss": 2.875,
      "step": 918
    },
    {
      "epoch": 0.6086092715231788,
      "grad_norm": 1.676860467349367,
      "learning_rate": 0.00013785,
      "loss": 2.7969,
      "step": 919
    },
    {
      "epoch": 0.609271523178808,
      "grad_norm": 2.4126555280345108,
      "learning_rate": 0.000138,
      "loss": 2.875,
      "step": 920
    },
    {
      "epoch": 0.6099337748344371,
      "grad_norm": 2.0035792229007314,
      "learning_rate": 0.00013815,
      "loss": 2.7656,
      "step": 921
    },
    {
      "epoch": 0.6105960264900663,
      "grad_norm": 1.6268260828469265,
      "learning_rate": 0.0001383,
      "loss": 2.7031,
      "step": 922
    },
    {
      "epoch": 0.6112582781456953,
      "grad_norm": 1.4593864314131886,
      "learning_rate": 0.00013845,
      "loss": 2.625,
      "step": 923
    },
    {
      "epoch": 0.6119205298013245,
      "grad_norm": 1.5424854855511951,
      "learning_rate": 0.0001386,
      "loss": 2.2031,
      "step": 924
    },
    {
      "epoch": 0.6125827814569537,
      "grad_norm": 2.75304929730052,
      "learning_rate": 0.00013874999999999998,
      "loss": 2.9219,
      "step": 925
    },
    {
      "epoch": 0.6132450331125828,
      "grad_norm": 2.0559003214152507,
      "learning_rate": 0.0001389,
      "loss": 2.7969,
      "step": 926
    },
    {
      "epoch": 0.6139072847682119,
      "grad_norm": 1.9994822373639374,
      "learning_rate": 0.00013905,
      "loss": 2.8594,
      "step": 927
    },
    {
      "epoch": 0.614569536423841,
      "grad_norm": 1.6849127577336724,
      "learning_rate": 0.0001392,
      "loss": 2.6875,
      "step": 928
    },
    {
      "epoch": 0.6152317880794702,
      "grad_norm": 1.8266263209664524,
      "learning_rate": 0.00013935,
      "loss": 2.7969,
      "step": 929
    },
    {
      "epoch": 0.6158940397350994,
      "grad_norm": 1.6556111186298015,
      "learning_rate": 0.0001395,
      "loss": 2.7188,
      "step": 930
    },
    {
      "epoch": 0.6165562913907284,
      "grad_norm": 1.5679796143528912,
      "learning_rate": 0.00013965,
      "loss": 2.5781,
      "step": 931
    },
    {
      "epoch": 0.6172185430463576,
      "grad_norm": 1.58608251661062,
      "learning_rate": 0.00013979999999999998,
      "loss": 2.7031,
      "step": 932
    },
    {
      "epoch": 0.6178807947019868,
      "grad_norm": 3.047964856430622,
      "learning_rate": 0.00013995,
      "loss": 2.7031,
      "step": 933
    },
    {
      "epoch": 0.6185430463576159,
      "grad_norm": 1.745642590387234,
      "learning_rate": 0.0001401,
      "loss": 2.625,
      "step": 934
    },
    {
      "epoch": 0.6192052980132451,
      "grad_norm": 1.6562030519640611,
      "learning_rate": 0.00014025,
      "loss": 2.7656,
      "step": 935
    },
    {
      "epoch": 0.6198675496688741,
      "grad_norm": 1.716649912453909,
      "learning_rate": 0.0001404,
      "loss": 2.75,
      "step": 936
    },
    {
      "epoch": 0.6205298013245033,
      "grad_norm": 1.5208361491671771,
      "learning_rate": 0.00014055,
      "loss": 2.75,
      "step": 937
    },
    {
      "epoch": 0.6211920529801325,
      "grad_norm": 1.70209635066516,
      "learning_rate": 0.00014069999999999998,
      "loss": 2.6719,
      "step": 938
    },
    {
      "epoch": 0.6218543046357616,
      "grad_norm": 1.6688952814411828,
      "learning_rate": 0.00014084999999999998,
      "loss": 2.7344,
      "step": 939
    },
    {
      "epoch": 0.6225165562913907,
      "grad_norm": 1.6280279993384466,
      "learning_rate": 0.00014099999999999998,
      "loss": 2.7969,
      "step": 940
    },
    {
      "epoch": 0.6231788079470199,
      "grad_norm": 1.6652758556387341,
      "learning_rate": 0.00014115,
      "loss": 2.7031,
      "step": 941
    },
    {
      "epoch": 0.623841059602649,
      "grad_norm": 1.739727663333885,
      "learning_rate": 0.0001413,
      "loss": 2.6562,
      "step": 942
    },
    {
      "epoch": 0.6245033112582782,
      "grad_norm": 1.737100148044706,
      "learning_rate": 0.00014144999999999997,
      "loss": 2.9531,
      "step": 943
    },
    {
      "epoch": 0.6251655629139072,
      "grad_norm": 1.8127289729274938,
      "learning_rate": 0.00014159999999999997,
      "loss": 2.7656,
      "step": 944
    },
    {
      "epoch": 0.6258278145695364,
      "grad_norm": 3.9350798549343593,
      "learning_rate": 0.00014174999999999998,
      "loss": 2.7812,
      "step": 945
    },
    {
      "epoch": 0.6264900662251656,
      "grad_norm": 1.840507817769587,
      "learning_rate": 0.00014189999999999998,
      "loss": 2.875,
      "step": 946
    },
    {
      "epoch": 0.6271523178807947,
      "grad_norm": 1.9119205332764484,
      "learning_rate": 0.00014204999999999998,
      "loss": 2.9375,
      "step": 947
    },
    {
      "epoch": 0.6278145695364239,
      "grad_norm": 1.7183680360721574,
      "learning_rate": 0.0001422,
      "loss": 2.8594,
      "step": 948
    },
    {
      "epoch": 0.628476821192053,
      "grad_norm": 1.6202688773542453,
      "learning_rate": 0.00014235,
      "loss": 2.625,
      "step": 949
    },
    {
      "epoch": 0.6291390728476821,
      "grad_norm": 1.7141548790172465,
      "learning_rate": 0.0001425,
      "loss": 2.75,
      "step": 950
    },
    {
      "epoch": 0.6298013245033113,
      "grad_norm": 1.623418135236663,
      "learning_rate": 0.00014264999999999997,
      "loss": 2.5625,
      "step": 951
    },
    {
      "epoch": 0.6304635761589404,
      "grad_norm": 1.6054248191490805,
      "learning_rate": 0.00014279999999999997,
      "loss": 2.3906,
      "step": 952
    },
    {
      "epoch": 0.6311258278145695,
      "grad_norm": 1.9691784389211007,
      "learning_rate": 0.00014294999999999998,
      "loss": 3.0625,
      "step": 953
    },
    {
      "epoch": 0.6317880794701987,
      "grad_norm": 1.912655902154441,
      "learning_rate": 0.00014309999999999998,
      "loss": 2.7188,
      "step": 954
    },
    {
      "epoch": 0.6324503311258278,
      "grad_norm": 1.6952050742878066,
      "learning_rate": 0.00014324999999999999,
      "loss": 2.5625,
      "step": 955
    },
    {
      "epoch": 0.633112582781457,
      "grad_norm": 3.434851730611255,
      "learning_rate": 0.0001434,
      "loss": 2.7812,
      "step": 956
    },
    {
      "epoch": 0.633774834437086,
      "grad_norm": 1.7388535992297727,
      "learning_rate": 0.00014355,
      "loss": 2.7969,
      "step": 957
    },
    {
      "epoch": 0.6344370860927152,
      "grad_norm": 1.6055065300055347,
      "learning_rate": 0.00014369999999999997,
      "loss": 2.8281,
      "step": 958
    },
    {
      "epoch": 0.6350993377483444,
      "grad_norm": 1.712475597222323,
      "learning_rate": 0.00014384999999999997,
      "loss": 2.6406,
      "step": 959
    },
    {
      "epoch": 0.6357615894039735,
      "grad_norm": 2.08335751499764,
      "learning_rate": 0.00014399999999999998,
      "loss": 2.8125,
      "step": 960
    },
    {
      "epoch": 0.6364238410596027,
      "grad_norm": 1.8171953911096144,
      "learning_rate": 0.00014414999999999998,
      "loss": 2.75,
      "step": 961
    },
    {
      "epoch": 0.6370860927152318,
      "grad_norm": 1.8735260700476095,
      "learning_rate": 0.00014429999999999998,
      "loss": 2.9219,
      "step": 962
    },
    {
      "epoch": 0.6377483443708609,
      "grad_norm": 1.8277704598450062,
      "learning_rate": 0.00014445,
      "loss": 2.8594,
      "step": 963
    },
    {
      "epoch": 0.6384105960264901,
      "grad_norm": 1.8210905265718296,
      "learning_rate": 0.0001446,
      "loss": 2.6562,
      "step": 964
    },
    {
      "epoch": 0.6390728476821192,
      "grad_norm": 1.505712954329362,
      "learning_rate": 0.00014475,
      "loss": 2.7656,
      "step": 965
    },
    {
      "epoch": 0.6397350993377483,
      "grad_norm": 1.918094753000461,
      "learning_rate": 0.00014489999999999997,
      "loss": 2.5938,
      "step": 966
    },
    {
      "epoch": 0.6403973509933775,
      "grad_norm": 1.767840050701229,
      "learning_rate": 0.00014504999999999997,
      "loss": 2.8281,
      "step": 967
    },
    {
      "epoch": 0.6410596026490066,
      "grad_norm": 1.653334231483156,
      "learning_rate": 0.00014519999999999998,
      "loss": 2.6875,
      "step": 968
    },
    {
      "epoch": 0.6417218543046358,
      "grad_norm": 2.0702035722707492,
      "learning_rate": 0.00014534999999999998,
      "loss": 3.0156,
      "step": 969
    },
    {
      "epoch": 0.6423841059602649,
      "grad_norm": 1.8596393640124391,
      "learning_rate": 0.00014549999999999999,
      "loss": 2.8281,
      "step": 970
    },
    {
      "epoch": 0.643046357615894,
      "grad_norm": 2.000203493554406,
      "learning_rate": 0.00014565,
      "loss": 2.6406,
      "step": 971
    },
    {
      "epoch": 0.6437086092715232,
      "grad_norm": 1.7274961291458457,
      "learning_rate": 0.0001458,
      "loss": 2.7969,
      "step": 972
    },
    {
      "epoch": 0.6443708609271523,
      "grad_norm": 1.9269270903513855,
      "learning_rate": 0.00014595,
      "loss": 2.3906,
      "step": 973
    },
    {
      "epoch": 0.6450331125827815,
      "grad_norm": 3.1400011087166444,
      "learning_rate": 0.00014609999999999997,
      "loss": 3.0625,
      "step": 974
    },
    {
      "epoch": 0.6456953642384106,
      "grad_norm": 1.7523819664212454,
      "learning_rate": 0.00014624999999999998,
      "loss": 2.6875,
      "step": 975
    },
    {
      "epoch": 0.6463576158940397,
      "grad_norm": 25.527572332226274,
      "learning_rate": 0.00014639999999999998,
      "loss": 3.0156,
      "step": 976
    },
    {
      "epoch": 0.6470198675496689,
      "grad_norm": 1.8407935632318109,
      "learning_rate": 0.00014654999999999998,
      "loss": 2.6719,
      "step": 977
    },
    {
      "epoch": 0.6476821192052981,
      "grad_norm": 1.9471448024888802,
      "learning_rate": 0.0001467,
      "loss": 2.6719,
      "step": 978
    },
    {
      "epoch": 0.6483443708609271,
      "grad_norm": 1.563538400749834,
      "learning_rate": 0.00014685,
      "loss": 2.5,
      "step": 979
    },
    {
      "epoch": 0.6490066225165563,
      "grad_norm": 1.916901970502345,
      "learning_rate": 0.000147,
      "loss": 2.7969,
      "step": 980
    },
    {
      "epoch": 0.6496688741721854,
      "grad_norm": 1.6939028458532706,
      "learning_rate": 0.00014714999999999997,
      "loss": 2.8594,
      "step": 981
    },
    {
      "epoch": 0.6503311258278146,
      "grad_norm": 1.671618081457123,
      "learning_rate": 0.00014729999999999998,
      "loss": 2.6406,
      "step": 982
    },
    {
      "epoch": 0.6509933774834437,
      "grad_norm": 2.3761553135046967,
      "learning_rate": 0.00014744999999999998,
      "loss": 3.1406,
      "step": 983
    },
    {
      "epoch": 0.6516556291390728,
      "grad_norm": 1.8718716012885332,
      "learning_rate": 0.00014759999999999998,
      "loss": 2.7031,
      "step": 984
    },
    {
      "epoch": 0.652317880794702,
      "grad_norm": 57.70429864989221,
      "learning_rate": 0.00014774999999999999,
      "loss": 3.1719,
      "step": 985
    },
    {
      "epoch": 0.6529801324503312,
      "grad_norm": 3.0096527716779415,
      "learning_rate": 0.0001479,
      "loss": 2.3281,
      "step": 986
    },
    {
      "epoch": 0.6536423841059603,
      "grad_norm": 82.97157024276366,
      "learning_rate": 0.00014805,
      "loss": 3.75,
      "step": 987
    },
    {
      "epoch": 0.6543046357615894,
      "grad_norm": 5.726878910172308,
      "learning_rate": 0.0001482,
      "loss": 3.1875,
      "step": 988
    },
    {
      "epoch": 0.6549668874172185,
      "grad_norm": 5.335748417256277,
      "learning_rate": 0.00014834999999999997,
      "loss": 3.125,
      "step": 989
    },
    {
      "epoch": 0.6556291390728477,
      "grad_norm": 3.9508229302334,
      "learning_rate": 0.00014849999999999998,
      "loss": 2.4688,
      "step": 990
    },
    {
      "epoch": 0.6562913907284769,
      "grad_norm": 3.9259101951527584,
      "learning_rate": 0.00014864999999999998,
      "loss": 2.8594,
      "step": 991
    },
    {
      "epoch": 0.6569536423841059,
      "grad_norm": 2.2498843377397946,
      "learning_rate": 0.00014879999999999998,
      "loss": 2.9219,
      "step": 992
    },
    {
      "epoch": 0.6576158940397351,
      "grad_norm": 2.7333916534135865,
      "learning_rate": 0.00014895,
      "loss": 2.7969,
      "step": 993
    },
    {
      "epoch": 0.6582781456953642,
      "grad_norm": 2.505994662255294,
      "learning_rate": 0.0001491,
      "loss": 2.8281,
      "step": 994
    },
    {
      "epoch": 0.6589403973509934,
      "grad_norm": 7.310629741063587,
      "learning_rate": 0.00014925,
      "loss": 2.7188,
      "step": 995
    },
    {
      "epoch": 0.6596026490066225,
      "grad_norm": 7.104904348913685,
      "learning_rate": 0.0001494,
      "loss": 2.875,
      "step": 996
    },
    {
      "epoch": 0.6602649006622516,
      "grad_norm": 2.574295802004421,
      "learning_rate": 0.00014954999999999998,
      "loss": 2.8281,
      "step": 997
    },
    {
      "epoch": 0.6609271523178808,
      "grad_norm": 2.1454345008062816,
      "learning_rate": 0.00014969999999999998,
      "loss": 2.9062,
      "step": 998
    },
    {
      "epoch": 0.66158940397351,
      "grad_norm": 10.795391745321732,
      "learning_rate": 0.00014984999999999998,
      "loss": 2.9062,
      "step": 999
    },
    {
      "epoch": 0.6622516556291391,
      "grad_norm": 10.886884174593044,
      "learning_rate": 0.00015,
      "loss": 2.6562,
      "step": 1000
    },
    {
      "epoch": 0.6629139072847682,
      "grad_norm": 2.6117790463445445,
      "learning_rate": 0.00015014999999999996,
      "loss": 2.3906,
      "step": 1001
    },
    {
      "epoch": 0.6635761589403973,
      "grad_norm": 1.9249744313079986,
      "learning_rate": 0.0001503,
      "loss": 2.4219,
      "step": 1002
    },
    {
      "epoch": 0.6642384105960265,
      "grad_norm": 1.8509413229809597,
      "learning_rate": 0.00015044999999999997,
      "loss": 2.5625,
      "step": 1003
    },
    {
      "epoch": 0.6649006622516557,
      "grad_norm": 4.9498943833068605,
      "learning_rate": 0.00015059999999999997,
      "loss": 3.3281,
      "step": 1004
    },
    {
      "epoch": 0.6655629139072847,
      "grad_norm": 2.4285161977290506,
      "learning_rate": 0.00015074999999999998,
      "loss": 2.7969,
      "step": 1005
    },
    {
      "epoch": 0.6662251655629139,
      "grad_norm": 1.983827443090624,
      "learning_rate": 0.00015089999999999998,
      "loss": 2.875,
      "step": 1006
    },
    {
      "epoch": 0.666887417218543,
      "grad_norm": 2.17557794399654,
      "learning_rate": 0.00015104999999999996,
      "loss": 2.6562,
      "step": 1007
    },
    {
      "epoch": 0.6675496688741722,
      "grad_norm": 1.9396356928330014,
      "learning_rate": 0.0001512,
      "loss": 2.7812,
      "step": 1008
    },
    {
      "epoch": 0.6682119205298013,
      "grad_norm": 1.7949651075224655,
      "learning_rate": 0.00015134999999999997,
      "loss": 2.6719,
      "step": 1009
    },
    {
      "epoch": 0.6688741721854304,
      "grad_norm": 1.7589255366193848,
      "learning_rate": 0.0001515,
      "loss": 2.8281,
      "step": 1010
    },
    {
      "epoch": 0.6695364238410596,
      "grad_norm": 1.738110057575885,
      "learning_rate": 0.00015164999999999997,
      "loss": 2.625,
      "step": 1011
    },
    {
      "epoch": 0.6701986754966888,
      "grad_norm": 1.827687088024332,
      "learning_rate": 0.00015179999999999998,
      "loss": 2.8125,
      "step": 1012
    },
    {
      "epoch": 0.6708609271523179,
      "grad_norm": 1.7129323981113438,
      "learning_rate": 0.00015194999999999998,
      "loss": 2.9688,
      "step": 1013
    },
    {
      "epoch": 0.671523178807947,
      "grad_norm": 1.8001785819020812,
      "learning_rate": 0.00015209999999999998,
      "loss": 2.5781,
      "step": 1014
    },
    {
      "epoch": 0.6721854304635762,
      "grad_norm": 6.554720271804624,
      "learning_rate": 0.00015224999999999996,
      "loss": 2.75,
      "step": 1015
    },
    {
      "epoch": 0.6728476821192053,
      "grad_norm": 1.7577700458290837,
      "learning_rate": 0.0001524,
      "loss": 2.1719,
      "step": 1016
    },
    {
      "epoch": 0.6735099337748345,
      "grad_norm": 1.9973660960273123,
      "learning_rate": 0.00015254999999999997,
      "loss": 2.8906,
      "step": 1017
    },
    {
      "epoch": 0.6741721854304635,
      "grad_norm": 4.426299592921524,
      "learning_rate": 0.0001527,
      "loss": 2.7344,
      "step": 1018
    },
    {
      "epoch": 0.6748344370860927,
      "grad_norm": 2.45074666365465,
      "learning_rate": 0.00015284999999999997,
      "loss": 2.625,
      "step": 1019
    },
    {
      "epoch": 0.6754966887417219,
      "grad_norm": 9.857719662659395,
      "learning_rate": 0.00015299999999999998,
      "loss": 3.1094,
      "step": 1020
    },
    {
      "epoch": 0.676158940397351,
      "grad_norm": 3.1033165148249817,
      "learning_rate": 0.00015314999999999998,
      "loss": 2.8125,
      "step": 1021
    },
    {
      "epoch": 0.6768211920529801,
      "grad_norm": 1.6451130588089793,
      "learning_rate": 0.00015329999999999999,
      "loss": 2.4844,
      "step": 1022
    },
    {
      "epoch": 0.6774834437086092,
      "grad_norm": 13.75871382888496,
      "learning_rate": 0.00015344999999999996,
      "loss": 3.0,
      "step": 1023
    },
    {
      "epoch": 0.6781456953642384,
      "grad_norm": 2.282975667806467,
      "learning_rate": 0.0001536,
      "loss": 2.9219,
      "step": 1024
    },
    {
      "epoch": 0.6788079470198676,
      "grad_norm": 7.514357103470541,
      "learning_rate": 0.00015374999999999997,
      "loss": 2.8594,
      "step": 1025
    },
    {
      "epoch": 0.6794701986754967,
      "grad_norm": 1.6854414816745074,
      "learning_rate": 0.0001539,
      "loss": 2.6094,
      "step": 1026
    },
    {
      "epoch": 0.6801324503311258,
      "grad_norm": 2.0967675538747717,
      "learning_rate": 0.00015404999999999998,
      "loss": 3.0312,
      "step": 1027
    },
    {
      "epoch": 0.680794701986755,
      "grad_norm": 6.4179893390034675,
      "learning_rate": 0.00015419999999999998,
      "loss": 3.1094,
      "step": 1028
    },
    {
      "epoch": 0.6814569536423841,
      "grad_norm": 1.9097735661715451,
      "learning_rate": 0.00015434999999999998,
      "loss": 3.0156,
      "step": 1029
    },
    {
      "epoch": 0.6821192052980133,
      "grad_norm": 2.5422131026892996,
      "learning_rate": 0.0001545,
      "loss": 2.8438,
      "step": 1030
    },
    {
      "epoch": 0.6827814569536423,
      "grad_norm": 1.927948696981671,
      "learning_rate": 0.00015464999999999996,
      "loss": 2.8438,
      "step": 1031
    },
    {
      "epoch": 0.6834437086092715,
      "grad_norm": 1.64953856454503,
      "learning_rate": 0.0001548,
      "loss": 2.2656,
      "step": 1032
    },
    {
      "epoch": 0.6841059602649007,
      "grad_norm": 1.6065647789242645,
      "learning_rate": 0.00015494999999999997,
      "loss": 2.5312,
      "step": 1033
    },
    {
      "epoch": 0.6847682119205298,
      "grad_norm": 2.040874751706338,
      "learning_rate": 0.0001551,
      "loss": 2.7656,
      "step": 1034
    },
    {
      "epoch": 0.6854304635761589,
      "grad_norm": 2.274138839387037,
      "learning_rate": 0.00015524999999999998,
      "loss": 2.7656,
      "step": 1035
    },
    {
      "epoch": 0.686092715231788,
      "grad_norm": 1.6322645562043083,
      "learning_rate": 0.00015539999999999998,
      "loss": 2.2031,
      "step": 1036
    },
    {
      "epoch": 0.6867549668874172,
      "grad_norm": 1.9914365557392222,
      "learning_rate": 0.00015554999999999999,
      "loss": 2.6719,
      "step": 1037
    },
    {
      "epoch": 0.6874172185430464,
      "grad_norm": 1.838952202627971,
      "learning_rate": 0.0001557,
      "loss": 2.8281,
      "step": 1038
    },
    {
      "epoch": 0.6880794701986755,
      "grad_norm": 1.7542502181405408,
      "learning_rate": 0.00015584999999999997,
      "loss": 2.5625,
      "step": 1039
    },
    {
      "epoch": 0.6887417218543046,
      "grad_norm": 1.8490427296896466,
      "learning_rate": 0.000156,
      "loss": 2.7656,
      "step": 1040
    },
    {
      "epoch": 0.6894039735099338,
      "grad_norm": 2.1268093649368414,
      "learning_rate": 0.00015614999999999997,
      "loss": 2.7656,
      "step": 1041
    },
    {
      "epoch": 0.6900662251655629,
      "grad_norm": 1.6469146169253956,
      "learning_rate": 0.0001563,
      "loss": 2.5312,
      "step": 1042
    },
    {
      "epoch": 0.6907284768211921,
      "grad_norm": 19.612814415289833,
      "learning_rate": 0.00015644999999999998,
      "loss": 2.875,
      "step": 1043
    },
    {
      "epoch": 0.6913907284768211,
      "grad_norm": 2.307580353944933,
      "learning_rate": 0.00015659999999999998,
      "loss": 2.7031,
      "step": 1044
    },
    {
      "epoch": 0.6920529801324503,
      "grad_norm": 3.7436321331711726,
      "learning_rate": 0.00015675,
      "loss": 2.6719,
      "step": 1045
    },
    {
      "epoch": 0.6927152317880795,
      "grad_norm": 3.4970479060494783,
      "learning_rate": 0.0001569,
      "loss": 2.6562,
      "step": 1046
    },
    {
      "epoch": 0.6933774834437086,
      "grad_norm": 4.035166923923399,
      "learning_rate": 0.00015704999999999997,
      "loss": 2.9688,
      "step": 1047
    },
    {
      "epoch": 0.6940397350993377,
      "grad_norm": 82.00195264618144,
      "learning_rate": 0.0001572,
      "loss": 3.0469,
      "step": 1048
    },
    {
      "epoch": 0.6947019867549669,
      "grad_norm": 2.111553657317677,
      "learning_rate": 0.00015734999999999998,
      "loss": 2.2812,
      "step": 1049
    },
    {
      "epoch": 0.695364238410596,
      "grad_norm": 1.848216894262673,
      "learning_rate": 0.00015749999999999998,
      "loss": 2.3125,
      "step": 1050
    },
    {
      "epoch": 0.6960264900662252,
      "grad_norm": 3.5719733930568136,
      "learning_rate": 0.00015764999999999998,
      "loss": 2.8438,
      "step": 1051
    },
    {
      "epoch": 0.6966887417218544,
      "grad_norm": 3.4325585709821635,
      "learning_rate": 0.0001578,
      "loss": 2.8281,
      "step": 1052
    },
    {
      "epoch": 0.6973509933774834,
      "grad_norm": 2.2526681835147433,
      "learning_rate": 0.00015794999999999996,
      "loss": 2.8906,
      "step": 1053
    },
    {
      "epoch": 0.6980132450331126,
      "grad_norm": 2.157125414703428,
      "learning_rate": 0.0001581,
      "loss": 2.7344,
      "step": 1054
    },
    {
      "epoch": 0.6986754966887417,
      "grad_norm": 2.3286727163211607,
      "learning_rate": 0.00015824999999999997,
      "loss": 2.9531,
      "step": 1055
    },
    {
      "epoch": 0.6993377483443709,
      "grad_norm": 13.342268056820014,
      "learning_rate": 0.0001584,
      "loss": 2.9375,
      "step": 1056
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.521314202240274,
      "learning_rate": 0.00015854999999999998,
      "loss": 2.875,
      "step": 1057
    },
    {
      "epoch": 0.7006622516556291,
      "grad_norm": 1.9241392967254913,
      "learning_rate": 0.00015869999999999998,
      "loss": 2.5,
      "step": 1058
    },
    {
      "epoch": 0.7013245033112583,
      "grad_norm": 7.157991129787909,
      "learning_rate": 0.00015884999999999999,
      "loss": 2.7344,
      "step": 1059
    },
    {
      "epoch": 0.7019867549668874,
      "grad_norm": 2.2843200199431157,
      "learning_rate": 0.000159,
      "loss": 2.7188,
      "step": 1060
    },
    {
      "epoch": 0.7026490066225165,
      "grad_norm": 1.670884075377115,
      "learning_rate": 0.00015914999999999997,
      "loss": 2.5625,
      "step": 1061
    },
    {
      "epoch": 0.7033112582781457,
      "grad_norm": 1.5757300597782906,
      "learning_rate": 0.0001593,
      "loss": 2.3438,
      "step": 1062
    },
    {
      "epoch": 0.7039735099337748,
      "grad_norm": 1.7203881141189485,
      "learning_rate": 0.00015944999999999997,
      "loss": 2.9375,
      "step": 1063
    },
    {
      "epoch": 0.704635761589404,
      "grad_norm": 1.8323788540640873,
      "learning_rate": 0.0001596,
      "loss": 3.1562,
      "step": 1064
    },
    {
      "epoch": 0.7052980132450332,
      "grad_norm": 4.966260446189065,
      "learning_rate": 0.00015974999999999998,
      "loss": 2.8594,
      "step": 1065
    },
    {
      "epoch": 0.7059602649006622,
      "grad_norm": 2.6036617400557485,
      "learning_rate": 0.00015989999999999998,
      "loss": 2.9219,
      "step": 1066
    },
    {
      "epoch": 0.7066225165562914,
      "grad_norm": 10.026339079934212,
      "learning_rate": 0.00016005,
      "loss": 2.8438,
      "step": 1067
    },
    {
      "epoch": 0.7072847682119205,
      "grad_norm": 1.779084842249553,
      "learning_rate": 0.0001602,
      "loss": 2.6719,
      "step": 1068
    },
    {
      "epoch": 0.7079470198675497,
      "grad_norm": 2.0472940875697088,
      "learning_rate": 0.00016034999999999997,
      "loss": 2.7656,
      "step": 1069
    },
    {
      "epoch": 0.7086092715231788,
      "grad_norm": 12.37951358298566,
      "learning_rate": 0.0001605,
      "loss": 2.7969,
      "step": 1070
    },
    {
      "epoch": 0.7092715231788079,
      "grad_norm": 2.5714672491432413,
      "learning_rate": 0.00016064999999999997,
      "loss": 2.8125,
      "step": 1071
    },
    {
      "epoch": 0.7099337748344371,
      "grad_norm": 3.2973828299741945,
      "learning_rate": 0.0001608,
      "loss": 3.125,
      "step": 1072
    },
    {
      "epoch": 0.7105960264900663,
      "grad_norm": 2.0773601601362355,
      "learning_rate": 0.00016094999999999998,
      "loss": 2.9219,
      "step": 1073
    },
    {
      "epoch": 0.7112582781456953,
      "grad_norm": 1.9576271793369373,
      "learning_rate": 0.00016109999999999999,
      "loss": 2.8594,
      "step": 1074
    },
    {
      "epoch": 0.7119205298013245,
      "grad_norm": 2.253081544525511,
      "learning_rate": 0.00016125,
      "loss": 2.8594,
      "step": 1075
    },
    {
      "epoch": 0.7125827814569536,
      "grad_norm": 17.849086542019425,
      "learning_rate": 0.0001614,
      "loss": 2.4688,
      "step": 1076
    },
    {
      "epoch": 0.7132450331125828,
      "grad_norm": 25.22103142092267,
      "learning_rate": 0.00016154999999999997,
      "loss": 3.2344,
      "step": 1077
    },
    {
      "epoch": 0.713907284768212,
      "grad_norm": 2.711755792004473,
      "learning_rate": 0.0001617,
      "loss": 2.9375,
      "step": 1078
    },
    {
      "epoch": 0.714569536423841,
      "grad_norm": 2.0429867634332206,
      "learning_rate": 0.00016184999999999998,
      "loss": 2.9531,
      "step": 1079
    },
    {
      "epoch": 0.7152317880794702,
      "grad_norm": 2.448420710327908,
      "learning_rate": 0.000162,
      "loss": 2.6875,
      "step": 1080
    },
    {
      "epoch": 0.7158940397350994,
      "grad_norm": 4.041013141288397,
      "learning_rate": 0.00016214999999999998,
      "loss": 3.2188,
      "step": 1081
    },
    {
      "epoch": 0.7165562913907285,
      "grad_norm": 2.207189529714852,
      "learning_rate": 0.0001623,
      "loss": 2.3906,
      "step": 1082
    },
    {
      "epoch": 0.7172185430463576,
      "grad_norm": 2.054389781140217,
      "learning_rate": 0.00016245,
      "loss": 3.1406,
      "step": 1083
    },
    {
      "epoch": 0.7178807947019867,
      "grad_norm": 9.055365864590497,
      "learning_rate": 0.0001626,
      "loss": 2.9688,
      "step": 1084
    },
    {
      "epoch": 0.7185430463576159,
      "grad_norm": 2.663922557828683,
      "learning_rate": 0.00016274999999999997,
      "loss": 2.8438,
      "step": 1085
    },
    {
      "epoch": 0.7192052980132451,
      "grad_norm": 2.7509920567552717,
      "learning_rate": 0.0001629,
      "loss": 2.8281,
      "step": 1086
    },
    {
      "epoch": 0.7198675496688741,
      "grad_norm": 2.1229102338276644,
      "learning_rate": 0.00016304999999999998,
      "loss": 2.5625,
      "step": 1087
    },
    {
      "epoch": 0.7205298013245033,
      "grad_norm": 4.655309629283799,
      "learning_rate": 0.0001632,
      "loss": 3.1406,
      "step": 1088
    },
    {
      "epoch": 0.7211920529801324,
      "grad_norm": 1.9126178330922006,
      "learning_rate": 0.00016334999999999999,
      "loss": 2.75,
      "step": 1089
    },
    {
      "epoch": 0.7218543046357616,
      "grad_norm": 1.8408699477695372,
      "learning_rate": 0.0001635,
      "loss": 2.7812,
      "step": 1090
    },
    {
      "epoch": 0.7225165562913908,
      "grad_norm": 1.7483772203157484,
      "learning_rate": 0.00016365,
      "loss": 2.7812,
      "step": 1091
    },
    {
      "epoch": 0.7231788079470198,
      "grad_norm": 1.8387604982818866,
      "learning_rate": 0.0001638,
      "loss": 2.7188,
      "step": 1092
    },
    {
      "epoch": 0.723841059602649,
      "grad_norm": 1.707319886269538,
      "learning_rate": 0.00016394999999999997,
      "loss": 2.8281,
      "step": 1093
    },
    {
      "epoch": 0.7245033112582782,
      "grad_norm": 3.0826758758914394,
      "learning_rate": 0.0001641,
      "loss": 3.1406,
      "step": 1094
    },
    {
      "epoch": 0.7251655629139073,
      "grad_norm": 2.3764448119774895,
      "learning_rate": 0.00016424999999999998,
      "loss": 2.4219,
      "step": 1095
    },
    {
      "epoch": 0.7258278145695364,
      "grad_norm": 3444.5459784831883,
      "learning_rate": 0.0001644,
      "loss": 10.25,
      "step": 1096
    },
    {
      "epoch": 0.7264900662251655,
      "grad_norm": 17.867142675853835,
      "learning_rate": 0.00016455,
      "loss": 4.1562,
      "step": 1097
    },
    {
      "epoch": 0.7271523178807947,
      "grad_norm": 49.17803503861254,
      "learning_rate": 0.0001647,
      "loss": 8.0,
      "step": 1098
    },
    {
      "epoch": 0.7278145695364239,
      "grad_norm": 13.75747538456727,
      "learning_rate": 0.00016485,
      "loss": 6.7188,
      "step": 1099
    },
    {
      "epoch": 0.7284768211920529,
      "grad_norm": 38.729371654488006,
      "learning_rate": 0.000165,
      "loss": 8.125,
      "step": 1100
    },
    {
      "epoch": 0.7291390728476821,
      "grad_norm": 16.74322854983273,
      "learning_rate": 0.00016514999999999998,
      "loss": 9.875,
      "step": 1101
    },
    {
      "epoch": 0.7298013245033113,
      "grad_norm": 9.992920049190671,
      "learning_rate": 0.0001653,
      "loss": 7.9375,
      "step": 1102
    },
    {
      "epoch": 0.7304635761589404,
      "grad_norm": 15.864214199617486,
      "learning_rate": 0.00016544999999999998,
      "loss": 6.8438,
      "step": 1103
    },
    {
      "epoch": 0.7311258278145696,
      "grad_norm": 8.989462769805998,
      "learning_rate": 0.0001656,
      "loss": 6.9688,
      "step": 1104
    },
    {
      "epoch": 0.7317880794701986,
      "grad_norm": 8.785505747260729,
      "learning_rate": 0.00016575,
      "loss": 7.3125,
      "step": 1105
    },
    {
      "epoch": 0.7324503311258278,
      "grad_norm": 11.864246352127116,
      "learning_rate": 0.0001659,
      "loss": 7.0625,
      "step": 1106
    },
    {
      "epoch": 0.733112582781457,
      "grad_norm": 11.880346608664418,
      "learning_rate": 0.00016604999999999997,
      "loss": 6.9688,
      "step": 1107
    },
    {
      "epoch": 0.7337748344370861,
      "grad_norm": 6.598027809605002,
      "learning_rate": 0.0001662,
      "loss": 6.25,
      "step": 1108
    },
    {
      "epoch": 0.7344370860927152,
      "grad_norm": 10.962003277808929,
      "learning_rate": 0.00016634999999999998,
      "loss": 6.3125,
      "step": 1109
    },
    {
      "epoch": 0.7350993377483444,
      "grad_norm": 11.981328536106767,
      "learning_rate": 0.0001665,
      "loss": 6.5938,
      "step": 1110
    },
    {
      "epoch": 0.7357615894039735,
      "grad_norm": 8.163918223592617,
      "learning_rate": 0.00016664999999999998,
      "loss": 6.2188,
      "step": 1111
    },
    {
      "epoch": 0.7364238410596027,
      "grad_norm": 8.281678213205572,
      "learning_rate": 0.0001668,
      "loss": 6.25,
      "step": 1112
    },
    {
      "epoch": 0.7370860927152317,
      "grad_norm": 4.036171711158652,
      "learning_rate": 0.00016695,
      "loss": 6.0938,
      "step": 1113
    },
    {
      "epoch": 0.7377483443708609,
      "grad_norm": 7.9542243143751215,
      "learning_rate": 0.0001671,
      "loss": 6.375,
      "step": 1114
    },
    {
      "epoch": 0.7384105960264901,
      "grad_norm": 3.367303358746958,
      "learning_rate": 0.00016724999999999997,
      "loss": 6.0,
      "step": 1115
    },
    {
      "epoch": 0.7390728476821192,
      "grad_norm": 4.621776213828051,
      "learning_rate": 0.0001674,
      "loss": 6.0938,
      "step": 1116
    },
    {
      "epoch": 0.7397350993377484,
      "grad_norm": 4.2742046020169315,
      "learning_rate": 0.00016754999999999998,
      "loss": 5.875,
      "step": 1117
    },
    {
      "epoch": 0.7403973509933774,
      "grad_norm": 2.6442755026214853,
      "learning_rate": 0.0001677,
      "loss": 5.8125,
      "step": 1118
    },
    {
      "epoch": 0.7410596026490066,
      "grad_norm": 4.969781805285279,
      "learning_rate": 0.00016785,
      "loss": 5.9062,
      "step": 1119
    },
    {
      "epoch": 0.7417218543046358,
      "grad_norm": 2.745704449580607,
      "learning_rate": 0.000168,
      "loss": 5.6562,
      "step": 1120
    },
    {
      "epoch": 0.7423841059602649,
      "grad_norm": 3.864828880700123,
      "learning_rate": 0.00016815,
      "loss": 5.6875,
      "step": 1121
    },
    {
      "epoch": 0.743046357615894,
      "grad_norm": 4.3791471954058085,
      "learning_rate": 0.0001683,
      "loss": 5.6875,
      "step": 1122
    },
    {
      "epoch": 0.7437086092715232,
      "grad_norm": 5.008083568549501,
      "learning_rate": 0.00016844999999999997,
      "loss": 5.5625,
      "step": 1123
    },
    {
      "epoch": 0.7443708609271523,
      "grad_norm": 3.6363444020555074,
      "learning_rate": 0.0001686,
      "loss": 5.6562,
      "step": 1124
    },
    {
      "epoch": 0.7450331125827815,
      "grad_norm": 3.721993365974379,
      "learning_rate": 0.00016874999999999998,
      "loss": 5.5938,
      "step": 1125
    },
    {
      "epoch": 0.7456953642384105,
      "grad_norm": 4.861834570471408,
      "learning_rate": 0.00016889999999999996,
      "loss": 5.5312,
      "step": 1126
    },
    {
      "epoch": 0.7463576158940397,
      "grad_norm": 3.1067774595793356,
      "learning_rate": 0.00016905,
      "loss": 5.4062,
      "step": 1127
    },
    {
      "epoch": 0.7470198675496689,
      "grad_norm": 2.939371966057707,
      "learning_rate": 0.00016919999999999997,
      "loss": 5.375,
      "step": 1128
    },
    {
      "epoch": 0.747682119205298,
      "grad_norm": 3.2585358377266367,
      "learning_rate": 0.00016935,
      "loss": 5.375,
      "step": 1129
    },
    {
      "epoch": 0.7483443708609272,
      "grad_norm": 3.3072919016763818,
      "learning_rate": 0.00016949999999999997,
      "loss": 5.25,
      "step": 1130
    },
    {
      "epoch": 0.7490066225165563,
      "grad_norm": 3.1650707861293674,
      "learning_rate": 0.00016964999999999998,
      "loss": 5.2188,
      "step": 1131
    },
    {
      "epoch": 0.7496688741721854,
      "grad_norm": 3.036197716763989,
      "learning_rate": 0.00016979999999999998,
      "loss": 5.2188,
      "step": 1132
    },
    {
      "epoch": 0.7503311258278146,
      "grad_norm": 3.127313215998794,
      "learning_rate": 0.00016994999999999998,
      "loss": 5.1875,
      "step": 1133
    },
    {
      "epoch": 0.7509933774834437,
      "grad_norm": 3.180130713822329,
      "learning_rate": 0.00017009999999999996,
      "loss": 5.2188,
      "step": 1134
    },
    {
      "epoch": 0.7516556291390728,
      "grad_norm": 2.4751777221760816,
      "learning_rate": 0.00017025,
      "loss": 5.0938,
      "step": 1135
    },
    {
      "epoch": 0.752317880794702,
      "grad_norm": 3.051657455882634,
      "learning_rate": 0.00017039999999999997,
      "loss": 5.125,
      "step": 1136
    },
    {
      "epoch": 0.7529801324503311,
      "grad_norm": 2.6189251628348194,
      "learning_rate": 0.00017055,
      "loss": 5.0625,
      "step": 1137
    },
    {
      "epoch": 0.7536423841059603,
      "grad_norm": 3.1082433355039942,
      "learning_rate": 0.00017069999999999998,
      "loss": 5.3125,
      "step": 1138
    },
    {
      "epoch": 0.7543046357615895,
      "grad_norm": 3.5379893101501905,
      "learning_rate": 0.00017084999999999998,
      "loss": 4.8438,
      "step": 1139
    },
    {
      "epoch": 0.7549668874172185,
      "grad_norm": 2.4137320275098935,
      "learning_rate": 0.00017099999999999998,
      "loss": 5.0938,
      "step": 1140
    },
    {
      "epoch": 0.7556291390728477,
      "grad_norm": 2.445364322628325,
      "learning_rate": 0.00017114999999999999,
      "loss": 4.875,
      "step": 1141
    },
    {
      "epoch": 0.7562913907284768,
      "grad_norm": 2.996873642910113,
      "learning_rate": 0.00017129999999999996,
      "loss": 5.0312,
      "step": 1142
    },
    {
      "epoch": 0.756953642384106,
      "grad_norm": 2.747291371714837,
      "learning_rate": 0.00017145,
      "loss": 4.875,
      "step": 1143
    },
    {
      "epoch": 0.7576158940397351,
      "grad_norm": 2.3558271492536136,
      "learning_rate": 0.00017159999999999997,
      "loss": 4.8438,
      "step": 1144
    },
    {
      "epoch": 0.7582781456953642,
      "grad_norm": 2.396160124853817,
      "learning_rate": 0.00017175,
      "loss": 4.9375,
      "step": 1145
    },
    {
      "epoch": 0.7589403973509934,
      "grad_norm": 2.969213369509106,
      "learning_rate": 0.00017189999999999998,
      "loss": 5.0,
      "step": 1146
    },
    {
      "epoch": 0.7596026490066226,
      "grad_norm": 2.2281532318148476,
      "learning_rate": 0.00017204999999999998,
      "loss": 5.0,
      "step": 1147
    },
    {
      "epoch": 0.7602649006622516,
      "grad_norm": 1.9678017337465405,
      "learning_rate": 0.00017219999999999998,
      "loss": 4.9375,
      "step": 1148
    },
    {
      "epoch": 0.7609271523178808,
      "grad_norm": 3.5458666927800766,
      "learning_rate": 0.00017235,
      "loss": 4.625,
      "step": 1149
    },
    {
      "epoch": 0.7615894039735099,
      "grad_norm": 2.1742722267479464,
      "learning_rate": 0.00017249999999999996,
      "loss": 4.9688,
      "step": 1150
    },
    {
      "epoch": 0.7622516556291391,
      "grad_norm": 3.906668526879908,
      "learning_rate": 0.00017265,
      "loss": 5.0312,
      "step": 1151
    },
    {
      "epoch": 0.7629139072847683,
      "grad_norm": 2.8571033177981904,
      "learning_rate": 0.00017279999999999997,
      "loss": 4.6875,
      "step": 1152
    },
    {
      "epoch": 0.7635761589403973,
      "grad_norm": 2.44066632890416,
      "learning_rate": 0.00017294999999999998,
      "loss": 4.75,
      "step": 1153
    },
    {
      "epoch": 0.7642384105960265,
      "grad_norm": 3.0635149614168893,
      "learning_rate": 0.00017309999999999998,
      "loss": 4.9062,
      "step": 1154
    },
    {
      "epoch": 0.7649006622516556,
      "grad_norm": 3.0418996654086006,
      "learning_rate": 0.00017324999999999998,
      "loss": 4.9688,
      "step": 1155
    },
    {
      "epoch": 0.7655629139072848,
      "grad_norm": 2.878370527327323,
      "learning_rate": 0.00017339999999999996,
      "loss": 4.8125,
      "step": 1156
    },
    {
      "epoch": 0.7662251655629139,
      "grad_norm": 3.566548171130857,
      "learning_rate": 0.00017355,
      "loss": 4.5938,
      "step": 1157
    },
    {
      "epoch": 0.766887417218543,
      "grad_norm": 2.3162439094836023,
      "learning_rate": 0.00017369999999999997,
      "loss": 4.5938,
      "step": 1158
    },
    {
      "epoch": 0.7675496688741722,
      "grad_norm": 3.617027262946174,
      "learning_rate": 0.00017385,
      "loss": 4.6562,
      "step": 1159
    },
    {
      "epoch": 0.7682119205298014,
      "grad_norm": 3.1546286741904983,
      "learning_rate": 0.00017399999999999997,
      "loss": 4.6562,
      "step": 1160
    },
    {
      "epoch": 0.7688741721854304,
      "grad_norm": 3.1997932902603687,
      "learning_rate": 0.00017414999999999998,
      "loss": 4.7812,
      "step": 1161
    },
    {
      "epoch": 0.7695364238410596,
      "grad_norm": 3.0074242982373787,
      "learning_rate": 0.00017429999999999998,
      "loss": 4.5625,
      "step": 1162
    },
    {
      "epoch": 0.7701986754966887,
      "grad_norm": 2.701669718999158,
      "learning_rate": 0.00017444999999999998,
      "loss": 4.75,
      "step": 1163
    },
    {
      "epoch": 0.7708609271523179,
      "grad_norm": 7.04521838374954,
      "learning_rate": 0.00017459999999999996,
      "loss": 4.6562,
      "step": 1164
    },
    {
      "epoch": 0.7715231788079471,
      "grad_norm": 2.7844368233784516,
      "learning_rate": 0.00017475,
      "loss": 4.7188,
      "step": 1165
    },
    {
      "epoch": 0.7721854304635761,
      "grad_norm": 2.5870788598183725,
      "learning_rate": 0.00017489999999999997,
      "loss": 4.5312,
      "step": 1166
    },
    {
      "epoch": 0.7728476821192053,
      "grad_norm": 3.0942102177798163,
      "learning_rate": 0.00017505,
      "loss": 4.625,
      "step": 1167
    },
    {
      "epoch": 0.7735099337748345,
      "grad_norm": 2.275111777542878,
      "learning_rate": 0.00017519999999999998,
      "loss": 4.625,
      "step": 1168
    },
    {
      "epoch": 0.7741721854304636,
      "grad_norm": 4.275960709796497,
      "learning_rate": 0.00017534999999999998,
      "loss": 4.4375,
      "step": 1169
    },
    {
      "epoch": 0.7748344370860927,
      "grad_norm": 3.6781087158113683,
      "learning_rate": 0.00017549999999999998,
      "loss": 4.5312,
      "step": 1170
    },
    {
      "epoch": 0.7754966887417218,
      "grad_norm": 3.550133481338501,
      "learning_rate": 0.00017565,
      "loss": 4.5312,
      "step": 1171
    },
    {
      "epoch": 0.776158940397351,
      "grad_norm": 2.328802091089066,
      "learning_rate": 0.00017579999999999996,
      "loss": 4.5938,
      "step": 1172
    },
    {
      "epoch": 0.7768211920529802,
      "grad_norm": 2.8815278322495557,
      "learning_rate": 0.00017595,
      "loss": 4.4688,
      "step": 1173
    },
    {
      "epoch": 0.7774834437086092,
      "grad_norm": 3.0685788093620197,
      "learning_rate": 0.00017609999999999997,
      "loss": 4.6562,
      "step": 1174
    },
    {
      "epoch": 0.7781456953642384,
      "grad_norm": 3.3633050476223203,
      "learning_rate": 0.00017625,
      "loss": 4.3125,
      "step": 1175
    },
    {
      "epoch": 0.7788079470198676,
      "grad_norm": 3.2921344274738247,
      "learning_rate": 0.00017639999999999998,
      "loss": 4.625,
      "step": 1176
    },
    {
      "epoch": 0.7794701986754967,
      "grad_norm": 3.0224334797484866,
      "learning_rate": 0.00017654999999999998,
      "loss": 4.3125,
      "step": 1177
    },
    {
      "epoch": 0.7801324503311259,
      "grad_norm": 4.310652612911602,
      "learning_rate": 0.00017669999999999999,
      "loss": 4.3438,
      "step": 1178
    },
    {
      "epoch": 0.7807947019867549,
      "grad_norm": 4.730327550566992,
      "learning_rate": 0.00017685,
      "loss": 4.5938,
      "step": 1179
    },
    {
      "epoch": 0.7814569536423841,
      "grad_norm": 3.7953680303299877,
      "learning_rate": 0.00017699999999999997,
      "loss": 4.3438,
      "step": 1180
    },
    {
      "epoch": 0.7821192052980133,
      "grad_norm": 3.9264451708090933,
      "learning_rate": 0.00017715,
      "loss": 4.375,
      "step": 1181
    },
    {
      "epoch": 0.7827814569536424,
      "grad_norm": 4.018045666660239,
      "learning_rate": 0.00017729999999999997,
      "loss": 4.7812,
      "step": 1182
    },
    {
      "epoch": 0.7834437086092715,
      "grad_norm": 4.8093140960148295,
      "learning_rate": 0.00017745,
      "loss": 4.2188,
      "step": 1183
    },
    {
      "epoch": 0.7841059602649006,
      "grad_norm": 3.7042612668074377,
      "learning_rate": 0.00017759999999999998,
      "loss": 4.4688,
      "step": 1184
    },
    {
      "epoch": 0.7847682119205298,
      "grad_norm": 2.8490327246696245,
      "learning_rate": 0.00017774999999999998,
      "loss": 4.25,
      "step": 1185
    },
    {
      "epoch": 0.785430463576159,
      "grad_norm": 3.0756689714784096,
      "learning_rate": 0.0001779,
      "loss": 4.25,
      "step": 1186
    },
    {
      "epoch": 0.786092715231788,
      "grad_norm": 3.210672943463751,
      "learning_rate": 0.00017805,
      "loss": 4.5,
      "step": 1187
    },
    {
      "epoch": 0.7867549668874172,
      "grad_norm": 2.6754651788415327,
      "learning_rate": 0.00017819999999999997,
      "loss": 4.25,
      "step": 1188
    },
    {
      "epoch": 0.7874172185430464,
      "grad_norm": 19.95414281009093,
      "learning_rate": 0.00017835,
      "loss": 4.1875,
      "step": 1189
    },
    {
      "epoch": 0.7880794701986755,
      "grad_norm": 9.188228181131823,
      "learning_rate": 0.00017849999999999997,
      "loss": 4.5312,
      "step": 1190
    },
    {
      "epoch": 0.7887417218543047,
      "grad_norm": 5.779872680125628,
      "learning_rate": 0.00017865,
      "loss": 4.4375,
      "step": 1191
    },
    {
      "epoch": 0.7894039735099337,
      "grad_norm": 5.593414000170351,
      "learning_rate": 0.00017879999999999998,
      "loss": 4.4062,
      "step": 1192
    },
    {
      "epoch": 0.7900662251655629,
      "grad_norm": 6.6034356606854185,
      "learning_rate": 0.00017894999999999999,
      "loss": 4.5312,
      "step": 1193
    },
    {
      "epoch": 0.7907284768211921,
      "grad_norm": 4.821609846684629,
      "learning_rate": 0.0001791,
      "loss": 4.0312,
      "step": 1194
    },
    {
      "epoch": 0.7913907284768212,
      "grad_norm": 4.829948455363373,
      "learning_rate": 0.00017925,
      "loss": 4.4688,
      "step": 1195
    },
    {
      "epoch": 0.7920529801324503,
      "grad_norm": 2.6500477647093903,
      "learning_rate": 0.00017939999999999997,
      "loss": 4.5,
      "step": 1196
    },
    {
      "epoch": 0.7927152317880795,
      "grad_norm": 2.8810942989297685,
      "learning_rate": 0.00017955,
      "loss": 4.1875,
      "step": 1197
    },
    {
      "epoch": 0.7933774834437086,
      "grad_norm": 3.1668885708924988,
      "learning_rate": 0.00017969999999999998,
      "loss": 4.3438,
      "step": 1198
    },
    {
      "epoch": 0.7940397350993378,
      "grad_norm": 3.712795619492887,
      "learning_rate": 0.00017984999999999998,
      "loss": 4.4375,
      "step": 1199
    },
    {
      "epoch": 0.7947019867549668,
      "grad_norm": 2.4360986873577852,
      "learning_rate": 0.00017999999999999998,
      "loss": 4.2188,
      "step": 1200
    },
    {
      "epoch": 0.795364238410596,
      "grad_norm": 3.2064211835310794,
      "learning_rate": 0.00018015,
      "loss": 4.0625,
      "step": 1201
    },
    {
      "epoch": 0.7960264900662252,
      "grad_norm": 2.837892589566235,
      "learning_rate": 0.00018029999999999996,
      "loss": 4.0,
      "step": 1202
    },
    {
      "epoch": 0.7966887417218543,
      "grad_norm": 2.845337413652026,
      "learning_rate": 0.00018045,
      "loss": 4.3125,
      "step": 1203
    },
    {
      "epoch": 0.7973509933774835,
      "grad_norm": 3.5258121209948476,
      "learning_rate": 0.00018059999999999997,
      "loss": 4.3438,
      "step": 1204
    },
    {
      "epoch": 0.7980132450331126,
      "grad_norm": 2.7474604714915842,
      "learning_rate": 0.00018075,
      "loss": 4.2188,
      "step": 1205
    },
    {
      "epoch": 0.7986754966887417,
      "grad_norm": 2.2968730034268514,
      "learning_rate": 0.00018089999999999998,
      "loss": 4.0312,
      "step": 1206
    },
    {
      "epoch": 0.7993377483443709,
      "grad_norm": 2.6714874146838268,
      "learning_rate": 0.00018104999999999998,
      "loss": 3.875,
      "step": 1207
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.07964517217086,
      "learning_rate": 0.00018119999999999999,
      "loss": 4.25,
      "step": 1208
    },
    {
      "epoch": 0.8006622516556291,
      "grad_norm": 2.435872017997588,
      "learning_rate": 0.00018135,
      "loss": 4.0312,
      "step": 1209
    },
    {
      "epoch": 0.8013245033112583,
      "grad_norm": 2.927330196530365,
      "learning_rate": 0.00018149999999999997,
      "loss": 4.0312,
      "step": 1210
    },
    {
      "epoch": 0.8019867549668874,
      "grad_norm": 2.5841002857107793,
      "learning_rate": 0.00018165,
      "loss": 4.2812,
      "step": 1211
    },
    {
      "epoch": 0.8026490066225166,
      "grad_norm": 2.5491835847603266,
      "learning_rate": 0.00018179999999999997,
      "loss": 3.8438,
      "step": 1212
    },
    {
      "epoch": 0.8033112582781456,
      "grad_norm": 3.121414494504496,
      "learning_rate": 0.00018195,
      "loss": 3.7188,
      "step": 1213
    },
    {
      "epoch": 0.8039735099337748,
      "grad_norm": 1.9150506810556445,
      "learning_rate": 0.00018209999999999998,
      "loss": 3.875,
      "step": 1214
    },
    {
      "epoch": 0.804635761589404,
      "grad_norm": 2.427306225432991,
      "learning_rate": 0.00018224999999999998,
      "loss": 3.9375,
      "step": 1215
    },
    {
      "epoch": 0.8052980132450331,
      "grad_norm": 2.3211485574252784,
      "learning_rate": 0.0001824,
      "loss": 4.1562,
      "step": 1216
    },
    {
      "epoch": 0.8059602649006623,
      "grad_norm": 2.3610528809973896,
      "learning_rate": 0.00018255,
      "loss": 3.7812,
      "step": 1217
    },
    {
      "epoch": 0.8066225165562914,
      "grad_norm": 2.331605604821229,
      "learning_rate": 0.00018269999999999997,
      "loss": 4.0625,
      "step": 1218
    },
    {
      "epoch": 0.8072847682119205,
      "grad_norm": 2.6490920241212472,
      "learning_rate": 0.00018285,
      "loss": 3.6406,
      "step": 1219
    },
    {
      "epoch": 0.8079470198675497,
      "grad_norm": 2.1481865914914713,
      "learning_rate": 0.00018299999999999998,
      "loss": 4.0312,
      "step": 1220
    },
    {
      "epoch": 0.8086092715231789,
      "grad_norm": 2.7477610714638305,
      "learning_rate": 0.00018315,
      "loss": 3.8438,
      "step": 1221
    },
    {
      "epoch": 0.8092715231788079,
      "grad_norm": 2.4444429311502294,
      "learning_rate": 0.00018329999999999998,
      "loss": 3.9531,
      "step": 1222
    },
    {
      "epoch": 0.8099337748344371,
      "grad_norm": 2.415903654978864,
      "learning_rate": 0.00018345,
      "loss": 3.9688,
      "step": 1223
    },
    {
      "epoch": 0.8105960264900662,
      "grad_norm": 2.301674740187517,
      "learning_rate": 0.0001836,
      "loss": 3.9219,
      "step": 1224
    },
    {
      "epoch": 0.8112582781456954,
      "grad_norm": 2.440961166859283,
      "learning_rate": 0.00018375,
      "loss": 3.7656,
      "step": 1225
    },
    {
      "epoch": 0.8119205298013245,
      "grad_norm": 2.038892872401617,
      "learning_rate": 0.00018389999999999997,
      "loss": 4.0,
      "step": 1226
    },
    {
      "epoch": 0.8125827814569536,
      "grad_norm": 3.1551766935288916,
      "learning_rate": 0.00018405,
      "loss": 3.9688,
      "step": 1227
    },
    {
      "epoch": 0.8132450331125828,
      "grad_norm": 2.1976280186158776,
      "learning_rate": 0.00018419999999999998,
      "loss": 4.0,
      "step": 1228
    },
    {
      "epoch": 0.813907284768212,
      "grad_norm": 2.97147560371195,
      "learning_rate": 0.00018435,
      "loss": 3.8125,
      "step": 1229
    },
    {
      "epoch": 0.8145695364238411,
      "grad_norm": 2.0671922356715418,
      "learning_rate": 0.00018449999999999999,
      "loss": 3.5938,
      "step": 1230
    },
    {
      "epoch": 0.8152317880794702,
      "grad_norm": 2.26941514711728,
      "learning_rate": 0.00018465,
      "loss": 3.8438,
      "step": 1231
    },
    {
      "epoch": 0.8158940397350993,
      "grad_norm": 2.3056419546058144,
      "learning_rate": 0.0001848,
      "loss": 3.9844,
      "step": 1232
    },
    {
      "epoch": 0.8165562913907285,
      "grad_norm": 4.82203660223364,
      "learning_rate": 0.00018495,
      "loss": 3.3906,
      "step": 1233
    },
    {
      "epoch": 0.8172185430463577,
      "grad_norm": 2.3594325093544266,
      "learning_rate": 0.00018509999999999997,
      "loss": 3.6875,
      "step": 1234
    },
    {
      "epoch": 0.8178807947019867,
      "grad_norm": 5.277144784562463,
      "learning_rate": 0.00018525,
      "loss": 3.375,
      "step": 1235
    },
    {
      "epoch": 0.8185430463576159,
      "grad_norm": 3.0511893627282034,
      "learning_rate": 0.00018539999999999998,
      "loss": 3.9531,
      "step": 1236
    },
    {
      "epoch": 0.819205298013245,
      "grad_norm": 2.8432612415683716,
      "learning_rate": 0.00018555,
      "loss": 3.8125,
      "step": 1237
    },
    {
      "epoch": 0.8198675496688742,
      "grad_norm": 3.606638322133848,
      "learning_rate": 0.0001857,
      "loss": 3.8906,
      "step": 1238
    },
    {
      "epoch": 0.8205298013245033,
      "grad_norm": 2.9541853612007865,
      "learning_rate": 0.00018585,
      "loss": 3.7969,
      "step": 1239
    },
    {
      "epoch": 0.8211920529801324,
      "grad_norm": 2.0184868079297926,
      "learning_rate": 0.000186,
      "loss": 3.875,
      "step": 1240
    },
    {
      "epoch": 0.8218543046357616,
      "grad_norm": 2.260662031836412,
      "learning_rate": 0.00018615,
      "loss": 3.9375,
      "step": 1241
    },
    {
      "epoch": 0.8225165562913908,
      "grad_norm": 2.7224847798881497,
      "learning_rate": 0.00018629999999999997,
      "loss": 3.7812,
      "step": 1242
    },
    {
      "epoch": 0.8231788079470199,
      "grad_norm": 3.646639369145067,
      "learning_rate": 0.00018645,
      "loss": 3.8281,
      "step": 1243
    },
    {
      "epoch": 0.823841059602649,
      "grad_norm": 3.0261305484025964,
      "learning_rate": 0.00018659999999999998,
      "loss": 3.2969,
      "step": 1244
    },
    {
      "epoch": 0.8245033112582781,
      "grad_norm": 2.542471918682892,
      "learning_rate": 0.00018675,
      "loss": 3.7344,
      "step": 1245
    },
    {
      "epoch": 0.8251655629139073,
      "grad_norm": 2.659117751392304,
      "learning_rate": 0.0001869,
      "loss": 4.0312,
      "step": 1246
    },
    {
      "epoch": 0.8258278145695365,
      "grad_norm": 2.134016312158818,
      "learning_rate": 0.00018705,
      "loss": 3.75,
      "step": 1247
    },
    {
      "epoch": 0.8264900662251655,
      "grad_norm": 2.3472444130810306,
      "learning_rate": 0.0001872,
      "loss": 3.75,
      "step": 1248
    },
    {
      "epoch": 0.8271523178807947,
      "grad_norm": 2.7658089412372338,
      "learning_rate": 0.00018735,
      "loss": 3.6719,
      "step": 1249
    },
    {
      "epoch": 0.8278145695364238,
      "grad_norm": 2.1061883116592552,
      "learning_rate": 0.00018749999999999998,
      "loss": 3.5156,
      "step": 1250
    },
    {
      "epoch": 0.828476821192053,
      "grad_norm": 2.8141025758938856,
      "learning_rate": 0.00018764999999999998,
      "loss": 3.6094,
      "step": 1251
    },
    {
      "epoch": 0.8291390728476821,
      "grad_norm": 2.0635282223933387,
      "learning_rate": 0.00018779999999999998,
      "loss": 3.6406,
      "step": 1252
    },
    {
      "epoch": 0.8298013245033112,
      "grad_norm": 2.247434687275756,
      "learning_rate": 0.00018794999999999996,
      "loss": 3.3125,
      "step": 1253
    },
    {
      "epoch": 0.8304635761589404,
      "grad_norm": 2.395702865018192,
      "learning_rate": 0.0001881,
      "loss": 3.4688,
      "step": 1254
    },
    {
      "epoch": 0.8311258278145696,
      "grad_norm": 2.0580642792743125,
      "learning_rate": 0.00018824999999999997,
      "loss": 3.5781,
      "step": 1255
    },
    {
      "epoch": 0.8317880794701987,
      "grad_norm": 2.2153006546281553,
      "learning_rate": 0.00018839999999999997,
      "loss": 3.375,
      "step": 1256
    },
    {
      "epoch": 0.8324503311258278,
      "grad_norm": 2.886146167349155,
      "learning_rate": 0.00018854999999999998,
      "loss": 3.0312,
      "step": 1257
    },
    {
      "epoch": 0.833112582781457,
      "grad_norm": 2.489217698079595,
      "learning_rate": 0.00018869999999999998,
      "loss": 3.8125,
      "step": 1258
    },
    {
      "epoch": 0.8337748344370861,
      "grad_norm": 3.8364593763290333,
      "learning_rate": 0.00018884999999999996,
      "loss": 3.4688,
      "step": 1259
    },
    {
      "epoch": 0.8344370860927153,
      "grad_norm": 2.3706238227239775,
      "learning_rate": 0.00018899999999999999,
      "loss": 3.3438,
      "step": 1260
    },
    {
      "epoch": 0.8350993377483443,
      "grad_norm": 2.6372026366170127,
      "learning_rate": 0.00018914999999999996,
      "loss": 3.0312,
      "step": 1261
    },
    {
      "epoch": 0.8357615894039735,
      "grad_norm": 2.3995036789981734,
      "learning_rate": 0.0001893,
      "loss": 3.1875,
      "step": 1262
    },
    {
      "epoch": 0.8364238410596027,
      "grad_norm": 2.9600273894537295,
      "learning_rate": 0.00018944999999999997,
      "loss": 3.0312,
      "step": 1263
    },
    {
      "epoch": 0.8370860927152318,
      "grad_norm": 2.801629215762054,
      "learning_rate": 0.00018959999999999997,
      "loss": 3.5312,
      "step": 1264
    },
    {
      "epoch": 0.8377483443708609,
      "grad_norm": 2.601399129858437,
      "learning_rate": 0.00018974999999999998,
      "loss": 3.625,
      "step": 1265
    },
    {
      "epoch": 0.83841059602649,
      "grad_norm": 2.442192861302586,
      "learning_rate": 0.00018989999999999998,
      "loss": 3.4531,
      "step": 1266
    },
    {
      "epoch": 0.8390728476821192,
      "grad_norm": 2.2439706155054155,
      "learning_rate": 0.00019004999999999996,
      "loss": 3.3125,
      "step": 1267
    },
    {
      "epoch": 0.8397350993377484,
      "grad_norm": 2.2463404716809983,
      "learning_rate": 0.0001902,
      "loss": 3.3281,
      "step": 1268
    },
    {
      "epoch": 0.8403973509933775,
      "grad_norm": 1.9614469501235225,
      "learning_rate": 0.00019034999999999996,
      "loss": 3.375,
      "step": 1269
    },
    {
      "epoch": 0.8410596026490066,
      "grad_norm": 1.797348593299781,
      "learning_rate": 0.0001905,
      "loss": 3.2188,
      "step": 1270
    },
    {
      "epoch": 0.8417218543046358,
      "grad_norm": 1.9667144495040036,
      "learning_rate": 0.00019064999999999997,
      "loss": 3.3125,
      "step": 1271
    },
    {
      "epoch": 0.8423841059602649,
      "grad_norm": 4.360274011206607,
      "learning_rate": 0.00019079999999999998,
      "loss": 2.875,
      "step": 1272
    },
    {
      "epoch": 0.8430463576158941,
      "grad_norm": 2.830328621433694,
      "learning_rate": 0.00019094999999999998,
      "loss": 3.3906,
      "step": 1273
    },
    {
      "epoch": 0.8437086092715231,
      "grad_norm": 2.0166246119722397,
      "learning_rate": 0.00019109999999999998,
      "loss": 3.3906,
      "step": 1274
    },
    {
      "epoch": 0.8443708609271523,
      "grad_norm": 2.4632450806613044,
      "learning_rate": 0.00019124999999999996,
      "loss": 3.0781,
      "step": 1275
    },
    {
      "epoch": 0.8450331125827815,
      "grad_norm": 3.041269482810716,
      "learning_rate": 0.0001914,
      "loss": 3.25,
      "step": 1276
    },
    {
      "epoch": 0.8456953642384106,
      "grad_norm": 2.094047835217932,
      "learning_rate": 0.00019154999999999997,
      "loss": 2.7656,
      "step": 1277
    },
    {
      "epoch": 0.8463576158940397,
      "grad_norm": 2.2279015293145012,
      "learning_rate": 0.0001917,
      "loss": 3.1562,
      "step": 1278
    },
    {
      "epoch": 0.8470198675496688,
      "grad_norm": 2.1197165579219033,
      "learning_rate": 0.00019184999999999997,
      "loss": 3.0625,
      "step": 1279
    },
    {
      "epoch": 0.847682119205298,
      "grad_norm": 1.9285421104999607,
      "learning_rate": 0.00019199999999999998,
      "loss": 3.0938,
      "step": 1280
    },
    {
      "epoch": 0.8483443708609272,
      "grad_norm": 1.9637823823706557,
      "learning_rate": 0.00019214999999999998,
      "loss": 3.0938,
      "step": 1281
    },
    {
      "epoch": 0.8490066225165563,
      "grad_norm": 2.102949037342207,
      "learning_rate": 0.00019229999999999999,
      "loss": 3.3125,
      "step": 1282
    },
    {
      "epoch": 0.8496688741721854,
      "grad_norm": 12.934159881862051,
      "learning_rate": 0.00019244999999999996,
      "loss": 3.0312,
      "step": 1283
    },
    {
      "epoch": 0.8503311258278146,
      "grad_norm": 2.5260629999943522,
      "learning_rate": 0.0001926,
      "loss": 3.1094,
      "step": 1284
    },
    {
      "epoch": 0.8509933774834437,
      "grad_norm": 3.0771487887927162,
      "learning_rate": 0.00019274999999999997,
      "loss": 2.4844,
      "step": 1285
    },
    {
      "epoch": 0.8516556291390729,
      "grad_norm": 12.293546759115715,
      "learning_rate": 0.0001929,
      "loss": 3.0781,
      "step": 1286
    },
    {
      "epoch": 0.8523178807947019,
      "grad_norm": 4.2734784355264335,
      "learning_rate": 0.00019304999999999998,
      "loss": 3.3281,
      "step": 1287
    },
    {
      "epoch": 0.8529801324503311,
      "grad_norm": 6.514369784301302,
      "learning_rate": 0.00019319999999999998,
      "loss": 3.2969,
      "step": 1288
    },
    {
      "epoch": 0.8536423841059603,
      "grad_norm": 2.480476949100863,
      "learning_rate": 0.00019334999999999998,
      "loss": 3.1094,
      "step": 1289
    },
    {
      "epoch": 0.8543046357615894,
      "grad_norm": 2.5250184781188767,
      "learning_rate": 0.0001935,
      "loss": 3.2969,
      "step": 1290
    },
    {
      "epoch": 0.8549668874172185,
      "grad_norm": 2.5897213788116296,
      "learning_rate": 0.00019364999999999996,
      "loss": 3.0,
      "step": 1291
    },
    {
      "epoch": 0.8556291390728477,
      "grad_norm": 2.1751238634615775,
      "learning_rate": 0.0001938,
      "loss": 3.0781,
      "step": 1292
    },
    {
      "epoch": 0.8562913907284768,
      "grad_norm": 2.003691957663341,
      "learning_rate": 0.00019394999999999997,
      "loss": 3.0312,
      "step": 1293
    },
    {
      "epoch": 0.856953642384106,
      "grad_norm": 3.425486841135925,
      "learning_rate": 0.0001941,
      "loss": 3.0156,
      "step": 1294
    },
    {
      "epoch": 0.8576158940397351,
      "grad_norm": 2.2281731877279447,
      "learning_rate": 0.00019424999999999998,
      "loss": 3.0312,
      "step": 1295
    },
    {
      "epoch": 0.8582781456953642,
      "grad_norm": 1.902503861135649,
      "learning_rate": 0.00019439999999999998,
      "loss": 3.1562,
      "step": 1296
    },
    {
      "epoch": 0.8589403973509934,
      "grad_norm": 2.937040665800666,
      "learning_rate": 0.00019454999999999999,
      "loss": 3.4688,
      "step": 1297
    },
    {
      "epoch": 0.8596026490066225,
      "grad_norm": 1.8101821516295078,
      "learning_rate": 0.0001947,
      "loss": 3.0781,
      "step": 1298
    },
    {
      "epoch": 0.8602649006622517,
      "grad_norm": 1.9145431161306536,
      "learning_rate": 0.00019484999999999997,
      "loss": 3.0156,
      "step": 1299
    },
    {
      "epoch": 0.8609271523178808,
      "grad_norm": 2.0362130933049416,
      "learning_rate": 0.000195,
      "loss": 2.9062,
      "step": 1300
    },
    {
      "epoch": 0.8615894039735099,
      "grad_norm": 1.6833551132776159,
      "learning_rate": 0.00019514999999999997,
      "loss": 2.8594,
      "step": 1301
    },
    {
      "epoch": 0.8622516556291391,
      "grad_norm": 3.2980914550470635,
      "learning_rate": 0.00019529999999999998,
      "loss": 3.0469,
      "step": 1302
    },
    {
      "epoch": 0.8629139072847682,
      "grad_norm": 2.2716150420434436,
      "learning_rate": 0.00019544999999999998,
      "loss": 2.7969,
      "step": 1303
    },
    {
      "epoch": 0.8635761589403973,
      "grad_norm": 1.866670695652504,
      "learning_rate": 0.00019559999999999998,
      "loss": 3.0312,
      "step": 1304
    },
    {
      "epoch": 0.8642384105960265,
      "grad_norm": 2.2407126827811457,
      "learning_rate": 0.00019574999999999996,
      "loss": 2.8906,
      "step": 1305
    },
    {
      "epoch": 0.8649006622516556,
      "grad_norm": 1.4297108464832682,
      "learning_rate": 0.0001959,
      "loss": 2.7812,
      "step": 1306
    },
    {
      "epoch": 0.8655629139072848,
      "grad_norm": 2.5384163063915834,
      "learning_rate": 0.00019604999999999997,
      "loss": 2.9375,
      "step": 1307
    },
    {
      "epoch": 0.866225165562914,
      "grad_norm": 1.7005327695549182,
      "learning_rate": 0.0001962,
      "loss": 2.9375,
      "step": 1308
    },
    {
      "epoch": 0.866887417218543,
      "grad_norm": 1.6533228202857764,
      "learning_rate": 0.00019634999999999998,
      "loss": 2.3438,
      "step": 1309
    },
    {
      "epoch": 0.8675496688741722,
      "grad_norm": 1.6390461491230774,
      "learning_rate": 0.00019649999999999998,
      "loss": 2.8281,
      "step": 1310
    },
    {
      "epoch": 0.8682119205298013,
      "grad_norm": 2.205631297152919,
      "learning_rate": 0.00019664999999999998,
      "loss": 3.1406,
      "step": 1311
    },
    {
      "epoch": 0.8688741721854305,
      "grad_norm": 1.7256990002269397,
      "learning_rate": 0.00019679999999999999,
      "loss": 2.3906,
      "step": 1312
    },
    {
      "epoch": 0.8695364238410596,
      "grad_norm": 1.8440750787405453,
      "learning_rate": 0.00019694999999999996,
      "loss": 3.0156,
      "step": 1313
    },
    {
      "epoch": 0.8701986754966887,
      "grad_norm": 1.6721072079437616,
      "learning_rate": 0.0001971,
      "loss": 3.1719,
      "step": 1314
    },
    {
      "epoch": 0.8708609271523179,
      "grad_norm": 3.0880752748553926,
      "learning_rate": 0.00019724999999999997,
      "loss": 2.875,
      "step": 1315
    },
    {
      "epoch": 0.871523178807947,
      "grad_norm": 1.591474979173941,
      "learning_rate": 0.0001974,
      "loss": 2.75,
      "step": 1316
    },
    {
      "epoch": 0.8721854304635761,
      "grad_norm": 1.499523289724405,
      "learning_rate": 0.00019754999999999998,
      "loss": 2.5625,
      "step": 1317
    },
    {
      "epoch": 0.8728476821192053,
      "grad_norm": 1.7969873592866519,
      "learning_rate": 0.00019769999999999998,
      "loss": 3.0781,
      "step": 1318
    },
    {
      "epoch": 0.8735099337748344,
      "grad_norm": 1.5605144901323431,
      "learning_rate": 0.00019784999999999998,
      "loss": 3.0312,
      "step": 1319
    },
    {
      "epoch": 0.8741721854304636,
      "grad_norm": 1.9429963964773207,
      "learning_rate": 0.000198,
      "loss": 3.0625,
      "step": 1320
    },
    {
      "epoch": 0.8748344370860928,
      "grad_norm": 1.55742681982741,
      "learning_rate": 0.00019814999999999996,
      "loss": 2.7969,
      "step": 1321
    },
    {
      "epoch": 0.8754966887417218,
      "grad_norm": 1.5818051331928482,
      "learning_rate": 0.0001983,
      "loss": 2.75,
      "step": 1322
    },
    {
      "epoch": 0.876158940397351,
      "grad_norm": 1.4961222930268283,
      "learning_rate": 0.00019844999999999997,
      "loss": 2.9062,
      "step": 1323
    },
    {
      "epoch": 0.8768211920529801,
      "grad_norm": 5.03589782682638,
      "learning_rate": 0.0001986,
      "loss": 2.8281,
      "step": 1324
    },
    {
      "epoch": 0.8774834437086093,
      "grad_norm": 1.8509996192290328,
      "learning_rate": 0.00019874999999999998,
      "loss": 2.9219,
      "step": 1325
    },
    {
      "epoch": 0.8781456953642384,
      "grad_norm": 1.83029097083417,
      "learning_rate": 0.00019889999999999998,
      "loss": 2.9375,
      "step": 1326
    },
    {
      "epoch": 0.8788079470198675,
      "grad_norm": 1.8718981928000682,
      "learning_rate": 0.00019905,
      "loss": 2.6406,
      "step": 1327
    },
    {
      "epoch": 0.8794701986754967,
      "grad_norm": 2.0359175764126305,
      "learning_rate": 0.0001992,
      "loss": 2.8281,
      "step": 1328
    },
    {
      "epoch": 0.8801324503311259,
      "grad_norm": 1.9720091395108976,
      "learning_rate": 0.00019934999999999997,
      "loss": 3.0625,
      "step": 1329
    },
    {
      "epoch": 0.8807947019867549,
      "grad_norm": 1.7165135233602165,
      "learning_rate": 0.0001995,
      "loss": 3.0312,
      "step": 1330
    },
    {
      "epoch": 0.8814569536423841,
      "grad_norm": 1.7387206413647913,
      "learning_rate": 0.00019964999999999997,
      "loss": 3.0,
      "step": 1331
    },
    {
      "epoch": 0.8821192052980132,
      "grad_norm": 1.5945304515390302,
      "learning_rate": 0.0001998,
      "loss": 2.8906,
      "step": 1332
    },
    {
      "epoch": 0.8827814569536424,
      "grad_norm": 1.7365850409684334,
      "learning_rate": 0.00019994999999999998,
      "loss": 2.8281,
      "step": 1333
    },
    {
      "epoch": 0.8834437086092716,
      "grad_norm": 1.6997557620462878,
      "learning_rate": 0.00020009999999999998,
      "loss": 2.375,
      "step": 1334
    },
    {
      "epoch": 0.8841059602649006,
      "grad_norm": 1.5913269309673954,
      "learning_rate": 0.00020025,
      "loss": 2.9062,
      "step": 1335
    },
    {
      "epoch": 0.8847682119205298,
      "grad_norm": 1.7332359839626117,
      "learning_rate": 0.0002004,
      "loss": 2.9219,
      "step": 1336
    },
    {
      "epoch": 0.885430463576159,
      "grad_norm": 1.4041720781314866,
      "learning_rate": 0.00020054999999999997,
      "loss": 2.4062,
      "step": 1337
    },
    {
      "epoch": 0.8860927152317881,
      "grad_norm": 1.74773019119878,
      "learning_rate": 0.0002007,
      "loss": 3.0312,
      "step": 1338
    },
    {
      "epoch": 0.8867549668874172,
      "grad_norm": 1.474956182698185,
      "learning_rate": 0.00020084999999999998,
      "loss": 2.9531,
      "step": 1339
    },
    {
      "epoch": 0.8874172185430463,
      "grad_norm": 2.463410637891758,
      "learning_rate": 0.000201,
      "loss": 2.6094,
      "step": 1340
    },
    {
      "epoch": 0.8880794701986755,
      "grad_norm": 1.4774451046450296,
      "learning_rate": 0.00020114999999999998,
      "loss": 3.1094,
      "step": 1341
    },
    {
      "epoch": 0.8887417218543047,
      "grad_norm": 1.5346116839122577,
      "learning_rate": 0.0002013,
      "loss": 2.6875,
      "step": 1342
    },
    {
      "epoch": 0.8894039735099337,
      "grad_norm": 1.6829863526467448,
      "learning_rate": 0.00020145,
      "loss": 2.9688,
      "step": 1343
    },
    {
      "epoch": 0.8900662251655629,
      "grad_norm": 1.4938465070677558,
      "learning_rate": 0.0002016,
      "loss": 2.7188,
      "step": 1344
    },
    {
      "epoch": 0.890728476821192,
      "grad_norm": 1.3753626821988953,
      "learning_rate": 0.00020174999999999997,
      "loss": 2.2031,
      "step": 1345
    },
    {
      "epoch": 0.8913907284768212,
      "grad_norm": 1.449124975048414,
      "learning_rate": 0.0002019,
      "loss": 2.9531,
      "step": 1346
    },
    {
      "epoch": 0.8920529801324504,
      "grad_norm": 1.4164700142217563,
      "learning_rate": 0.00020204999999999998,
      "loss": 2.8438,
      "step": 1347
    },
    {
      "epoch": 0.8927152317880794,
      "grad_norm": 1.561906060182665,
      "learning_rate": 0.0002022,
      "loss": 2.7656,
      "step": 1348
    },
    {
      "epoch": 0.8933774834437086,
      "grad_norm": 1.50374427364562,
      "learning_rate": 0.00020234999999999999,
      "loss": 2.9219,
      "step": 1349
    },
    {
      "epoch": 0.8940397350993378,
      "grad_norm": 1.7678186446785271,
      "learning_rate": 0.0002025,
      "loss": 2.8281,
      "step": 1350
    },
    {
      "epoch": 0.8947019867549669,
      "grad_norm": 1.7658211908889372,
      "learning_rate": 0.00020264999999999997,
      "loss": 2.8125,
      "step": 1351
    },
    {
      "epoch": 0.895364238410596,
      "grad_norm": 1.4397884138710912,
      "learning_rate": 0.0002028,
      "loss": 2.75,
      "step": 1352
    },
    {
      "epoch": 0.8960264900662251,
      "grad_norm": 1.4548613283237661,
      "learning_rate": 0.00020294999999999997,
      "loss": 2.3594,
      "step": 1353
    },
    {
      "epoch": 0.8966887417218543,
      "grad_norm": 1.7180858602217157,
      "learning_rate": 0.0002031,
      "loss": 2.8125,
      "step": 1354
    },
    {
      "epoch": 0.8973509933774835,
      "grad_norm": 1.4663944084563982,
      "learning_rate": 0.00020324999999999998,
      "loss": 2.7344,
      "step": 1355
    },
    {
      "epoch": 0.8980132450331125,
      "grad_norm": 1.533038867656512,
      "learning_rate": 0.00020339999999999998,
      "loss": 2.9531,
      "step": 1356
    },
    {
      "epoch": 0.8986754966887417,
      "grad_norm": 1.620822625361471,
      "learning_rate": 0.00020355,
      "loss": 2.6094,
      "step": 1357
    },
    {
      "epoch": 0.8993377483443709,
      "grad_norm": 1.447237712739006,
      "learning_rate": 0.0002037,
      "loss": 2.7031,
      "step": 1358
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6105157965875507,
      "learning_rate": 0.00020384999999999997,
      "loss": 2.9531,
      "step": 1359
    },
    {
      "epoch": 0.9006622516556292,
      "grad_norm": 4.571638031607098,
      "learning_rate": 0.000204,
      "loss": 2.875,
      "step": 1360
    },
    {
      "epoch": 0.9013245033112582,
      "grad_norm": 1.7738790930088484,
      "learning_rate": 0.00020414999999999997,
      "loss": 2.6562,
      "step": 1361
    },
    {
      "epoch": 0.9019867549668874,
      "grad_norm": 1.5899087195644968,
      "learning_rate": 0.0002043,
      "loss": 2.7969,
      "step": 1362
    },
    {
      "epoch": 0.9026490066225166,
      "grad_norm": 1.5037662290863272,
      "learning_rate": 0.00020444999999999998,
      "loss": 2.8125,
      "step": 1363
    },
    {
      "epoch": 0.9033112582781457,
      "grad_norm": 1.5222018208359926,
      "learning_rate": 0.00020459999999999999,
      "loss": 2.3906,
      "step": 1364
    },
    {
      "epoch": 0.9039735099337748,
      "grad_norm": 1.6118353293337373,
      "learning_rate": 0.00020475,
      "loss": 2.8594,
      "step": 1365
    },
    {
      "epoch": 0.904635761589404,
      "grad_norm": 1.502161009638808,
      "learning_rate": 0.0002049,
      "loss": 2.3906,
      "step": 1366
    },
    {
      "epoch": 0.9052980132450331,
      "grad_norm": 1.5597003886186351,
      "learning_rate": 0.00020504999999999997,
      "loss": 2.8438,
      "step": 1367
    },
    {
      "epoch": 0.9059602649006623,
      "grad_norm": 1.5554827438380323,
      "learning_rate": 0.0002052,
      "loss": 2.2812,
      "step": 1368
    },
    {
      "epoch": 0.9066225165562913,
      "grad_norm": 1.5836107931876346,
      "learning_rate": 0.00020534999999999998,
      "loss": 2.8594,
      "step": 1369
    },
    {
      "epoch": 0.9072847682119205,
      "grad_norm": 1.78331122867097,
      "learning_rate": 0.0002055,
      "loss": 3.0625,
      "step": 1370
    },
    {
      "epoch": 0.9079470198675497,
      "grad_norm": 1.4242900341146445,
      "learning_rate": 0.00020564999999999998,
      "loss": 2.6875,
      "step": 1371
    },
    {
      "epoch": 0.9086092715231788,
      "grad_norm": 1.357918874171911,
      "learning_rate": 0.0002058,
      "loss": 2.8438,
      "step": 1372
    },
    {
      "epoch": 0.909271523178808,
      "grad_norm": 1.4618490258419004,
      "learning_rate": 0.00020595,
      "loss": 2.6719,
      "step": 1373
    },
    {
      "epoch": 0.909933774834437,
      "grad_norm": 1.5027818471955123,
      "learning_rate": 0.0002061,
      "loss": 3.1562,
      "step": 1374
    },
    {
      "epoch": 0.9105960264900662,
      "grad_norm": 1.4237550740618516,
      "learning_rate": 0.00020624999999999997,
      "loss": 2.8906,
      "step": 1375
    },
    {
      "epoch": 0.9112582781456954,
      "grad_norm": 1.3951903615686183,
      "learning_rate": 0.00020639999999999998,
      "loss": 2.8125,
      "step": 1376
    },
    {
      "epoch": 0.9119205298013245,
      "grad_norm": 1.5472104206974824,
      "learning_rate": 0.00020654999999999998,
      "loss": 2.9219,
      "step": 1377
    },
    {
      "epoch": 0.9125827814569536,
      "grad_norm": 1.393140708389783,
      "learning_rate": 0.00020669999999999996,
      "loss": 2.9219,
      "step": 1378
    },
    {
      "epoch": 0.9132450331125828,
      "grad_norm": 1.7274961045766346,
      "learning_rate": 0.00020684999999999999,
      "loss": 2.9531,
      "step": 1379
    },
    {
      "epoch": 0.9139072847682119,
      "grad_norm": 1.4538649519869518,
      "learning_rate": 0.00020699999999999996,
      "loss": 2.75,
      "step": 1380
    },
    {
      "epoch": 0.9145695364238411,
      "grad_norm": 1.279091826306969,
      "learning_rate": 0.00020715,
      "loss": 2.7656,
      "step": 1381
    },
    {
      "epoch": 0.9152317880794701,
      "grad_norm": 1.5609246305548645,
      "learning_rate": 0.00020729999999999997,
      "loss": 3.0312,
      "step": 1382
    },
    {
      "epoch": 0.9158940397350993,
      "grad_norm": 1.538887972920399,
      "learning_rate": 0.00020744999999999997,
      "loss": 2.7188,
      "step": 1383
    },
    {
      "epoch": 0.9165562913907285,
      "grad_norm": 1.4381341232781528,
      "learning_rate": 0.00020759999999999998,
      "loss": 2.875,
      "step": 1384
    },
    {
      "epoch": 0.9172185430463576,
      "grad_norm": 1.369532596945704,
      "learning_rate": 0.00020774999999999998,
      "loss": 2.6562,
      "step": 1385
    },
    {
      "epoch": 0.9178807947019868,
      "grad_norm": 1.5035880126747518,
      "learning_rate": 0.00020789999999999996,
      "loss": 3.0938,
      "step": 1386
    },
    {
      "epoch": 0.9185430463576159,
      "grad_norm": 1.7210000314337632,
      "learning_rate": 0.00020805,
      "loss": 2.6094,
      "step": 1387
    },
    {
      "epoch": 0.919205298013245,
      "grad_norm": 1.499158745828499,
      "learning_rate": 0.00020819999999999996,
      "loss": 2.75,
      "step": 1388
    },
    {
      "epoch": 0.9198675496688742,
      "grad_norm": 1.5287500508620049,
      "learning_rate": 0.00020835,
      "loss": 2.8906,
      "step": 1389
    },
    {
      "epoch": 0.9205298013245033,
      "grad_norm": 1.3536655453120798,
      "learning_rate": 0.00020849999999999997,
      "loss": 2.4219,
      "step": 1390
    },
    {
      "epoch": 0.9211920529801324,
      "grad_norm": 1.6649626616424975,
      "learning_rate": 0.00020864999999999998,
      "loss": 3.0,
      "step": 1391
    },
    {
      "epoch": 0.9218543046357616,
      "grad_norm": 1.3470106061149882,
      "learning_rate": 0.00020879999999999998,
      "loss": 2.875,
      "step": 1392
    },
    {
      "epoch": 0.9225165562913907,
      "grad_norm": 1.3212770437782715,
      "learning_rate": 0.00020894999999999998,
      "loss": 2.75,
      "step": 1393
    },
    {
      "epoch": 0.9231788079470199,
      "grad_norm": 1.463062616171229,
      "learning_rate": 0.00020909999999999996,
      "loss": 2.375,
      "step": 1394
    },
    {
      "epoch": 0.9238410596026491,
      "grad_norm": 1.4628881973612466,
      "learning_rate": 0.00020925,
      "loss": 2.8125,
      "step": 1395
    },
    {
      "epoch": 0.9245033112582781,
      "grad_norm": 1.5518256163080537,
      "learning_rate": 0.00020939999999999997,
      "loss": 2.8438,
      "step": 1396
    },
    {
      "epoch": 0.9251655629139073,
      "grad_norm": 1.390316201113342,
      "learning_rate": 0.00020955,
      "loss": 2.8438,
      "step": 1397
    },
    {
      "epoch": 0.9258278145695364,
      "grad_norm": 1.3945650958185598,
      "learning_rate": 0.00020969999999999997,
      "loss": 2.7812,
      "step": 1398
    },
    {
      "epoch": 0.9264900662251656,
      "grad_norm": 2.5940424767660013,
      "learning_rate": 0.00020984999999999998,
      "loss": 2.875,
      "step": 1399
    },
    {
      "epoch": 0.9271523178807947,
      "grad_norm": 1.4273371665104149,
      "learning_rate": 0.00020999999999999998,
      "loss": 2.6562,
      "step": 1400
    },
    {
      "epoch": 0.9278145695364238,
      "grad_norm": 1.470668390919937,
      "learning_rate": 0.00021014999999999999,
      "loss": 2.8125,
      "step": 1401
    },
    {
      "epoch": 0.928476821192053,
      "grad_norm": 2.449382176929562,
      "learning_rate": 0.00021029999999999996,
      "loss": 2.4531,
      "step": 1402
    },
    {
      "epoch": 0.9291390728476822,
      "grad_norm": 1.4482714390262088,
      "learning_rate": 0.00021045,
      "loss": 2.9688,
      "step": 1403
    },
    {
      "epoch": 0.9298013245033112,
      "grad_norm": 1.5969689405241236,
      "learning_rate": 0.00021059999999999997,
      "loss": 2.7969,
      "step": 1404
    },
    {
      "epoch": 0.9304635761589404,
      "grad_norm": 1.48817199692799,
      "learning_rate": 0.00021074999999999997,
      "loss": 2.8594,
      "step": 1405
    },
    {
      "epoch": 0.9311258278145695,
      "grad_norm": 1.5629706737677924,
      "learning_rate": 0.00021089999999999998,
      "loss": 2.8281,
      "step": 1406
    },
    {
      "epoch": 0.9317880794701987,
      "grad_norm": 1.519313146047688,
      "learning_rate": 0.00021104999999999998,
      "loss": 2.875,
      "step": 1407
    },
    {
      "epoch": 0.9324503311258279,
      "grad_norm": 1.2564930413360058,
      "learning_rate": 0.00021119999999999996,
      "loss": 2.6562,
      "step": 1408
    },
    {
      "epoch": 0.9331125827814569,
      "grad_norm": 1.2642865821304912,
      "learning_rate": 0.00021135,
      "loss": 2.5625,
      "step": 1409
    },
    {
      "epoch": 0.9337748344370861,
      "grad_norm": 1.2435791296096543,
      "learning_rate": 0.00021149999999999996,
      "loss": 2.7031,
      "step": 1410
    },
    {
      "epoch": 0.9344370860927153,
      "grad_norm": 1.4854880780755986,
      "learning_rate": 0.00021165,
      "loss": 3.0469,
      "step": 1411
    },
    {
      "epoch": 0.9350993377483444,
      "grad_norm": 1.378053261340132,
      "learning_rate": 0.00021179999999999997,
      "loss": 2.75,
      "step": 1412
    },
    {
      "epoch": 0.9357615894039735,
      "grad_norm": 1.4096308458980293,
      "learning_rate": 0.00021194999999999997,
      "loss": 3.0156,
      "step": 1413
    },
    {
      "epoch": 0.9364238410596026,
      "grad_norm": 1.6356674337850206,
      "learning_rate": 0.00021209999999999998,
      "loss": 2.7344,
      "step": 1414
    },
    {
      "epoch": 0.9370860927152318,
      "grad_norm": 1.5192142930479984,
      "learning_rate": 0.00021224999999999998,
      "loss": 2.75,
      "step": 1415
    },
    {
      "epoch": 0.937748344370861,
      "grad_norm": 1.501254080021138,
      "learning_rate": 0.00021239999999999996,
      "loss": 2.8906,
      "step": 1416
    },
    {
      "epoch": 0.93841059602649,
      "grad_norm": 1.394566827391653,
      "learning_rate": 0.00021255,
      "loss": 2.8125,
      "step": 1417
    },
    {
      "epoch": 0.9390728476821192,
      "grad_norm": 1.316154987913069,
      "learning_rate": 0.00021269999999999997,
      "loss": 2.4531,
      "step": 1418
    },
    {
      "epoch": 0.9397350993377483,
      "grad_norm": 1.270056255112331,
      "learning_rate": 0.00021285,
      "loss": 2.5781,
      "step": 1419
    },
    {
      "epoch": 0.9403973509933775,
      "grad_norm": 1.5399642317981366,
      "learning_rate": 0.00021299999999999997,
      "loss": 2.8281,
      "step": 1420
    },
    {
      "epoch": 0.9410596026490067,
      "grad_norm": 1.2269499858746964,
      "learning_rate": 0.00021314999999999998,
      "loss": 2.7344,
      "step": 1421
    },
    {
      "epoch": 0.9417218543046357,
      "grad_norm": 1.3888560618557948,
      "learning_rate": 0.00021329999999999998,
      "loss": 2.8594,
      "step": 1422
    },
    {
      "epoch": 0.9423841059602649,
      "grad_norm": 1.3530706227235787,
      "learning_rate": 0.00021344999999999998,
      "loss": 2.6406,
      "step": 1423
    },
    {
      "epoch": 0.9430463576158941,
      "grad_norm": 1.3544176318893908,
      "learning_rate": 0.00021359999999999996,
      "loss": 2.625,
      "step": 1424
    },
    {
      "epoch": 0.9437086092715232,
      "grad_norm": 1.4032200727843989,
      "learning_rate": 0.00021375,
      "loss": 2.7969,
      "step": 1425
    },
    {
      "epoch": 0.9443708609271523,
      "grad_norm": 1.4600841457367302,
      "learning_rate": 0.00021389999999999997,
      "loss": 2.6875,
      "step": 1426
    },
    {
      "epoch": 0.9450331125827814,
      "grad_norm": 1.3333425334918014,
      "learning_rate": 0.00021405,
      "loss": 2.8281,
      "step": 1427
    },
    {
      "epoch": 0.9456953642384106,
      "grad_norm": 1.5317319637065545,
      "learning_rate": 0.00021419999999999998,
      "loss": 2.7656,
      "step": 1428
    },
    {
      "epoch": 0.9463576158940398,
      "grad_norm": 1.3257516968812817,
      "learning_rate": 0.00021434999999999998,
      "loss": 2.7812,
      "step": 1429
    },
    {
      "epoch": 0.9470198675496688,
      "grad_norm": 4.545829716436066,
      "learning_rate": 0.00021449999999999998,
      "loss": 2.6875,
      "step": 1430
    },
    {
      "epoch": 0.947682119205298,
      "grad_norm": 1.3742459724148652,
      "learning_rate": 0.00021464999999999999,
      "loss": 2.7031,
      "step": 1431
    },
    {
      "epoch": 0.9483443708609272,
      "grad_norm": 1.5111068358181001,
      "learning_rate": 0.00021479999999999996,
      "loss": 2.8281,
      "step": 1432
    },
    {
      "epoch": 0.9490066225165563,
      "grad_norm": 1.3401822541097885,
      "learning_rate": 0.00021495,
      "loss": 2.9688,
      "step": 1433
    },
    {
      "epoch": 0.9496688741721855,
      "grad_norm": 1.2808650841947398,
      "learning_rate": 0.00021509999999999997,
      "loss": 2.7031,
      "step": 1434
    },
    {
      "epoch": 0.9503311258278145,
      "grad_norm": 1.393405026829189,
      "learning_rate": 0.00021525,
      "loss": 2.625,
      "step": 1435
    },
    {
      "epoch": 0.9509933774834437,
      "grad_norm": 1.3190900170936124,
      "learning_rate": 0.00021539999999999998,
      "loss": 2.2031,
      "step": 1436
    },
    {
      "epoch": 0.9516556291390729,
      "grad_norm": 1.4694674081498431,
      "learning_rate": 0.00021554999999999998,
      "loss": 2.9219,
      "step": 1437
    },
    {
      "epoch": 0.952317880794702,
      "grad_norm": 1.716467710084653,
      "learning_rate": 0.00021569999999999998,
      "loss": 2.6875,
      "step": 1438
    },
    {
      "epoch": 0.9529801324503311,
      "grad_norm": 1.3657841980107903,
      "learning_rate": 0.00021585,
      "loss": 2.6562,
      "step": 1439
    },
    {
      "epoch": 0.9536423841059603,
      "grad_norm": 1.3364184085402722,
      "learning_rate": 0.00021599999999999996,
      "loss": 2.75,
      "step": 1440
    },
    {
      "epoch": 0.9543046357615894,
      "grad_norm": 1.5110590546223863,
      "learning_rate": 0.00021615,
      "loss": 2.9844,
      "step": 1441
    },
    {
      "epoch": 0.9549668874172186,
      "grad_norm": 1.4377294927026996,
      "learning_rate": 0.00021629999999999997,
      "loss": 2.9375,
      "step": 1442
    },
    {
      "epoch": 0.9556291390728476,
      "grad_norm": 1.294891873077855,
      "learning_rate": 0.00021645,
      "loss": 2.4844,
      "step": 1443
    },
    {
      "epoch": 0.9562913907284768,
      "grad_norm": 1.317993300421824,
      "learning_rate": 0.00021659999999999998,
      "loss": 2.6406,
      "step": 1444
    },
    {
      "epoch": 0.956953642384106,
      "grad_norm": 1.3512037382647744,
      "learning_rate": 0.00021674999999999998,
      "loss": 2.8281,
      "step": 1445
    },
    {
      "epoch": 0.9576158940397351,
      "grad_norm": 1.3238621982766001,
      "learning_rate": 0.0002169,
      "loss": 2.8438,
      "step": 1446
    },
    {
      "epoch": 0.9582781456953643,
      "grad_norm": 1.4602664147481539,
      "learning_rate": 0.00021705,
      "loss": 3.0,
      "step": 1447
    },
    {
      "epoch": 0.9589403973509933,
      "grad_norm": 1.3174064478315253,
      "learning_rate": 0.00021719999999999997,
      "loss": 2.7969,
      "step": 1448
    },
    {
      "epoch": 0.9596026490066225,
      "grad_norm": 1.312557581970755,
      "learning_rate": 0.00021735,
      "loss": 2.7188,
      "step": 1449
    },
    {
      "epoch": 0.9602649006622517,
      "grad_norm": 1.2076507764658608,
      "learning_rate": 0.00021749999999999997,
      "loss": 2.5312,
      "step": 1450
    },
    {
      "epoch": 0.9609271523178808,
      "grad_norm": 1.3350720202564774,
      "learning_rate": 0.00021764999999999998,
      "loss": 2.8438,
      "step": 1451
    },
    {
      "epoch": 0.9615894039735099,
      "grad_norm": 1.3047772135309976,
      "learning_rate": 0.00021779999999999998,
      "loss": 2.4844,
      "step": 1452
    },
    {
      "epoch": 0.9622516556291391,
      "grad_norm": 1.352514392344182,
      "learning_rate": 0.00021794999999999999,
      "loss": 2.8438,
      "step": 1453
    },
    {
      "epoch": 0.9629139072847682,
      "grad_norm": 1.288692855024458,
      "learning_rate": 0.00021809999999999996,
      "loss": 2.4062,
      "step": 1454
    },
    {
      "epoch": 0.9635761589403974,
      "grad_norm": 1.324059219810981,
      "learning_rate": 0.00021825,
      "loss": 2.6875,
      "step": 1455
    },
    {
      "epoch": 0.9642384105960264,
      "grad_norm": 1.3247575800966827,
      "learning_rate": 0.00021839999999999997,
      "loss": 2.8438,
      "step": 1456
    },
    {
      "epoch": 0.9649006622516556,
      "grad_norm": 1.3968875007793182,
      "learning_rate": 0.00021855,
      "loss": 2.8906,
      "step": 1457
    },
    {
      "epoch": 0.9655629139072848,
      "grad_norm": 1.271406317706869,
      "learning_rate": 0.00021869999999999998,
      "loss": 2.5781,
      "step": 1458
    },
    {
      "epoch": 0.9662251655629139,
      "grad_norm": 1.27000537797962,
      "learning_rate": 0.00021884999999999998,
      "loss": 2.6094,
      "step": 1459
    },
    {
      "epoch": 0.9668874172185431,
      "grad_norm": 1.2720938100691048,
      "learning_rate": 0.00021899999999999998,
      "loss": 2.7656,
      "step": 1460
    },
    {
      "epoch": 0.9675496688741722,
      "grad_norm": 1.4034429717771355,
      "learning_rate": 0.00021915,
      "loss": 2.8281,
      "step": 1461
    },
    {
      "epoch": 0.9682119205298013,
      "grad_norm": 1.281269090662935,
      "learning_rate": 0.00021929999999999996,
      "loss": 2.4375,
      "step": 1462
    },
    {
      "epoch": 0.9688741721854305,
      "grad_norm": 1.3428427551068691,
      "learning_rate": 0.00021945,
      "loss": 2.6562,
      "step": 1463
    },
    {
      "epoch": 0.9695364238410596,
      "grad_norm": 1.3036853360893546,
      "learning_rate": 0.00021959999999999997,
      "loss": 2.6094,
      "step": 1464
    },
    {
      "epoch": 0.9701986754966887,
      "grad_norm": 1.3470144851290966,
      "learning_rate": 0.00021975,
      "loss": 2.8281,
      "step": 1465
    },
    {
      "epoch": 0.9708609271523179,
      "grad_norm": 1.4369795624670356,
      "learning_rate": 0.00021989999999999998,
      "loss": 2.6875,
      "step": 1466
    },
    {
      "epoch": 0.971523178807947,
      "grad_norm": 1.4456095040656771,
      "learning_rate": 0.00022004999999999998,
      "loss": 2.875,
      "step": 1467
    },
    {
      "epoch": 0.9721854304635762,
      "grad_norm": 1.4529721352808063,
      "learning_rate": 0.00022019999999999999,
      "loss": 2.7344,
      "step": 1468
    },
    {
      "epoch": 0.9728476821192052,
      "grad_norm": 1.254798333969733,
      "learning_rate": 0.00022035,
      "loss": 2.7656,
      "step": 1469
    },
    {
      "epoch": 0.9735099337748344,
      "grad_norm": 1.507387495013173,
      "learning_rate": 0.00022049999999999997,
      "loss": 2.6562,
      "step": 1470
    },
    {
      "epoch": 0.9741721854304636,
      "grad_norm": 1.478097855643656,
      "learning_rate": 0.00022065,
      "loss": 2.9531,
      "step": 1471
    },
    {
      "epoch": 0.9748344370860927,
      "grad_norm": 1.3885810640699483,
      "learning_rate": 0.00022079999999999997,
      "loss": 2.7031,
      "step": 1472
    },
    {
      "epoch": 0.9754966887417219,
      "grad_norm": 1.3301416729862565,
      "learning_rate": 0.00022095,
      "loss": 2.6406,
      "step": 1473
    },
    {
      "epoch": 0.976158940397351,
      "grad_norm": 1.3593749113187064,
      "learning_rate": 0.00022109999999999998,
      "loss": 2.2656,
      "step": 1474
    },
    {
      "epoch": 0.9768211920529801,
      "grad_norm": 1.3769146432485133,
      "learning_rate": 0.00022124999999999998,
      "loss": 2.5781,
      "step": 1475
    },
    {
      "epoch": 0.9774834437086093,
      "grad_norm": 1.369152044166675,
      "learning_rate": 0.0002214,
      "loss": 2.6406,
      "step": 1476
    },
    {
      "epoch": 0.9781456953642385,
      "grad_norm": 1.6658279351130625,
      "learning_rate": 0.00022155,
      "loss": 2.8594,
      "step": 1477
    },
    {
      "epoch": 0.9788079470198675,
      "grad_norm": 1.39317837183439,
      "learning_rate": 0.00022169999999999997,
      "loss": 2.9531,
      "step": 1478
    },
    {
      "epoch": 0.9794701986754967,
      "grad_norm": 1.2817190345444938,
      "learning_rate": 0.00022185,
      "loss": 2.7031,
      "step": 1479
    },
    {
      "epoch": 0.9801324503311258,
      "grad_norm": 1.3399915666046203,
      "learning_rate": 0.00022199999999999998,
      "loss": 2.7812,
      "step": 1480
    },
    {
      "epoch": 0.980794701986755,
      "grad_norm": 1.3255547633109914,
      "learning_rate": 0.00022215,
      "loss": 2.4688,
      "step": 1481
    },
    {
      "epoch": 0.9814569536423841,
      "grad_norm": 1.398686569630114,
      "learning_rate": 0.00022229999999999998,
      "loss": 2.7344,
      "step": 1482
    },
    {
      "epoch": 0.9821192052980132,
      "grad_norm": 1.4177146429090006,
      "learning_rate": 0.00022244999999999999,
      "loss": 2.7812,
      "step": 1483
    },
    {
      "epoch": 0.9827814569536424,
      "grad_norm": 1.2256578714309039,
      "learning_rate": 0.0002226,
      "loss": 2.5938,
      "step": 1484
    },
    {
      "epoch": 0.9834437086092715,
      "grad_norm": 1.3343181263324553,
      "learning_rate": 0.00022275,
      "loss": 2.6406,
      "step": 1485
    },
    {
      "epoch": 0.9841059602649007,
      "grad_norm": 1.449724238441013,
      "learning_rate": 0.00022289999999999997,
      "loss": 2.9688,
      "step": 1486
    },
    {
      "epoch": 0.9847682119205298,
      "grad_norm": 1.1672030860650229,
      "learning_rate": 0.00022305,
      "loss": 2.4219,
      "step": 1487
    },
    {
      "epoch": 0.9854304635761589,
      "grad_norm": 1.4158786412734308,
      "learning_rate": 0.00022319999999999998,
      "loss": 2.7344,
      "step": 1488
    },
    {
      "epoch": 0.9860927152317881,
      "grad_norm": 1.3640840389080837,
      "learning_rate": 0.00022335,
      "loss": 2.9531,
      "step": 1489
    },
    {
      "epoch": 0.9867549668874173,
      "grad_norm": 1.296678057042255,
      "learning_rate": 0.00022349999999999998,
      "loss": 2.8438,
      "step": 1490
    },
    {
      "epoch": 0.9874172185430463,
      "grad_norm": 1.3785105930686457,
      "learning_rate": 0.00022365,
      "loss": 2.1719,
      "step": 1491
    },
    {
      "epoch": 0.9880794701986755,
      "grad_norm": 1.336866249430873,
      "learning_rate": 0.0002238,
      "loss": 2.5469,
      "step": 1492
    },
    {
      "epoch": 0.9887417218543046,
      "grad_norm": 1.2826570369271129,
      "learning_rate": 0.00022395,
      "loss": 2.5469,
      "step": 1493
    },
    {
      "epoch": 0.9894039735099338,
      "grad_norm": 1.368318593098838,
      "learning_rate": 0.00022409999999999997,
      "loss": 2.7031,
      "step": 1494
    },
    {
      "epoch": 0.9900662251655629,
      "grad_norm": 1.3391828909053034,
      "learning_rate": 0.00022425,
      "loss": 2.625,
      "step": 1495
    },
    {
      "epoch": 0.990728476821192,
      "grad_norm": 1.358492861399515,
      "learning_rate": 0.00022439999999999998,
      "loss": 2.5469,
      "step": 1496
    },
    {
      "epoch": 0.9913907284768212,
      "grad_norm": 1.3565271494003894,
      "learning_rate": 0.00022455,
      "loss": 2.8281,
      "step": 1497
    },
    {
      "epoch": 0.9920529801324504,
      "grad_norm": 1.1957861482563874,
      "learning_rate": 0.0002247,
      "loss": 2.1406,
      "step": 1498
    },
    {
      "epoch": 0.9927152317880795,
      "grad_norm": 1.3805488818206735,
      "learning_rate": 0.00022485,
      "loss": 2.7344,
      "step": 1499
    },
    {
      "epoch": 0.9933774834437086,
      "grad_norm": 1.4317760344078536,
      "learning_rate": 0.000225,
      "loss": 2.6562,
      "step": 1500
    },
    {
      "epoch": 0.9940397350993377,
      "grad_norm": 1.3471406939170751,
      "learning_rate": 0.00022514999999999997,
      "loss": 2.8125,
      "step": 1501
    },
    {
      "epoch": 0.9947019867549669,
      "grad_norm": 1.270885753278136,
      "learning_rate": 0.00022529999999999997,
      "loss": 2.7344,
      "step": 1502
    },
    {
      "epoch": 0.9953642384105961,
      "grad_norm": 1.5625251777065006,
      "learning_rate": 0.00022544999999999995,
      "loss": 2.7188,
      "step": 1503
    },
    {
      "epoch": 0.9960264900662251,
      "grad_norm": 1.4056014700380517,
      "learning_rate": 0.00022559999999999998,
      "loss": 2.7031,
      "step": 1504
    },
    {
      "epoch": 0.9966887417218543,
      "grad_norm": 1.4440078609057376,
      "learning_rate": 0.00022574999999999996,
      "loss": 2.8906,
      "step": 1505
    },
    {
      "epoch": 0.9973509933774835,
      "grad_norm": 1.4636963433470624,
      "learning_rate": 0.0002259,
      "loss": 3.0156,
      "step": 1506
    },
    {
      "epoch": 0.9980132450331126,
      "grad_norm": 1.4122316232867143,
      "learning_rate": 0.00022604999999999997,
      "loss": 2.6562,
      "step": 1507
    },
    {
      "epoch": 0.9986754966887417,
      "grad_norm": 1.1385473039829144,
      "learning_rate": 0.00022619999999999997,
      "loss": 2.5,
      "step": 1508
    },
    {
      "epoch": 0.9993377483443708,
      "grad_norm": 1.3077440560589277,
      "learning_rate": 0.00022634999999999997,
      "loss": 2.6562,
      "step": 1509
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2105287244431298,
      "learning_rate": 0.00022649999999999998,
      "loss": 2.7969,
      "step": 1510
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.625924587249756,
      "eval_runtime": 34.0853,
      "eval_samples_per_second": 9.916,
      "eval_steps_per_second": 9.916,
      "step": 1510
    },
    {
      "epoch": 1.0006622516556292,
      "grad_norm": 1.2584246200918825,
      "learning_rate": 0.00022664999999999995,
      "loss": 2.6406,
      "step": 1511
    },
    {
      "epoch": 1.0013245033112583,
      "grad_norm": 1.4771743635453922,
      "learning_rate": 0.00022679999999999998,
      "loss": 2.7969,
      "step": 1512
    },
    {
      "epoch": 1.0019867549668875,
      "grad_norm": 1.4118511689346829,
      "learning_rate": 0.00022694999999999996,
      "loss": 2.5625,
      "step": 1513
    },
    {
      "epoch": 1.0026490066225167,
      "grad_norm": 1.2780776327953083,
      "learning_rate": 0.0002271,
      "loss": 2.6875,
      "step": 1514
    },
    {
      "epoch": 1.0033112582781456,
      "grad_norm": 1.2600893150163668,
      "learning_rate": 0.00022724999999999997,
      "loss": 2.7031,
      "step": 1515
    },
    {
      "epoch": 1.0039735099337748,
      "grad_norm": 1.2340528109906057,
      "learning_rate": 0.00022739999999999997,
      "loss": 2.6562,
      "step": 1516
    },
    {
      "epoch": 1.004635761589404,
      "grad_norm": 1.4209038924647308,
      "learning_rate": 0.00022754999999999997,
      "loss": 2.7344,
      "step": 1517
    },
    {
      "epoch": 1.005298013245033,
      "grad_norm": 1.2009470774561128,
      "learning_rate": 0.00022769999999999998,
      "loss": 2.9062,
      "step": 1518
    },
    {
      "epoch": 1.0059602649006623,
      "grad_norm": 1.4607176197867633,
      "learning_rate": 0.00022784999999999995,
      "loss": 2.5625,
      "step": 1519
    },
    {
      "epoch": 1.0066225165562914,
      "grad_norm": 1.329691605058324,
      "learning_rate": 0.00022799999999999999,
      "loss": 2.6719,
      "step": 1520
    },
    {
      "epoch": 1.0072847682119206,
      "grad_norm": 1.2032114553547824,
      "learning_rate": 0.00022814999999999996,
      "loss": 2.5156,
      "step": 1521
    },
    {
      "epoch": 1.0079470198675498,
      "grad_norm": 1.3572314784316437,
      "learning_rate": 0.0002283,
      "loss": 2.7188,
      "step": 1522
    },
    {
      "epoch": 1.008609271523179,
      "grad_norm": 1.2772922516951868,
      "learning_rate": 0.00022844999999999997,
      "loss": 2.7344,
      "step": 1523
    },
    {
      "epoch": 1.0092715231788079,
      "grad_norm": 1.329747253519728,
      "learning_rate": 0.00022859999999999997,
      "loss": 2.6875,
      "step": 1524
    },
    {
      "epoch": 1.009933774834437,
      "grad_norm": 1.2757010985116841,
      "learning_rate": 0.00022874999999999998,
      "loss": 2.7031,
      "step": 1525
    },
    {
      "epoch": 1.0105960264900662,
      "grad_norm": 1.2903175490841643,
      "learning_rate": 0.00022889999999999998,
      "loss": 2.7188,
      "step": 1526
    },
    {
      "epoch": 1.0112582781456954,
      "grad_norm": 1.2892690991267763,
      "learning_rate": 0.00022904999999999996,
      "loss": 2.9375,
      "step": 1527
    },
    {
      "epoch": 1.0119205298013245,
      "grad_norm": 1.3138798604299393,
      "learning_rate": 0.0002292,
      "loss": 2.9219,
      "step": 1528
    },
    {
      "epoch": 1.0125827814569537,
      "grad_norm": 1.2873922465196805,
      "learning_rate": 0.00022934999999999996,
      "loss": 2.2188,
      "step": 1529
    },
    {
      "epoch": 1.0132450331125828,
      "grad_norm": 1.3068908586498,
      "learning_rate": 0.0002295,
      "loss": 2.7344,
      "step": 1530
    },
    {
      "epoch": 1.013907284768212,
      "grad_norm": 1.2519276550832668,
      "learning_rate": 0.00022964999999999997,
      "loss": 2.6875,
      "step": 1531
    },
    {
      "epoch": 1.014569536423841,
      "grad_norm": 1.5501490897814858,
      "learning_rate": 0.00022979999999999997,
      "loss": 2.9062,
      "step": 1532
    },
    {
      "epoch": 1.0152317880794701,
      "grad_norm": 1.169911303649165,
      "learning_rate": 0.00022994999999999998,
      "loss": 2.1875,
      "step": 1533
    },
    {
      "epoch": 1.0158940397350993,
      "grad_norm": 1.1796369624920715,
      "learning_rate": 0.00023009999999999998,
      "loss": 2.6406,
      "step": 1534
    },
    {
      "epoch": 1.0165562913907285,
      "grad_norm": 1.317158553982051,
      "learning_rate": 0.00023024999999999996,
      "loss": 2.8125,
      "step": 1535
    },
    {
      "epoch": 1.0172185430463576,
      "grad_norm": 1.4408749064344644,
      "learning_rate": 0.0002304,
      "loss": 2.7344,
      "step": 1536
    },
    {
      "epoch": 1.0178807947019868,
      "grad_norm": 1.2105410731807122,
      "learning_rate": 0.00023054999999999997,
      "loss": 2.2812,
      "step": 1537
    },
    {
      "epoch": 1.018543046357616,
      "grad_norm": 1.3307348657727653,
      "learning_rate": 0.0002307,
      "loss": 2.6094,
      "step": 1538
    },
    {
      "epoch": 1.019205298013245,
      "grad_norm": 1.4312476891927959,
      "learning_rate": 0.00023084999999999997,
      "loss": 2.1719,
      "step": 1539
    },
    {
      "epoch": 1.0198675496688743,
      "grad_norm": 1.3398737919756114,
      "learning_rate": 0.00023099999999999998,
      "loss": 2.6094,
      "step": 1540
    },
    {
      "epoch": 1.0205298013245032,
      "grad_norm": 1.441897175374856,
      "learning_rate": 0.00023114999999999998,
      "loss": 2.7031,
      "step": 1541
    },
    {
      "epoch": 1.0211920529801324,
      "grad_norm": 1.3154895184656525,
      "learning_rate": 0.00023129999999999998,
      "loss": 2.6406,
      "step": 1542
    },
    {
      "epoch": 1.0218543046357615,
      "grad_norm": 1.0524561937157482,
      "learning_rate": 0.00023144999999999996,
      "loss": 2.0312,
      "step": 1543
    },
    {
      "epoch": 1.0225165562913907,
      "grad_norm": 1.3752767501037098,
      "learning_rate": 0.0002316,
      "loss": 2.5312,
      "step": 1544
    },
    {
      "epoch": 1.0231788079470199,
      "grad_norm": 1.4277261508161245,
      "learning_rate": 0.00023174999999999997,
      "loss": 2.8438,
      "step": 1545
    },
    {
      "epoch": 1.023841059602649,
      "grad_norm": 1.2234578987637423,
      "learning_rate": 0.0002319,
      "loss": 2.3125,
      "step": 1546
    },
    {
      "epoch": 1.0245033112582782,
      "grad_norm": 1.2494664433120775,
      "learning_rate": 0.00023204999999999998,
      "loss": 2.5469,
      "step": 1547
    },
    {
      "epoch": 1.0251655629139074,
      "grad_norm": 1.1840993924120304,
      "learning_rate": 0.00023219999999999998,
      "loss": 2.4219,
      "step": 1548
    },
    {
      "epoch": 1.0258278145695363,
      "grad_norm": 1.1999999896759703,
      "learning_rate": 0.00023234999999999998,
      "loss": 2.4531,
      "step": 1549
    },
    {
      "epoch": 1.0264900662251655,
      "grad_norm": 1.1472696922454548,
      "learning_rate": 0.00023249999999999999,
      "loss": 2.7656,
      "step": 1550
    },
    {
      "epoch": 1.0271523178807946,
      "grad_norm": 1.2133862408287155,
      "learning_rate": 0.00023264999999999996,
      "loss": 2.4688,
      "step": 1551
    },
    {
      "epoch": 1.0278145695364238,
      "grad_norm": 1.2015173124294138,
      "learning_rate": 0.0002328,
      "loss": 2.7031,
      "step": 1552
    },
    {
      "epoch": 1.028476821192053,
      "grad_norm": 1.1870154402479058,
      "learning_rate": 0.00023294999999999997,
      "loss": 2.5312,
      "step": 1553
    },
    {
      "epoch": 1.0291390728476821,
      "grad_norm": 1.2392996326108534,
      "learning_rate": 0.00023309999999999997,
      "loss": 2.6094,
      "step": 1554
    },
    {
      "epoch": 1.0298013245033113,
      "grad_norm": 1.1190456785007425,
      "learning_rate": 0.00023324999999999998,
      "loss": 2.6094,
      "step": 1555
    },
    {
      "epoch": 1.0304635761589405,
      "grad_norm": 1.2978419302281687,
      "learning_rate": 0.00023339999999999998,
      "loss": 2.5781,
      "step": 1556
    },
    {
      "epoch": 1.0311258278145696,
      "grad_norm": 1.3405730590307243,
      "learning_rate": 0.00023354999999999996,
      "loss": 2.7188,
      "step": 1557
    },
    {
      "epoch": 1.0317880794701986,
      "grad_norm": 1.3161156613368452,
      "learning_rate": 0.0002337,
      "loss": 2.7188,
      "step": 1558
    },
    {
      "epoch": 1.0324503311258277,
      "grad_norm": 1.289421967288709,
      "learning_rate": 0.00023384999999999997,
      "loss": 2.7656,
      "step": 1559
    },
    {
      "epoch": 1.033112582781457,
      "grad_norm": 1.1236335256311452,
      "learning_rate": 0.000234,
      "loss": 2.4531,
      "step": 1560
    },
    {
      "epoch": 1.033774834437086,
      "grad_norm": 1.1581344660593105,
      "learning_rate": 0.00023414999999999997,
      "loss": 2.7969,
      "step": 1561
    },
    {
      "epoch": 1.0344370860927152,
      "grad_norm": 1.1788368884894007,
      "learning_rate": 0.00023429999999999998,
      "loss": 2.6562,
      "step": 1562
    },
    {
      "epoch": 1.0350993377483444,
      "grad_norm": 1.260191740547071,
      "learning_rate": 0.00023444999999999998,
      "loss": 2.6875,
      "step": 1563
    },
    {
      "epoch": 1.0357615894039736,
      "grad_norm": 1.2217913822762283,
      "learning_rate": 0.00023459999999999998,
      "loss": 2.8125,
      "step": 1564
    },
    {
      "epoch": 1.0364238410596027,
      "grad_norm": 1.2824185359919227,
      "learning_rate": 0.00023474999999999996,
      "loss": 2.4844,
      "step": 1565
    },
    {
      "epoch": 1.0370860927152319,
      "grad_norm": 1.096439129107709,
      "learning_rate": 0.0002349,
      "loss": 2.5312,
      "step": 1566
    },
    {
      "epoch": 1.0377483443708608,
      "grad_norm": 1.3737963039997907,
      "learning_rate": 0.00023504999999999997,
      "loss": 2.6562,
      "step": 1567
    },
    {
      "epoch": 1.03841059602649,
      "grad_norm": 1.3094219829606462,
      "learning_rate": 0.0002352,
      "loss": 2.5469,
      "step": 1568
    },
    {
      "epoch": 1.0390728476821192,
      "grad_norm": 1.2621084239078817,
      "learning_rate": 0.00023534999999999997,
      "loss": 2.5156,
      "step": 1569
    },
    {
      "epoch": 1.0397350993377483,
      "grad_norm": 1.321421061278166,
      "learning_rate": 0.00023549999999999998,
      "loss": 2.7188,
      "step": 1570
    },
    {
      "epoch": 1.0403973509933775,
      "grad_norm": 1.2486199499154973,
      "learning_rate": 0.00023564999999999998,
      "loss": 2.7031,
      "step": 1571
    },
    {
      "epoch": 1.0410596026490067,
      "grad_norm": 1.0981257589378932,
      "learning_rate": 0.00023579999999999999,
      "loss": 2.2031,
      "step": 1572
    },
    {
      "epoch": 1.0417218543046358,
      "grad_norm": 1.343385244858492,
      "learning_rate": 0.00023594999999999996,
      "loss": 2.7344,
      "step": 1573
    },
    {
      "epoch": 1.042384105960265,
      "grad_norm": 1.1794000908581665,
      "learning_rate": 0.0002361,
      "loss": 2.625,
      "step": 1574
    },
    {
      "epoch": 1.0430463576158941,
      "grad_norm": 1.1591135173877467,
      "learning_rate": 0.00023624999999999997,
      "loss": 2.8438,
      "step": 1575
    },
    {
      "epoch": 1.043708609271523,
      "grad_norm": 1.1676763218705613,
      "learning_rate": 0.0002364,
      "loss": 2.4062,
      "step": 1576
    },
    {
      "epoch": 1.0443708609271523,
      "grad_norm": 1.2244496284831055,
      "learning_rate": 0.00023654999999999998,
      "loss": 2.2031,
      "step": 1577
    },
    {
      "epoch": 1.0450331125827814,
      "grad_norm": 1.3276820451425495,
      "learning_rate": 0.00023669999999999998,
      "loss": 2.6094,
      "step": 1578
    },
    {
      "epoch": 1.0456953642384106,
      "grad_norm": 1.6104766702295796,
      "learning_rate": 0.00023684999999999998,
      "loss": 2.5312,
      "step": 1579
    },
    {
      "epoch": 1.0463576158940397,
      "grad_norm": 1.4810454758411034,
      "learning_rate": 0.000237,
      "loss": 2.5781,
      "step": 1580
    },
    {
      "epoch": 1.047019867549669,
      "grad_norm": 1.172700561529911,
      "learning_rate": 0.00023714999999999996,
      "loss": 2.6562,
      "step": 1581
    },
    {
      "epoch": 1.047682119205298,
      "grad_norm": 1.453269326375865,
      "learning_rate": 0.0002373,
      "loss": 3.0469,
      "step": 1582
    },
    {
      "epoch": 1.0483443708609272,
      "grad_norm": 1.2922896956287713,
      "learning_rate": 0.00023744999999999997,
      "loss": 2.5781,
      "step": 1583
    },
    {
      "epoch": 1.0490066225165562,
      "grad_norm": 1.2038122677186902,
      "learning_rate": 0.0002376,
      "loss": 2.5156,
      "step": 1584
    },
    {
      "epoch": 1.0496688741721854,
      "grad_norm": 1.2329090917741925,
      "learning_rate": 0.00023774999999999998,
      "loss": 2.6719,
      "step": 1585
    },
    {
      "epoch": 1.0503311258278145,
      "grad_norm": 1.5076944294027734,
      "learning_rate": 0.00023789999999999998,
      "loss": 2.9844,
      "step": 1586
    },
    {
      "epoch": 1.0509933774834437,
      "grad_norm": 1.2657993003787582,
      "learning_rate": 0.00023804999999999999,
      "loss": 2.7812,
      "step": 1587
    },
    {
      "epoch": 1.0516556291390728,
      "grad_norm": 1.2728062099835233,
      "learning_rate": 0.0002382,
      "loss": 2.7656,
      "step": 1588
    },
    {
      "epoch": 1.052317880794702,
      "grad_norm": 1.245135169605465,
      "learning_rate": 0.00023834999999999997,
      "loss": 2.6875,
      "step": 1589
    },
    {
      "epoch": 1.0529801324503312,
      "grad_norm": 1.4250464798427807,
      "learning_rate": 0.0002385,
      "loss": 2.1875,
      "step": 1590
    },
    {
      "epoch": 1.0536423841059603,
      "grad_norm": 1.1954242051112403,
      "learning_rate": 0.00023864999999999997,
      "loss": 2.7812,
      "step": 1591
    },
    {
      "epoch": 1.0543046357615895,
      "grad_norm": 1.1388735172418343,
      "learning_rate": 0.0002388,
      "loss": 2.6719,
      "step": 1592
    },
    {
      "epoch": 1.0549668874172184,
      "grad_norm": 1.3377942515109469,
      "learning_rate": 0.00023894999999999998,
      "loss": 2.7812,
      "step": 1593
    },
    {
      "epoch": 1.0556291390728476,
      "grad_norm": 1.3001557888599227,
      "learning_rate": 0.00023909999999999998,
      "loss": 2.3438,
      "step": 1594
    },
    {
      "epoch": 1.0562913907284768,
      "grad_norm": 1.2221372185665975,
      "learning_rate": 0.00023925,
      "loss": 2.625,
      "step": 1595
    },
    {
      "epoch": 1.056953642384106,
      "grad_norm": 1.1734110574068584,
      "learning_rate": 0.0002394,
      "loss": 2.6406,
      "step": 1596
    },
    {
      "epoch": 1.057615894039735,
      "grad_norm": 1.1710055444077583,
      "learning_rate": 0.00023954999999999997,
      "loss": 2.4375,
      "step": 1597
    },
    {
      "epoch": 1.0582781456953643,
      "grad_norm": 1.1088485707544646,
      "learning_rate": 0.0002397,
      "loss": 2.5938,
      "step": 1598
    },
    {
      "epoch": 1.0589403973509934,
      "grad_norm": 1.1576755756800217,
      "learning_rate": 0.00023984999999999998,
      "loss": 2.625,
      "step": 1599
    },
    {
      "epoch": 1.0596026490066226,
      "grad_norm": 1.3247251491326573,
      "learning_rate": 0.00023999999999999998,
      "loss": 2.8906,
      "step": 1600
    },
    {
      "epoch": 1.0602649006622515,
      "grad_norm": 1.1981481143974533,
      "learning_rate": 0.00024014999999999998,
      "loss": 2.75,
      "step": 1601
    },
    {
      "epoch": 1.0609271523178807,
      "grad_norm": 1.3537667539769362,
      "learning_rate": 0.00024029999999999999,
      "loss": 2.8281,
      "step": 1602
    },
    {
      "epoch": 1.0615894039735099,
      "grad_norm": 1.2772292682770163,
      "learning_rate": 0.00024044999999999996,
      "loss": 2.3594,
      "step": 1603
    },
    {
      "epoch": 1.062251655629139,
      "grad_norm": 1.2365721765500288,
      "learning_rate": 0.0002406,
      "loss": 2.5312,
      "step": 1604
    },
    {
      "epoch": 1.0629139072847682,
      "grad_norm": 1.250090677626946,
      "learning_rate": 0.00024074999999999997,
      "loss": 2.7188,
      "step": 1605
    },
    {
      "epoch": 1.0635761589403974,
      "grad_norm": 1.27388031390634,
      "learning_rate": 0.0002409,
      "loss": 2.5781,
      "step": 1606
    },
    {
      "epoch": 1.0642384105960265,
      "grad_norm": 1.2312540272219987,
      "learning_rate": 0.00024104999999999998,
      "loss": 2.4062,
      "step": 1607
    },
    {
      "epoch": 1.0649006622516557,
      "grad_norm": 1.3555542209281282,
      "learning_rate": 0.00024119999999999998,
      "loss": 2.7031,
      "step": 1608
    },
    {
      "epoch": 1.0655629139072849,
      "grad_norm": 1.1456001352380034,
      "learning_rate": 0.00024134999999999998,
      "loss": 2.625,
      "step": 1609
    },
    {
      "epoch": 1.0662251655629138,
      "grad_norm": 1.2047348141768965,
      "learning_rate": 0.0002415,
      "loss": 2.5156,
      "step": 1610
    },
    {
      "epoch": 1.066887417218543,
      "grad_norm": 1.26752882240382,
      "learning_rate": 0.00024164999999999996,
      "loss": 2.5625,
      "step": 1611
    },
    {
      "epoch": 1.0675496688741721,
      "grad_norm": 1.1995893800531765,
      "learning_rate": 0.0002418,
      "loss": 2.7656,
      "step": 1612
    },
    {
      "epoch": 1.0682119205298013,
      "grad_norm": 1.273182808041412,
      "learning_rate": 0.00024194999999999997,
      "loss": 2.3281,
      "step": 1613
    },
    {
      "epoch": 1.0688741721854305,
      "grad_norm": 1.231035323469265,
      "learning_rate": 0.0002421,
      "loss": 2.8125,
      "step": 1614
    },
    {
      "epoch": 1.0695364238410596,
      "grad_norm": 1.332825608098906,
      "learning_rate": 0.00024224999999999998,
      "loss": 2.7031,
      "step": 1615
    },
    {
      "epoch": 1.0701986754966888,
      "grad_norm": 1.2130123545552294,
      "learning_rate": 0.00024239999999999998,
      "loss": 2.2031,
      "step": 1616
    },
    {
      "epoch": 1.070860927152318,
      "grad_norm": 1.2671768177985228,
      "learning_rate": 0.00024255,
      "loss": 2.8281,
      "step": 1617
    },
    {
      "epoch": 1.0715231788079471,
      "grad_norm": 1.333311804670691,
      "learning_rate": 0.0002427,
      "loss": 2.5,
      "step": 1618
    },
    {
      "epoch": 1.072185430463576,
      "grad_norm": 1.2858043740631016,
      "learning_rate": 0.00024284999999999997,
      "loss": 2.5312,
      "step": 1619
    },
    {
      "epoch": 1.0728476821192052,
      "grad_norm": 1.3361191899024603,
      "learning_rate": 0.000243,
      "loss": 2.875,
      "step": 1620
    },
    {
      "epoch": 1.0735099337748344,
      "grad_norm": 1.2260415764590065,
      "learning_rate": 0.00024314999999999997,
      "loss": 2.75,
      "step": 1621
    },
    {
      "epoch": 1.0741721854304636,
      "grad_norm": 1.076658716628758,
      "learning_rate": 0.0002433,
      "loss": 2.5469,
      "step": 1622
    },
    {
      "epoch": 1.0748344370860927,
      "grad_norm": 1.4613902109563353,
      "learning_rate": 0.00024344999999999998,
      "loss": 2.8438,
      "step": 1623
    },
    {
      "epoch": 1.0754966887417219,
      "grad_norm": 1.210706024203546,
      "learning_rate": 0.00024359999999999999,
      "loss": 2.5781,
      "step": 1624
    },
    {
      "epoch": 1.076158940397351,
      "grad_norm": 1.1166941573679585,
      "learning_rate": 0.00024375,
      "loss": 2.6875,
      "step": 1625
    },
    {
      "epoch": 1.0768211920529802,
      "grad_norm": 1.2132920715234536,
      "learning_rate": 0.00024389999999999997,
      "loss": 2.4531,
      "step": 1626
    },
    {
      "epoch": 1.0774834437086094,
      "grad_norm": 1.4011719166615808,
      "learning_rate": 0.00024404999999999997,
      "loss": 2.8125,
      "step": 1627
    },
    {
      "epoch": 1.0781456953642383,
      "grad_norm": 1.4141550951982462,
      "learning_rate": 0.00024419999999999997,
      "loss": 2.7344,
      "step": 1628
    },
    {
      "epoch": 1.0788079470198675,
      "grad_norm": 1.1537325697006269,
      "learning_rate": 0.00024435,
      "loss": 2.25,
      "step": 1629
    },
    {
      "epoch": 1.0794701986754967,
      "grad_norm": 1.421258144061272,
      "learning_rate": 0.0002445,
      "loss": 2.9375,
      "step": 1630
    },
    {
      "epoch": 1.0801324503311258,
      "grad_norm": 1.2052956901126441,
      "learning_rate": 0.00024464999999999996,
      "loss": 2.6406,
      "step": 1631
    },
    {
      "epoch": 1.080794701986755,
      "grad_norm": 1.1730647179923297,
      "learning_rate": 0.0002448,
      "loss": 2.9062,
      "step": 1632
    },
    {
      "epoch": 1.0814569536423841,
      "grad_norm": 1.2074540396578712,
      "learning_rate": 0.00024494999999999996,
      "loss": 2.4219,
      "step": 1633
    },
    {
      "epoch": 1.0821192052980133,
      "grad_norm": 1.1492915684813103,
      "learning_rate": 0.00024509999999999994,
      "loss": 2.8906,
      "step": 1634
    },
    {
      "epoch": 1.0827814569536425,
      "grad_norm": 1.238133783864804,
      "learning_rate": 0.00024524999999999997,
      "loss": 2.5938,
      "step": 1635
    },
    {
      "epoch": 1.0834437086092716,
      "grad_norm": 1.1827514201750082,
      "learning_rate": 0.00024539999999999995,
      "loss": 2.3594,
      "step": 1636
    },
    {
      "epoch": 1.0841059602649006,
      "grad_norm": 1.052229407613025,
      "learning_rate": 0.00024555,
      "loss": 2.5,
      "step": 1637
    },
    {
      "epoch": 1.0847682119205297,
      "grad_norm": 1.2732819129812245,
      "learning_rate": 0.00024569999999999995,
      "loss": 2.6094,
      "step": 1638
    },
    {
      "epoch": 1.085430463576159,
      "grad_norm": 1.2875860681034854,
      "learning_rate": 0.00024585,
      "loss": 2.7344,
      "step": 1639
    },
    {
      "epoch": 1.086092715231788,
      "grad_norm": 1.1405165421664474,
      "learning_rate": 0.00024599999999999996,
      "loss": 2.5781,
      "step": 1640
    },
    {
      "epoch": 1.0867549668874172,
      "grad_norm": 1.2120375235367489,
      "learning_rate": 0.00024615,
      "loss": 2.4531,
      "step": 1641
    },
    {
      "epoch": 1.0874172185430464,
      "grad_norm": 1.2397354601004646,
      "learning_rate": 0.00024629999999999997,
      "loss": 2.625,
      "step": 1642
    },
    {
      "epoch": 1.0880794701986756,
      "grad_norm": 1.1435946704433604,
      "learning_rate": 0.00024645,
      "loss": 2.75,
      "step": 1643
    },
    {
      "epoch": 1.0887417218543047,
      "grad_norm": 2.353588092960578,
      "learning_rate": 0.0002466,
      "loss": 2.4219,
      "step": 1644
    },
    {
      "epoch": 1.0894039735099337,
      "grad_norm": 1.2343085685969089,
      "learning_rate": 0.00024675,
      "loss": 2.5312,
      "step": 1645
    },
    {
      "epoch": 1.0900662251655628,
      "grad_norm": 1.246652446039875,
      "learning_rate": 0.0002469,
      "loss": 2.75,
      "step": 1646
    },
    {
      "epoch": 1.090728476821192,
      "grad_norm": 1.2455743322228694,
      "learning_rate": 0.00024704999999999996,
      "loss": 2.4531,
      "step": 1647
    },
    {
      "epoch": 1.0913907284768212,
      "grad_norm": 1.1590575700470773,
      "learning_rate": 0.0002472,
      "loss": 2.625,
      "step": 1648
    },
    {
      "epoch": 1.0920529801324503,
      "grad_norm": 1.188198469518599,
      "learning_rate": 0.00024734999999999997,
      "loss": 2.5625,
      "step": 1649
    },
    {
      "epoch": 1.0927152317880795,
      "grad_norm": 1.3896858328668413,
      "learning_rate": 0.00024749999999999994,
      "loss": 3.1094,
      "step": 1650
    },
    {
      "epoch": 1.0933774834437087,
      "grad_norm": 1.3182136720279225,
      "learning_rate": 0.00024765,
      "loss": 2.8906,
      "step": 1651
    },
    {
      "epoch": 1.0940397350993378,
      "grad_norm": 1.3558794789532171,
      "learning_rate": 0.00024779999999999995,
      "loss": 2.8281,
      "step": 1652
    },
    {
      "epoch": 1.0947019867549668,
      "grad_norm": 1.2896461186363017,
      "learning_rate": 0.00024795,
      "loss": 2.6094,
      "step": 1653
    },
    {
      "epoch": 1.095364238410596,
      "grad_norm": 1.3043000889636394,
      "learning_rate": 0.00024809999999999996,
      "loss": 2.5938,
      "step": 1654
    },
    {
      "epoch": 1.096026490066225,
      "grad_norm": 1.2844762333350255,
      "learning_rate": 0.00024825,
      "loss": 2.625,
      "step": 1655
    },
    {
      "epoch": 1.0966887417218543,
      "grad_norm": 1.2866248801237585,
      "learning_rate": 0.00024839999999999997,
      "loss": 2.2812,
      "step": 1656
    },
    {
      "epoch": 1.0973509933774834,
      "grad_norm": 1.312814696444647,
      "learning_rate": 0.00024855,
      "loss": 2.7188,
      "step": 1657
    },
    {
      "epoch": 1.0980132450331126,
      "grad_norm": 1.630454701991641,
      "learning_rate": 0.0002487,
      "loss": 2.6406,
      "step": 1658
    },
    {
      "epoch": 1.0986754966887418,
      "grad_norm": 1.1246236115513686,
      "learning_rate": 0.00024885,
      "loss": 2.6719,
      "step": 1659
    },
    {
      "epoch": 1.099337748344371,
      "grad_norm": 1.1126051895827604,
      "learning_rate": 0.000249,
      "loss": 2.2031,
      "step": 1660
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0972085560166904,
      "learning_rate": 0.00024914999999999996,
      "loss": 2.5312,
      "step": 1661
    },
    {
      "epoch": 1.100662251655629,
      "grad_norm": 1.3072905531956498,
      "learning_rate": 0.0002493,
      "loss": 2.9062,
      "step": 1662
    },
    {
      "epoch": 1.1013245033112582,
      "grad_norm": 1.4518018413097984,
      "learning_rate": 0.00024944999999999996,
      "loss": 2.8281,
      "step": 1663
    },
    {
      "epoch": 1.1019867549668874,
      "grad_norm": 1.20203731948028,
      "learning_rate": 0.00024959999999999994,
      "loss": 2.625,
      "step": 1664
    },
    {
      "epoch": 1.1026490066225165,
      "grad_norm": 1.1615835788212872,
      "learning_rate": 0.00024974999999999997,
      "loss": 2.4688,
      "step": 1665
    },
    {
      "epoch": 1.1033112582781457,
      "grad_norm": 1.2546269774309333,
      "learning_rate": 0.00024989999999999995,
      "loss": 2.7969,
      "step": 1666
    },
    {
      "epoch": 1.1039735099337749,
      "grad_norm": 1.1754226402658847,
      "learning_rate": 0.00025005,
      "loss": 2.6719,
      "step": 1667
    },
    {
      "epoch": 1.104635761589404,
      "grad_norm": 1.176987767046394,
      "learning_rate": 0.00025019999999999996,
      "loss": 2.3594,
      "step": 1668
    },
    {
      "epoch": 1.1052980132450332,
      "grad_norm": 1.1554922206862372,
      "learning_rate": 0.00025035,
      "loss": 2.4844,
      "step": 1669
    },
    {
      "epoch": 1.1059602649006623,
      "grad_norm": 1.0759195957639154,
      "learning_rate": 0.00025049999999999996,
      "loss": 2.4531,
      "step": 1670
    },
    {
      "epoch": 1.1066225165562913,
      "grad_norm": 1.1755267738732726,
      "learning_rate": 0.00025065,
      "loss": 2.7812,
      "step": 1671
    },
    {
      "epoch": 1.1072847682119205,
      "grad_norm": 1.3691510355513095,
      "learning_rate": 0.00025079999999999997,
      "loss": 2.8125,
      "step": 1672
    },
    {
      "epoch": 1.1079470198675496,
      "grad_norm": 1.133783529547844,
      "learning_rate": 0.00025095,
      "loss": 2.5781,
      "step": 1673
    },
    {
      "epoch": 1.1086092715231788,
      "grad_norm": 1.162411982950755,
      "learning_rate": 0.0002511,
      "loss": 2.5781,
      "step": 1674
    },
    {
      "epoch": 1.109271523178808,
      "grad_norm": 1.1715697240606722,
      "learning_rate": 0.00025125,
      "loss": 2.7344,
      "step": 1675
    },
    {
      "epoch": 1.1099337748344371,
      "grad_norm": 1.4625296971749757,
      "learning_rate": 0.0002514,
      "loss": 2.6094,
      "step": 1676
    },
    {
      "epoch": 1.1105960264900663,
      "grad_norm": 1.1710803086685577,
      "learning_rate": 0.00025154999999999996,
      "loss": 2.7656,
      "step": 1677
    },
    {
      "epoch": 1.1112582781456954,
      "grad_norm": 2.323879291813002,
      "learning_rate": 0.0002517,
      "loss": 2.1562,
      "step": 1678
    },
    {
      "epoch": 1.1119205298013246,
      "grad_norm": 1.1364833908535334,
      "learning_rate": 0.00025184999999999997,
      "loss": 2.625,
      "step": 1679
    },
    {
      "epoch": 1.1125827814569536,
      "grad_norm": 2.1486042567064962,
      "learning_rate": 0.00025199999999999995,
      "loss": 2.6406,
      "step": 1680
    },
    {
      "epoch": 1.1132450331125827,
      "grad_norm": 1.1634430751352385,
      "learning_rate": 0.00025215,
      "loss": 2.6094,
      "step": 1681
    },
    {
      "epoch": 1.1139072847682119,
      "grad_norm": 1.3609180209389597,
      "learning_rate": 0.00025229999999999995,
      "loss": 2.75,
      "step": 1682
    },
    {
      "epoch": 1.114569536423841,
      "grad_norm": 1.1500410418858928,
      "learning_rate": 0.00025245,
      "loss": 2.7031,
      "step": 1683
    },
    {
      "epoch": 1.1152317880794702,
      "grad_norm": 1.1877100251352473,
      "learning_rate": 0.00025259999999999996,
      "loss": 2.4844,
      "step": 1684
    },
    {
      "epoch": 1.1158940397350994,
      "grad_norm": 1.104707764604417,
      "learning_rate": 0.00025275,
      "loss": 2.5,
      "step": 1685
    },
    {
      "epoch": 1.1165562913907285,
      "grad_norm": 1.3214005237825592,
      "learning_rate": 0.00025289999999999997,
      "loss": 2.8281,
      "step": 1686
    },
    {
      "epoch": 1.1172185430463577,
      "grad_norm": 1.278809972119621,
      "learning_rate": 0.00025305,
      "loss": 2.5625,
      "step": 1687
    },
    {
      "epoch": 1.1178807947019869,
      "grad_norm": 1.091011274451508,
      "learning_rate": 0.0002532,
      "loss": 2.5156,
      "step": 1688
    },
    {
      "epoch": 1.1185430463576158,
      "grad_norm": 1.0410842105575855,
      "learning_rate": 0.00025335,
      "loss": 2.7344,
      "step": 1689
    },
    {
      "epoch": 1.119205298013245,
      "grad_norm": 1.1961207006253298,
      "learning_rate": 0.0002535,
      "loss": 2.7031,
      "step": 1690
    },
    {
      "epoch": 1.1198675496688741,
      "grad_norm": 1.3488806828599484,
      "learning_rate": 0.00025365,
      "loss": 2.875,
      "step": 1691
    },
    {
      "epoch": 1.1205298013245033,
      "grad_norm": 1.0579380401956666,
      "learning_rate": 0.0002538,
      "loss": 2.3594,
      "step": 1692
    },
    {
      "epoch": 1.1211920529801325,
      "grad_norm": 1.1650879118266533,
      "learning_rate": 0.00025394999999999997,
      "loss": 2.6562,
      "step": 1693
    },
    {
      "epoch": 1.1218543046357616,
      "grad_norm": 1.2260495678601286,
      "learning_rate": 0.0002541,
      "loss": 2.625,
      "step": 1694
    },
    {
      "epoch": 1.1225165562913908,
      "grad_norm": 1.072137291018581,
      "learning_rate": 0.00025425,
      "loss": 2.4531,
      "step": 1695
    },
    {
      "epoch": 1.12317880794702,
      "grad_norm": 1.3105079553922625,
      "learning_rate": 0.00025439999999999995,
      "loss": 2.7031,
      "step": 1696
    },
    {
      "epoch": 1.123841059602649,
      "grad_norm": 1.2354398698425346,
      "learning_rate": 0.00025455,
      "loss": 2.6562,
      "step": 1697
    },
    {
      "epoch": 1.124503311258278,
      "grad_norm": 1.0373715050130747,
      "learning_rate": 0.00025469999999999996,
      "loss": 2.4219,
      "step": 1698
    },
    {
      "epoch": 1.1251655629139072,
      "grad_norm": 1.2978610918200406,
      "learning_rate": 0.00025485,
      "loss": 2.8906,
      "step": 1699
    },
    {
      "epoch": 1.1258278145695364,
      "grad_norm": 1.2790280840366408,
      "learning_rate": 0.00025499999999999996,
      "loss": 2.7656,
      "step": 1700
    },
    {
      "epoch": 1.1264900662251656,
      "grad_norm": 1.159972164201933,
      "learning_rate": 0.00025515,
      "loss": 2.625,
      "step": 1701
    },
    {
      "epoch": 1.1271523178807947,
      "grad_norm": 1.0967399683412424,
      "learning_rate": 0.00025529999999999997,
      "loss": 2.4844,
      "step": 1702
    },
    {
      "epoch": 1.127814569536424,
      "grad_norm": 1.142315209595986,
      "learning_rate": 0.00025545,
      "loss": 2.5625,
      "step": 1703
    },
    {
      "epoch": 1.128476821192053,
      "grad_norm": 1.2966698100551113,
      "learning_rate": 0.0002556,
      "loss": 2.5781,
      "step": 1704
    },
    {
      "epoch": 1.129139072847682,
      "grad_norm": 1.165333404824752,
      "learning_rate": 0.00025575,
      "loss": 2.5625,
      "step": 1705
    },
    {
      "epoch": 1.1298013245033112,
      "grad_norm": 1.20836265587115,
      "learning_rate": 0.0002559,
      "loss": 2.6094,
      "step": 1706
    },
    {
      "epoch": 1.1304635761589403,
      "grad_norm": 1.3134063937569378,
      "learning_rate": 0.00025604999999999996,
      "loss": 2.6562,
      "step": 1707
    },
    {
      "epoch": 1.1311258278145695,
      "grad_norm": 1.1331269296615698,
      "learning_rate": 0.0002562,
      "loss": 2.5781,
      "step": 1708
    },
    {
      "epoch": 1.1317880794701987,
      "grad_norm": 1.3363666135147227,
      "learning_rate": 0.00025634999999999997,
      "loss": 2.7031,
      "step": 1709
    },
    {
      "epoch": 1.1324503311258278,
      "grad_norm": 1.4190394179186674,
      "learning_rate": 0.00025649999999999995,
      "loss": 2.7656,
      "step": 1710
    },
    {
      "epoch": 1.133112582781457,
      "grad_norm": 1.4684299382954138,
      "learning_rate": 0.00025665,
      "loss": 2.0938,
      "step": 1711
    },
    {
      "epoch": 1.1337748344370862,
      "grad_norm": 1.1017845971765399,
      "learning_rate": 0.00025679999999999995,
      "loss": 2.75,
      "step": 1712
    },
    {
      "epoch": 1.1344370860927153,
      "grad_norm": 1.136612360153799,
      "learning_rate": 0.00025695,
      "loss": 2.5781,
      "step": 1713
    },
    {
      "epoch": 1.1350993377483443,
      "grad_norm": 1.2545248225133467,
      "learning_rate": 0.00025709999999999996,
      "loss": 2.5312,
      "step": 1714
    },
    {
      "epoch": 1.1357615894039734,
      "grad_norm": 1.2591202972033981,
      "learning_rate": 0.00025725,
      "loss": 2.5938,
      "step": 1715
    },
    {
      "epoch": 1.1364238410596026,
      "grad_norm": 1.30873961947508,
      "learning_rate": 0.00025739999999999997,
      "loss": 2.6562,
      "step": 1716
    },
    {
      "epoch": 1.1370860927152318,
      "grad_norm": 1.2035134867073713,
      "learning_rate": 0.00025755,
      "loss": 2.4531,
      "step": 1717
    },
    {
      "epoch": 1.137748344370861,
      "grad_norm": 1.158631499261634,
      "learning_rate": 0.0002577,
      "loss": 2.2812,
      "step": 1718
    },
    {
      "epoch": 1.13841059602649,
      "grad_norm": 2.0932306768215154,
      "learning_rate": 0.00025785,
      "loss": 2.6406,
      "step": 1719
    },
    {
      "epoch": 1.1390728476821192,
      "grad_norm": 1.2269890503169205,
      "learning_rate": 0.000258,
      "loss": 2.2656,
      "step": 1720
    },
    {
      "epoch": 1.1397350993377484,
      "grad_norm": 1.066683277383856,
      "learning_rate": 0.00025815,
      "loss": 2.0938,
      "step": 1721
    },
    {
      "epoch": 1.1403973509933776,
      "grad_norm": 1.188651566352273,
      "learning_rate": 0.0002583,
      "loss": 2.5469,
      "step": 1722
    },
    {
      "epoch": 1.1410596026490065,
      "grad_norm": 4.702704389759291,
      "learning_rate": 0.00025844999999999997,
      "loss": 2.5781,
      "step": 1723
    },
    {
      "epoch": 1.1417218543046357,
      "grad_norm": 1.1918284630421399,
      "learning_rate": 0.0002586,
      "loss": 2.2969,
      "step": 1724
    },
    {
      "epoch": 1.1423841059602649,
      "grad_norm": 1.1595351168687273,
      "learning_rate": 0.00025875,
      "loss": 2.625,
      "step": 1725
    },
    {
      "epoch": 1.143046357615894,
      "grad_norm": 1.1159134780954056,
      "learning_rate": 0.00025889999999999995,
      "loss": 2.7188,
      "step": 1726
    },
    {
      "epoch": 1.1437086092715232,
      "grad_norm": 1.8241405672567241,
      "learning_rate": 0.00025905,
      "loss": 2.5469,
      "step": 1727
    },
    {
      "epoch": 1.1443708609271523,
      "grad_norm": 1.1784729568836416,
      "learning_rate": 0.00025919999999999996,
      "loss": 2.3125,
      "step": 1728
    },
    {
      "epoch": 1.1450331125827815,
      "grad_norm": 1.1386849788674025,
      "learning_rate": 0.00025935,
      "loss": 2.375,
      "step": 1729
    },
    {
      "epoch": 1.1456953642384107,
      "grad_norm": 1.23385921784263,
      "learning_rate": 0.00025949999999999997,
      "loss": 2.7031,
      "step": 1730
    },
    {
      "epoch": 1.1463576158940398,
      "grad_norm": 1.1210794209092474,
      "learning_rate": 0.00025965,
      "loss": 2.4688,
      "step": 1731
    },
    {
      "epoch": 1.1470198675496688,
      "grad_norm": 1.1569069294558514,
      "learning_rate": 0.00025979999999999997,
      "loss": 2.6562,
      "step": 1732
    },
    {
      "epoch": 1.147682119205298,
      "grad_norm": 10.521586044719268,
      "learning_rate": 0.00025995,
      "loss": 2.8438,
      "step": 1733
    },
    {
      "epoch": 1.148344370860927,
      "grad_norm": 1.3301959596358284,
      "learning_rate": 0.0002601,
      "loss": 2.6406,
      "step": 1734
    },
    {
      "epoch": 1.1490066225165563,
      "grad_norm": 8.120003968526621,
      "learning_rate": 0.00026025,
      "loss": 2.4219,
      "step": 1735
    },
    {
      "epoch": 1.1496688741721854,
      "grad_norm": 1.1207312445208972,
      "learning_rate": 0.0002604,
      "loss": 2.4844,
      "step": 1736
    },
    {
      "epoch": 1.1503311258278146,
      "grad_norm": 1.3120139170323604,
      "learning_rate": 0.00026055,
      "loss": 2.5469,
      "step": 1737
    },
    {
      "epoch": 1.1509933774834438,
      "grad_norm": 1.2958513980527286,
      "learning_rate": 0.0002607,
      "loss": 2.75,
      "step": 1738
    },
    {
      "epoch": 1.151655629139073,
      "grad_norm": 1.4140541985633366,
      "learning_rate": 0.00026084999999999997,
      "loss": 2.6875,
      "step": 1739
    },
    {
      "epoch": 1.152317880794702,
      "grad_norm": 1.207831720792249,
      "learning_rate": 0.000261,
      "loss": 2.6406,
      "step": 1740
    },
    {
      "epoch": 1.152980132450331,
      "grad_norm": 1.0937650766914155,
      "learning_rate": 0.00026115,
      "loss": 2.75,
      "step": 1741
    },
    {
      "epoch": 1.1536423841059602,
      "grad_norm": 1.1399716762029208,
      "learning_rate": 0.00026129999999999995,
      "loss": 2.5938,
      "step": 1742
    },
    {
      "epoch": 1.1543046357615894,
      "grad_norm": 1.1600664033985133,
      "learning_rate": 0.00026145,
      "loss": 2.7656,
      "step": 1743
    },
    {
      "epoch": 1.1549668874172185,
      "grad_norm": 1.2212739781012554,
      "learning_rate": 0.00026159999999999996,
      "loss": 2.7031,
      "step": 1744
    },
    {
      "epoch": 1.1556291390728477,
      "grad_norm": 1.138208459770132,
      "learning_rate": 0.00026175,
      "loss": 2.4844,
      "step": 1745
    },
    {
      "epoch": 1.1562913907284769,
      "grad_norm": 1.132639228330785,
      "learning_rate": 0.00026189999999999997,
      "loss": 2.4219,
      "step": 1746
    },
    {
      "epoch": 1.156953642384106,
      "grad_norm": 1.2796335914899115,
      "learning_rate": 0.00026205,
      "loss": 2.8125,
      "step": 1747
    },
    {
      "epoch": 1.1576158940397352,
      "grad_norm": 1.0897491600739748,
      "learning_rate": 0.0002622,
      "loss": 2.5625,
      "step": 1748
    },
    {
      "epoch": 1.1582781456953644,
      "grad_norm": 1.2290771899974522,
      "learning_rate": 0.00026235,
      "loss": 2.9531,
      "step": 1749
    },
    {
      "epoch": 1.1589403973509933,
      "grad_norm": 1.2031799640137273,
      "learning_rate": 0.0002625,
      "loss": 2.6562,
      "step": 1750
    },
    {
      "epoch": 1.1596026490066225,
      "grad_norm": 2.9651636154760133,
      "learning_rate": 0.00026264999999999996,
      "loss": 2.4375,
      "step": 1751
    },
    {
      "epoch": 1.1602649006622516,
      "grad_norm": 1.1813701470718891,
      "learning_rate": 0.0002628,
      "loss": 2.7188,
      "step": 1752
    },
    {
      "epoch": 1.1609271523178808,
      "grad_norm": 1.1737875231560162,
      "learning_rate": 0.00026294999999999997,
      "loss": 2.6094,
      "step": 1753
    },
    {
      "epoch": 1.16158940397351,
      "grad_norm": 1.285964083835266,
      "learning_rate": 0.0002631,
      "loss": 2.6406,
      "step": 1754
    },
    {
      "epoch": 1.1622516556291391,
      "grad_norm": 1.1717580706899449,
      "learning_rate": 0.00026325,
      "loss": 2.7344,
      "step": 1755
    },
    {
      "epoch": 1.1629139072847683,
      "grad_norm": 1.1143003248234185,
      "learning_rate": 0.00026339999999999995,
      "loss": 2.2969,
      "step": 1756
    },
    {
      "epoch": 1.1635761589403972,
      "grad_norm": 1.1414163884818587,
      "learning_rate": 0.00026355,
      "loss": 2.6719,
      "step": 1757
    },
    {
      "epoch": 1.1642384105960264,
      "grad_norm": 1.0742409182865917,
      "learning_rate": 0.00026369999999999996,
      "loss": 2.1875,
      "step": 1758
    },
    {
      "epoch": 1.1649006622516556,
      "grad_norm": 1.1572617454121166,
      "learning_rate": 0.00026384999999999994,
      "loss": 2.5469,
      "step": 1759
    },
    {
      "epoch": 1.1655629139072847,
      "grad_norm": 1.050866407227508,
      "learning_rate": 0.00026399999999999997,
      "loss": 2.1094,
      "step": 1760
    },
    {
      "epoch": 1.166225165562914,
      "grad_norm": 1.1222401063498422,
      "learning_rate": 0.00026414999999999994,
      "loss": 2.6562,
      "step": 1761
    },
    {
      "epoch": 1.166887417218543,
      "grad_norm": 1.1194411807069933,
      "learning_rate": 0.0002643,
      "loss": 2.8125,
      "step": 1762
    },
    {
      "epoch": 1.1675496688741722,
      "grad_norm": 1.2329332430223954,
      "learning_rate": 0.00026444999999999995,
      "loss": 3.0156,
      "step": 1763
    },
    {
      "epoch": 1.1682119205298014,
      "grad_norm": 1.131044583189528,
      "learning_rate": 0.0002646,
      "loss": 2.7656,
      "step": 1764
    },
    {
      "epoch": 1.1688741721854305,
      "grad_norm": 1.276013354663163,
      "learning_rate": 0.00026474999999999996,
      "loss": 2.625,
      "step": 1765
    },
    {
      "epoch": 1.1695364238410595,
      "grad_norm": 1.2043168266830409,
      "learning_rate": 0.0002649,
      "loss": 2.1719,
      "step": 1766
    },
    {
      "epoch": 1.1701986754966887,
      "grad_norm": 1.305034018992924,
      "learning_rate": 0.00026504999999999996,
      "loss": 2.5469,
      "step": 1767
    },
    {
      "epoch": 1.1708609271523178,
      "grad_norm": 1.049354475796462,
      "learning_rate": 0.0002652,
      "loss": 2.3906,
      "step": 1768
    },
    {
      "epoch": 1.171523178807947,
      "grad_norm": 1.2748653446990472,
      "learning_rate": 0.00026534999999999997,
      "loss": 2.7031,
      "step": 1769
    },
    {
      "epoch": 1.1721854304635762,
      "grad_norm": 1.1656653287452228,
      "learning_rate": 0.0002655,
      "loss": 2.6562,
      "step": 1770
    },
    {
      "epoch": 1.1728476821192053,
      "grad_norm": 1.147529976887952,
      "learning_rate": 0.00026565,
      "loss": 2.5938,
      "step": 1771
    },
    {
      "epoch": 1.1735099337748345,
      "grad_norm": 2.791023841360208,
      "learning_rate": 0.00026579999999999996,
      "loss": 2.8125,
      "step": 1772
    },
    {
      "epoch": 1.1741721854304636,
      "grad_norm": 1.1588687716075585,
      "learning_rate": 0.00026595,
      "loss": 2.4375,
      "step": 1773
    },
    {
      "epoch": 1.1748344370860928,
      "grad_norm": 1.1222842084178932,
      "learning_rate": 0.00026609999999999996,
      "loss": 2.5469,
      "step": 1774
    },
    {
      "epoch": 1.1754966887417218,
      "grad_norm": 1.1452537517746582,
      "learning_rate": 0.00026624999999999994,
      "loss": 2.5469,
      "step": 1775
    },
    {
      "epoch": 1.176158940397351,
      "grad_norm": 1.0870029544867892,
      "learning_rate": 0.00026639999999999997,
      "loss": 2.6406,
      "step": 1776
    },
    {
      "epoch": 1.17682119205298,
      "grad_norm": 1.1280680366200098,
      "learning_rate": 0.00026654999999999995,
      "loss": 2.6875,
      "step": 1777
    },
    {
      "epoch": 1.1774834437086092,
      "grad_norm": 1.3989267118848066,
      "learning_rate": 0.0002667,
      "loss": 2.75,
      "step": 1778
    },
    {
      "epoch": 1.1781456953642384,
      "grad_norm": 1.25831780808853,
      "learning_rate": 0.00026684999999999995,
      "loss": 2.875,
      "step": 1779
    },
    {
      "epoch": 1.1788079470198676,
      "grad_norm": 1.1252858803886931,
      "learning_rate": 0.000267,
      "loss": 2.5938,
      "step": 1780
    },
    {
      "epoch": 1.1794701986754967,
      "grad_norm": 1.324177542464379,
      "learning_rate": 0.00026714999999999996,
      "loss": 2.9062,
      "step": 1781
    },
    {
      "epoch": 1.180132450331126,
      "grad_norm": 1.1241767300947374,
      "learning_rate": 0.0002673,
      "loss": 2.4688,
      "step": 1782
    },
    {
      "epoch": 1.180794701986755,
      "grad_norm": 1.0944027695429912,
      "learning_rate": 0.00026744999999999997,
      "loss": 2.4375,
      "step": 1783
    },
    {
      "epoch": 1.181456953642384,
      "grad_norm": 1.0478072975769979,
      "learning_rate": 0.0002676,
      "loss": 2.625,
      "step": 1784
    },
    {
      "epoch": 1.1821192052980132,
      "grad_norm": 1.1388623333905465,
      "learning_rate": 0.00026775,
      "loss": 2.8594,
      "step": 1785
    },
    {
      "epoch": 1.1827814569536423,
      "grad_norm": 1.0126028834819254,
      "learning_rate": 0.0002679,
      "loss": 2.6094,
      "step": 1786
    },
    {
      "epoch": 1.1834437086092715,
      "grad_norm": 1.1487591291413297,
      "learning_rate": 0.00026805,
      "loss": 2.6406,
      "step": 1787
    },
    {
      "epoch": 1.1841059602649007,
      "grad_norm": 1.0703098414937817,
      "learning_rate": 0.00026819999999999996,
      "loss": 2.6562,
      "step": 1788
    },
    {
      "epoch": 1.1847682119205298,
      "grad_norm": 1.0900945590520366,
      "learning_rate": 0.00026835,
      "loss": 2.7188,
      "step": 1789
    },
    {
      "epoch": 1.185430463576159,
      "grad_norm": 1.231901960880779,
      "learning_rate": 0.00026849999999999997,
      "loss": 2.5469,
      "step": 1790
    },
    {
      "epoch": 1.1860927152317882,
      "grad_norm": 1.1578911679305506,
      "learning_rate": 0.00026864999999999994,
      "loss": 2.7812,
      "step": 1791
    },
    {
      "epoch": 1.1867549668874173,
      "grad_norm": 1.0761349538375902,
      "learning_rate": 0.0002688,
      "loss": 2.4844,
      "step": 1792
    },
    {
      "epoch": 1.1874172185430463,
      "grad_norm": 1.166318238093845,
      "learning_rate": 0.00026894999999999995,
      "loss": 2.7656,
      "step": 1793
    },
    {
      "epoch": 1.1880794701986754,
      "grad_norm": 1.1443551272558634,
      "learning_rate": 0.0002691,
      "loss": 2.0781,
      "step": 1794
    },
    {
      "epoch": 1.1887417218543046,
      "grad_norm": 1.103815135595975,
      "learning_rate": 0.00026924999999999996,
      "loss": 2.4844,
      "step": 1795
    },
    {
      "epoch": 1.1894039735099338,
      "grad_norm": 1.2259929291235874,
      "learning_rate": 0.0002694,
      "loss": 2.8125,
      "step": 1796
    },
    {
      "epoch": 1.190066225165563,
      "grad_norm": 1.0364629181624325,
      "learning_rate": 0.00026954999999999997,
      "loss": 2.3594,
      "step": 1797
    },
    {
      "epoch": 1.190728476821192,
      "grad_norm": 1.2654999942913223,
      "learning_rate": 0.0002697,
      "loss": 2.2188,
      "step": 1798
    },
    {
      "epoch": 1.1913907284768213,
      "grad_norm": 1.1907409607502855,
      "learning_rate": 0.00026984999999999997,
      "loss": 2.6875,
      "step": 1799
    },
    {
      "epoch": 1.1920529801324504,
      "grad_norm": 1.1686414699543444,
      "learning_rate": 0.00027,
      "loss": 2.5312,
      "step": 1800
    },
    {
      "epoch": 1.1927152317880796,
      "grad_norm": 1.1773070773761583,
      "learning_rate": 0.00027015,
      "loss": 2.5312,
      "step": 1801
    },
    {
      "epoch": 1.1933774834437085,
      "grad_norm": 1.1743247615145778,
      "learning_rate": 0.00027029999999999996,
      "loss": 2.4375,
      "step": 1802
    },
    {
      "epoch": 1.1940397350993377,
      "grad_norm": 1.30979549916338,
      "learning_rate": 0.00027045,
      "loss": 2.5,
      "step": 1803
    },
    {
      "epoch": 1.1947019867549669,
      "grad_norm": 1.0878316245196766,
      "learning_rate": 0.00027059999999999996,
      "loss": 2.5156,
      "step": 1804
    },
    {
      "epoch": 1.195364238410596,
      "grad_norm": 1.2648602433390705,
      "learning_rate": 0.00027074999999999994,
      "loss": 2.5156,
      "step": 1805
    },
    {
      "epoch": 1.1960264900662252,
      "grad_norm": 1.2260850181319787,
      "learning_rate": 0.00027089999999999997,
      "loss": 2.5781,
      "step": 1806
    },
    {
      "epoch": 1.1966887417218544,
      "grad_norm": 1.0678250441718888,
      "learning_rate": 0.00027104999999999995,
      "loss": 2.4062,
      "step": 1807
    },
    {
      "epoch": 1.1973509933774835,
      "grad_norm": 1.1413106256524905,
      "learning_rate": 0.0002712,
      "loss": 2.6562,
      "step": 1808
    },
    {
      "epoch": 1.1980132450331125,
      "grad_norm": 1.3364971931152663,
      "learning_rate": 0.00027134999999999995,
      "loss": 2.7344,
      "step": 1809
    },
    {
      "epoch": 1.1986754966887416,
      "grad_norm": 1.072806367865523,
      "learning_rate": 0.0002715,
      "loss": 2.5156,
      "step": 1810
    },
    {
      "epoch": 1.1993377483443708,
      "grad_norm": 1.0494632182021033,
      "learning_rate": 0.00027164999999999996,
      "loss": 2.4375,
      "step": 1811
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0825141596084191,
      "learning_rate": 0.0002718,
      "loss": 2.1875,
      "step": 1812
    },
    {
      "epoch": 1.2006622516556291,
      "grad_norm": 1.241389117892049,
      "learning_rate": 0.00027194999999999997,
      "loss": 2.5312,
      "step": 1813
    },
    {
      "epoch": 1.2013245033112583,
      "grad_norm": 1.16805167446297,
      "learning_rate": 0.0002721,
      "loss": 2.5469,
      "step": 1814
    },
    {
      "epoch": 1.2019867549668874,
      "grad_norm": 1.1045410633803046,
      "learning_rate": 0.00027225,
      "loss": 2.5625,
      "step": 1815
    },
    {
      "epoch": 1.2026490066225166,
      "grad_norm": 1.1040235397770117,
      "learning_rate": 0.0002724,
      "loss": 1.9922,
      "step": 1816
    },
    {
      "epoch": 1.2033112582781458,
      "grad_norm": 1.0583540769047224,
      "learning_rate": 0.00027255,
      "loss": 2.6094,
      "step": 1817
    },
    {
      "epoch": 1.2039735099337747,
      "grad_norm": 1.1678403807291557,
      "learning_rate": 0.00027269999999999996,
      "loss": 2.6562,
      "step": 1818
    },
    {
      "epoch": 1.2046357615894039,
      "grad_norm": 1.2511723249734839,
      "learning_rate": 0.00027285,
      "loss": 2.875,
      "step": 1819
    },
    {
      "epoch": 1.205298013245033,
      "grad_norm": 1.138811259626744,
      "learning_rate": 0.00027299999999999997,
      "loss": 2.5781,
      "step": 1820
    },
    {
      "epoch": 1.2059602649006622,
      "grad_norm": 1.0547316711985497,
      "learning_rate": 0.00027314999999999994,
      "loss": 2.625,
      "step": 1821
    },
    {
      "epoch": 1.2066225165562914,
      "grad_norm": 1.1177023229551604,
      "learning_rate": 0.0002733,
      "loss": 2.4375,
      "step": 1822
    },
    {
      "epoch": 1.2072847682119205,
      "grad_norm": 1.0351003788206876,
      "learning_rate": 0.00027344999999999995,
      "loss": 2.5938,
      "step": 1823
    },
    {
      "epoch": 1.2079470198675497,
      "grad_norm": 1.0750611256258018,
      "learning_rate": 0.0002736,
      "loss": 2.5,
      "step": 1824
    },
    {
      "epoch": 1.2086092715231789,
      "grad_norm": 1.212251069032006,
      "learning_rate": 0.00027374999999999996,
      "loss": 2.6875,
      "step": 1825
    },
    {
      "epoch": 1.209271523178808,
      "grad_norm": 1.0892714684363998,
      "learning_rate": 0.0002739,
      "loss": 2.6562,
      "step": 1826
    },
    {
      "epoch": 1.209933774834437,
      "grad_norm": 1.045955630657899,
      "learning_rate": 0.00027404999999999997,
      "loss": 2.6094,
      "step": 1827
    },
    {
      "epoch": 1.2105960264900661,
      "grad_norm": 1.0131862121952844,
      "learning_rate": 0.0002742,
      "loss": 2.0625,
      "step": 1828
    },
    {
      "epoch": 1.2112582781456953,
      "grad_norm": 1.0616344338813846,
      "learning_rate": 0.00027435,
      "loss": 2.6094,
      "step": 1829
    },
    {
      "epoch": 1.2119205298013245,
      "grad_norm": 0.9947857381903571,
      "learning_rate": 0.0002745,
      "loss": 2.375,
      "step": 1830
    },
    {
      "epoch": 1.2125827814569536,
      "grad_norm": 1.0467204551351708,
      "learning_rate": 0.00027465,
      "loss": 2.375,
      "step": 1831
    },
    {
      "epoch": 1.2132450331125828,
      "grad_norm": 1.0531189716657996,
      "learning_rate": 0.0002748,
      "loss": 2.1875,
      "step": 1832
    },
    {
      "epoch": 1.213907284768212,
      "grad_norm": 1.0478528868072308,
      "learning_rate": 0.00027495,
      "loss": 2.0938,
      "step": 1833
    },
    {
      "epoch": 1.2145695364238411,
      "grad_norm": 1.114171507747835,
      "learning_rate": 0.00027509999999999996,
      "loss": 2.6562,
      "step": 1834
    },
    {
      "epoch": 1.2152317880794703,
      "grad_norm": 1.4247476459311124,
      "learning_rate": 0.00027525,
      "loss": 2.5938,
      "step": 1835
    },
    {
      "epoch": 1.2158940397350992,
      "grad_norm": 1.0829716627580173,
      "learning_rate": 0.00027539999999999997,
      "loss": 2.5625,
      "step": 1836
    },
    {
      "epoch": 1.2165562913907284,
      "grad_norm": 1.1803416416483359,
      "learning_rate": 0.00027554999999999995,
      "loss": 2.625,
      "step": 1837
    },
    {
      "epoch": 1.2172185430463576,
      "grad_norm": 1.0615957829610734,
      "learning_rate": 0.0002757,
      "loss": 2.6094,
      "step": 1838
    },
    {
      "epoch": 1.2178807947019867,
      "grad_norm": 1.0915048001427294,
      "learning_rate": 0.00027584999999999996,
      "loss": 2.7031,
      "step": 1839
    },
    {
      "epoch": 1.218543046357616,
      "grad_norm": 1.0812911145445732,
      "learning_rate": 0.000276,
      "loss": 2.5156,
      "step": 1840
    },
    {
      "epoch": 1.219205298013245,
      "grad_norm": 1.1531567311431212,
      "learning_rate": 0.00027614999999999996,
      "loss": 2.5469,
      "step": 1841
    },
    {
      "epoch": 1.2198675496688742,
      "grad_norm": 1.1680675093906006,
      "learning_rate": 0.0002763,
      "loss": 2.6875,
      "step": 1842
    },
    {
      "epoch": 1.2205298013245034,
      "grad_norm": 1.0898100247359879,
      "learning_rate": 0.00027644999999999997,
      "loss": 2.0312,
      "step": 1843
    },
    {
      "epoch": 1.2211920529801326,
      "grad_norm": 1.040513833175346,
      "learning_rate": 0.0002766,
      "loss": 2.5156,
      "step": 1844
    },
    {
      "epoch": 1.2218543046357615,
      "grad_norm": 1.103313297572885,
      "learning_rate": 0.00027675,
      "loss": 2.75,
      "step": 1845
    },
    {
      "epoch": 1.2225165562913907,
      "grad_norm": 1.174291331822553,
      "learning_rate": 0.0002769,
      "loss": 2.75,
      "step": 1846
    },
    {
      "epoch": 1.2231788079470198,
      "grad_norm": 1.101086799159118,
      "learning_rate": 0.00027705,
      "loss": 2.1094,
      "step": 1847
    },
    {
      "epoch": 1.223841059602649,
      "grad_norm": 1.076112265393698,
      "learning_rate": 0.0002772,
      "loss": 2.1562,
      "step": 1848
    },
    {
      "epoch": 1.2245033112582782,
      "grad_norm": 1.1576205632588799,
      "learning_rate": 0.00027735,
      "loss": 2.6562,
      "step": 1849
    },
    {
      "epoch": 1.2251655629139073,
      "grad_norm": 1.0103135757337511,
      "learning_rate": 0.00027749999999999997,
      "loss": 2.5938,
      "step": 1850
    },
    {
      "epoch": 1.2258278145695365,
      "grad_norm": 1.2025434183382095,
      "learning_rate": 0.00027764999999999995,
      "loss": 2.6719,
      "step": 1851
    },
    {
      "epoch": 1.2264900662251657,
      "grad_norm": 1.2376545000528463,
      "learning_rate": 0.0002778,
      "loss": 2.5469,
      "step": 1852
    },
    {
      "epoch": 1.2271523178807948,
      "grad_norm": 1.1023399903362123,
      "learning_rate": 0.00027794999999999995,
      "loss": 2.8125,
      "step": 1853
    },
    {
      "epoch": 1.2278145695364238,
      "grad_norm": 1.1365387951273755,
      "learning_rate": 0.0002781,
      "loss": 2.5625,
      "step": 1854
    },
    {
      "epoch": 1.228476821192053,
      "grad_norm": 1.1832089637703747,
      "learning_rate": 0.00027824999999999996,
      "loss": 2.6562,
      "step": 1855
    },
    {
      "epoch": 1.229139072847682,
      "grad_norm": 1.1236279296599083,
      "learning_rate": 0.0002784,
      "loss": 2.5,
      "step": 1856
    },
    {
      "epoch": 1.2298013245033113,
      "grad_norm": 1.0810551324569708,
      "learning_rate": 0.00027854999999999997,
      "loss": 2.2031,
      "step": 1857
    },
    {
      "epoch": 1.2304635761589404,
      "grad_norm": 1.0537838403223088,
      "learning_rate": 0.0002787,
      "loss": 2.5469,
      "step": 1858
    },
    {
      "epoch": 1.2311258278145696,
      "grad_norm": 1.0704641850473486,
      "learning_rate": 0.00027885,
      "loss": 2.5156,
      "step": 1859
    },
    {
      "epoch": 1.2317880794701987,
      "grad_norm": 1.5470857420778297,
      "learning_rate": 0.000279,
      "loss": 2.6875,
      "step": 1860
    },
    {
      "epoch": 1.2324503311258277,
      "grad_norm": 1.0171333371766393,
      "learning_rate": 0.00027915,
      "loss": 2.0625,
      "step": 1861
    },
    {
      "epoch": 1.233112582781457,
      "grad_norm": 1.0813066921819676,
      "learning_rate": 0.0002793,
      "loss": 2.5938,
      "step": 1862
    },
    {
      "epoch": 1.233774834437086,
      "grad_norm": 1.2930044956571096,
      "learning_rate": 0.00027945,
      "loss": 3.0312,
      "step": 1863
    },
    {
      "epoch": 1.2344370860927152,
      "grad_norm": 1.1085237276168365,
      "learning_rate": 0.00027959999999999997,
      "loss": 2.6406,
      "step": 1864
    },
    {
      "epoch": 1.2350993377483444,
      "grad_norm": 1.0685194282931558,
      "learning_rate": 0.00027975,
      "loss": 2.6719,
      "step": 1865
    },
    {
      "epoch": 1.2357615894039735,
      "grad_norm": 1.1148534327997999,
      "learning_rate": 0.0002799,
      "loss": 2.375,
      "step": 1866
    },
    {
      "epoch": 1.2364238410596027,
      "grad_norm": 1.0274912984756877,
      "learning_rate": 0.00028004999999999995,
      "loss": 2.75,
      "step": 1867
    },
    {
      "epoch": 1.2370860927152318,
      "grad_norm": 1.0478849488062358,
      "learning_rate": 0.0002802,
      "loss": 2.7656,
      "step": 1868
    },
    {
      "epoch": 1.237748344370861,
      "grad_norm": 0.9550521220643732,
      "learning_rate": 0.00028034999999999996,
      "loss": 2.5625,
      "step": 1869
    },
    {
      "epoch": 1.23841059602649,
      "grad_norm": 1.0309101286007627,
      "learning_rate": 0.0002805,
      "loss": 2.5781,
      "step": 1870
    },
    {
      "epoch": 1.2390728476821191,
      "grad_norm": 1.0792114337977892,
      "learning_rate": 0.00028064999999999996,
      "loss": 2.5781,
      "step": 1871
    },
    {
      "epoch": 1.2397350993377483,
      "grad_norm": 1.1951294477952854,
      "learning_rate": 0.0002808,
      "loss": 2.8438,
      "step": 1872
    },
    {
      "epoch": 1.2403973509933774,
      "grad_norm": 1.029946520128323,
      "learning_rate": 0.00028094999999999997,
      "loss": 2.6562,
      "step": 1873
    },
    {
      "epoch": 1.2410596026490066,
      "grad_norm": 0.9641967429063424,
      "learning_rate": 0.0002811,
      "loss": 2.3438,
      "step": 1874
    },
    {
      "epoch": 1.2417218543046358,
      "grad_norm": 1.162198009973546,
      "learning_rate": 0.00028125,
      "loss": 2.7656,
      "step": 1875
    },
    {
      "epoch": 1.242384105960265,
      "grad_norm": 1.0442571370476585,
      "learning_rate": 0.00028139999999999996,
      "loss": 2.5156,
      "step": 1876
    },
    {
      "epoch": 1.243046357615894,
      "grad_norm": 1.0398861604629954,
      "learning_rate": 0.00028155,
      "loss": 2.5,
      "step": 1877
    },
    {
      "epoch": 1.2437086092715233,
      "grad_norm": 0.9622858880520134,
      "learning_rate": 0.00028169999999999996,
      "loss": 2.1875,
      "step": 1878
    },
    {
      "epoch": 1.2443708609271522,
      "grad_norm": 1.0660310584231711,
      "learning_rate": 0.00028185,
      "loss": 2.6719,
      "step": 1879
    },
    {
      "epoch": 1.2450331125827814,
      "grad_norm": 1.0188204358198052,
      "learning_rate": 0.00028199999999999997,
      "loss": 2.4375,
      "step": 1880
    },
    {
      "epoch": 1.2456953642384105,
      "grad_norm": 1.1510881996047657,
      "learning_rate": 0.00028215,
      "loss": 2.5156,
      "step": 1881
    },
    {
      "epoch": 1.2463576158940397,
      "grad_norm": 1.0173357767415563,
      "learning_rate": 0.0002823,
      "loss": 2.6875,
      "step": 1882
    },
    {
      "epoch": 1.2470198675496689,
      "grad_norm": 1.1103666033790824,
      "learning_rate": 0.00028244999999999995,
      "loss": 2.8281,
      "step": 1883
    },
    {
      "epoch": 1.247682119205298,
      "grad_norm": 1.051364719679405,
      "learning_rate": 0.0002826,
      "loss": 2.5938,
      "step": 1884
    },
    {
      "epoch": 1.2483443708609272,
      "grad_norm": 0.9689593068921656,
      "learning_rate": 0.00028274999999999996,
      "loss": 2.4844,
      "step": 1885
    },
    {
      "epoch": 1.2490066225165564,
      "grad_norm": 1.0351034112023918,
      "learning_rate": 0.00028289999999999994,
      "loss": 2.8906,
      "step": 1886
    },
    {
      "epoch": 1.2496688741721855,
      "grad_norm": 1.0992985825572599,
      "learning_rate": 0.00028304999999999997,
      "loss": 1.9766,
      "step": 1887
    },
    {
      "epoch": 1.2503311258278145,
      "grad_norm": 1.0478148931367806,
      "learning_rate": 0.00028319999999999994,
      "loss": 2.5625,
      "step": 1888
    },
    {
      "epoch": 1.2509933774834436,
      "grad_norm": 1.0780593475405176,
      "learning_rate": 0.00028335,
      "loss": 2.7656,
      "step": 1889
    },
    {
      "epoch": 1.2516556291390728,
      "grad_norm": 0.9860731831857292,
      "learning_rate": 0.00028349999999999995,
      "loss": 2.5312,
      "step": 1890
    },
    {
      "epoch": 1.252317880794702,
      "grad_norm": 0.9696993325504033,
      "learning_rate": 0.00028365,
      "loss": 2.1562,
      "step": 1891
    },
    {
      "epoch": 1.2529801324503311,
      "grad_norm": 1.077802281550881,
      "learning_rate": 0.00028379999999999996,
      "loss": 2.8281,
      "step": 1892
    },
    {
      "epoch": 1.2536423841059603,
      "grad_norm": 1.0798474477816749,
      "learning_rate": 0.00028395,
      "loss": 2.625,
      "step": 1893
    },
    {
      "epoch": 1.2543046357615895,
      "grad_norm": 1.1323584111272793,
      "learning_rate": 0.00028409999999999997,
      "loss": 2.7656,
      "step": 1894
    },
    {
      "epoch": 1.2549668874172186,
      "grad_norm": 1.114729870633954,
      "learning_rate": 0.00028425,
      "loss": 2.4062,
      "step": 1895
    },
    {
      "epoch": 1.2556291390728478,
      "grad_norm": 1.2209638086839982,
      "learning_rate": 0.0002844,
      "loss": 2.0781,
      "step": 1896
    },
    {
      "epoch": 1.2562913907284767,
      "grad_norm": 1.3049261131915915,
      "learning_rate": 0.00028455,
      "loss": 2.7812,
      "step": 1897
    },
    {
      "epoch": 1.256953642384106,
      "grad_norm": 1.0604805948446407,
      "learning_rate": 0.0002847,
      "loss": 2.5781,
      "step": 1898
    },
    {
      "epoch": 1.257615894039735,
      "grad_norm": 1.1194736129673266,
      "learning_rate": 0.00028484999999999996,
      "loss": 2.6719,
      "step": 1899
    },
    {
      "epoch": 1.2582781456953642,
      "grad_norm": 1.1360705332281893,
      "learning_rate": 0.000285,
      "loss": 2.6406,
      "step": 1900
    },
    {
      "epoch": 1.2589403973509934,
      "grad_norm": 1.168718456941622,
      "learning_rate": 0.00028514999999999997,
      "loss": 2.5625,
      "step": 1901
    },
    {
      "epoch": 1.2596026490066226,
      "grad_norm": 1.0200250638359925,
      "learning_rate": 0.00028529999999999994,
      "loss": 2.5469,
      "step": 1902
    },
    {
      "epoch": 1.2602649006622517,
      "grad_norm": 1.1208829067947539,
      "learning_rate": 0.00028544999999999997,
      "loss": 2.1875,
      "step": 1903
    },
    {
      "epoch": 1.2609271523178807,
      "grad_norm": 1.0525553334609878,
      "learning_rate": 0.00028559999999999995,
      "loss": 2.6875,
      "step": 1904
    },
    {
      "epoch": 1.26158940397351,
      "grad_norm": 1.0604350845208788,
      "learning_rate": 0.00028575,
      "loss": 2.5312,
      "step": 1905
    },
    {
      "epoch": 1.262251655629139,
      "grad_norm": 0.992399333560451,
      "learning_rate": 0.00028589999999999996,
      "loss": 2.0938,
      "step": 1906
    },
    {
      "epoch": 1.2629139072847682,
      "grad_norm": 1.0111889050599285,
      "learning_rate": 0.00028605,
      "loss": 2.5938,
      "step": 1907
    },
    {
      "epoch": 1.2635761589403973,
      "grad_norm": 0.9681452529335135,
      "learning_rate": 0.00028619999999999996,
      "loss": 2.2031,
      "step": 1908
    },
    {
      "epoch": 1.2642384105960265,
      "grad_norm": 1.1407885430181062,
      "learning_rate": 0.00028635,
      "loss": 2.7031,
      "step": 1909
    },
    {
      "epoch": 1.2649006622516556,
      "grad_norm": 1.118260853278201,
      "learning_rate": 0.00028649999999999997,
      "loss": 2.7188,
      "step": 1910
    },
    {
      "epoch": 1.2655629139072848,
      "grad_norm": 1.055249747682395,
      "learning_rate": 0.00028665,
      "loss": 2.1875,
      "step": 1911
    },
    {
      "epoch": 1.266225165562914,
      "grad_norm": 1.1026575482125525,
      "learning_rate": 0.0002868,
      "loss": 2.4219,
      "step": 1912
    },
    {
      "epoch": 1.266887417218543,
      "grad_norm": 1.1169894504203735,
      "learning_rate": 0.00028694999999999995,
      "loss": 2.4219,
      "step": 1913
    },
    {
      "epoch": 1.2675496688741723,
      "grad_norm": 0.9843326738823613,
      "learning_rate": 0.0002871,
      "loss": 2.1562,
      "step": 1914
    },
    {
      "epoch": 1.2682119205298013,
      "grad_norm": 1.2241118134432818,
      "learning_rate": 0.00028724999999999996,
      "loss": 3.0469,
      "step": 1915
    },
    {
      "epoch": 1.2688741721854304,
      "grad_norm": 1.1400944025068844,
      "learning_rate": 0.00028739999999999994,
      "loss": 3.0,
      "step": 1916
    },
    {
      "epoch": 1.2695364238410596,
      "grad_norm": 1.100338586503263,
      "learning_rate": 0.00028754999999999997,
      "loss": 2.6875,
      "step": 1917
    },
    {
      "epoch": 1.2701986754966887,
      "grad_norm": 1.0469257858662373,
      "learning_rate": 0.00028769999999999995,
      "loss": 2.3281,
      "step": 1918
    },
    {
      "epoch": 1.270860927152318,
      "grad_norm": 1.155515575239003,
      "learning_rate": 0.00028785,
      "loss": 2.7344,
      "step": 1919
    },
    {
      "epoch": 1.271523178807947,
      "grad_norm": 1.0409552474106014,
      "learning_rate": 0.00028799999999999995,
      "loss": 2.0938,
      "step": 1920
    },
    {
      "epoch": 1.2721854304635762,
      "grad_norm": 1.03696624615095,
      "learning_rate": 0.00028815,
      "loss": 2.5,
      "step": 1921
    },
    {
      "epoch": 1.2728476821192052,
      "grad_norm": 1.1270970282997816,
      "learning_rate": 0.00028829999999999996,
      "loss": 2.1719,
      "step": 1922
    },
    {
      "epoch": 1.2735099337748346,
      "grad_norm": 1.2352258964071394,
      "learning_rate": 0.00028845,
      "loss": 2.7344,
      "step": 1923
    },
    {
      "epoch": 1.2741721854304635,
      "grad_norm": 1.1379863833456227,
      "learning_rate": 0.00028859999999999997,
      "loss": 2.4531,
      "step": 1924
    },
    {
      "epoch": 1.2748344370860927,
      "grad_norm": 1.3774584956460922,
      "learning_rate": 0.00028875,
      "loss": 2.2031,
      "step": 1925
    },
    {
      "epoch": 1.2754966887417218,
      "grad_norm": 1.0019107181300786,
      "learning_rate": 0.0002889,
      "loss": 2.2031,
      "step": 1926
    },
    {
      "epoch": 1.276158940397351,
      "grad_norm": 1.1805687541288026,
      "learning_rate": 0.00028905,
      "loss": 2.6562,
      "step": 1927
    },
    {
      "epoch": 1.2768211920529802,
      "grad_norm": 1.0791646104057475,
      "learning_rate": 0.0002892,
      "loss": 2.5625,
      "step": 1928
    },
    {
      "epoch": 1.2774834437086093,
      "grad_norm": 1.0959949474657371,
      "learning_rate": 0.00028934999999999996,
      "loss": 2.0625,
      "step": 1929
    },
    {
      "epoch": 1.2781456953642385,
      "grad_norm": 1.0354366485198145,
      "learning_rate": 0.0002895,
      "loss": 2.0938,
      "step": 1930
    },
    {
      "epoch": 1.2788079470198674,
      "grad_norm": 1.1501986369387915,
      "learning_rate": 0.00028964999999999997,
      "loss": 2.5,
      "step": 1931
    },
    {
      "epoch": 1.2794701986754966,
      "grad_norm": 1.0645195298901111,
      "learning_rate": 0.00028979999999999994,
      "loss": 2.5469,
      "step": 1932
    },
    {
      "epoch": 1.2801324503311258,
      "grad_norm": 1.2568649516174393,
      "learning_rate": 0.00028995,
      "loss": 2.6562,
      "step": 1933
    },
    {
      "epoch": 1.280794701986755,
      "grad_norm": 1.0184071562805015,
      "learning_rate": 0.00029009999999999995,
      "loss": 2.5781,
      "step": 1934
    },
    {
      "epoch": 1.281456953642384,
      "grad_norm": 0.9511248085741806,
      "learning_rate": 0.00029025,
      "loss": 2.5469,
      "step": 1935
    },
    {
      "epoch": 1.2821192052980133,
      "grad_norm": 1.0350583274553433,
      "learning_rate": 0.00029039999999999996,
      "loss": 2.4531,
      "step": 1936
    },
    {
      "epoch": 1.2827814569536424,
      "grad_norm": 1.1329353201398,
      "learning_rate": 0.00029055,
      "loss": 2.7031,
      "step": 1937
    },
    {
      "epoch": 1.2834437086092716,
      "grad_norm": 1.0894534422556255,
      "learning_rate": 0.00029069999999999996,
      "loss": 2.5938,
      "step": 1938
    },
    {
      "epoch": 1.2841059602649008,
      "grad_norm": 1.0967781765222333,
      "learning_rate": 0.00029085,
      "loss": 2.6719,
      "step": 1939
    },
    {
      "epoch": 1.2847682119205297,
      "grad_norm": 1.2096398116942162,
      "learning_rate": 0.00029099999999999997,
      "loss": 2.375,
      "step": 1940
    },
    {
      "epoch": 1.2854304635761589,
      "grad_norm": 1.036185112772001,
      "learning_rate": 0.00029115,
      "loss": 2.6719,
      "step": 1941
    },
    {
      "epoch": 1.286092715231788,
      "grad_norm": 1.0788777078109564,
      "learning_rate": 0.0002913,
      "loss": 2.6406,
      "step": 1942
    },
    {
      "epoch": 1.2867549668874172,
      "grad_norm": 1.2141745514818192,
      "learning_rate": 0.00029145,
      "loss": 2.6562,
      "step": 1943
    },
    {
      "epoch": 1.2874172185430464,
      "grad_norm": 1.1644640957341408,
      "learning_rate": 0.0002916,
      "loss": 2.5938,
      "step": 1944
    },
    {
      "epoch": 1.2880794701986755,
      "grad_norm": 1.0881825031520775,
      "learning_rate": 0.00029174999999999996,
      "loss": 2.7969,
      "step": 1945
    },
    {
      "epoch": 1.2887417218543047,
      "grad_norm": 1.1861048268415086,
      "learning_rate": 0.0002919,
      "loss": 2.5781,
      "step": 1946
    },
    {
      "epoch": 1.2894039735099339,
      "grad_norm": 1.126688345733708,
      "learning_rate": 0.00029204999999999997,
      "loss": 2.5,
      "step": 1947
    },
    {
      "epoch": 1.290066225165563,
      "grad_norm": 0.9650983365409912,
      "learning_rate": 0.00029219999999999995,
      "loss": 2.5469,
      "step": 1948
    },
    {
      "epoch": 1.290728476821192,
      "grad_norm": 1.1985797671040255,
      "learning_rate": 0.00029235,
      "loss": 2.2031,
      "step": 1949
    },
    {
      "epoch": 1.2913907284768211,
      "grad_norm": 1.3279822585227978,
      "learning_rate": 0.00029249999999999995,
      "loss": 2.8281,
      "step": 1950
    },
    {
      "epoch": 1.2920529801324503,
      "grad_norm": 1.109824070413775,
      "learning_rate": 0.00029265,
      "loss": 2.7969,
      "step": 1951
    },
    {
      "epoch": 1.2927152317880795,
      "grad_norm": 1.2355286789036353,
      "learning_rate": 0.00029279999999999996,
      "loss": 2.5781,
      "step": 1952
    },
    {
      "epoch": 1.2933774834437086,
      "grad_norm": 0.9902368225212206,
      "learning_rate": 0.00029295,
      "loss": 2.5469,
      "step": 1953
    },
    {
      "epoch": 1.2940397350993378,
      "grad_norm": 1.2500663286055769,
      "learning_rate": 0.00029309999999999997,
      "loss": 2.8281,
      "step": 1954
    },
    {
      "epoch": 1.294701986754967,
      "grad_norm": 1.1390262695226614,
      "learning_rate": 0.00029325,
      "loss": 2.6719,
      "step": 1955
    },
    {
      "epoch": 1.295364238410596,
      "grad_norm": 1.0569373115836096,
      "learning_rate": 0.0002934,
      "loss": 2.6094,
      "step": 1956
    },
    {
      "epoch": 1.2960264900662253,
      "grad_norm": 1.121002100150029,
      "learning_rate": 0.00029355,
      "loss": 2.6094,
      "step": 1957
    },
    {
      "epoch": 1.2966887417218542,
      "grad_norm": 1.1086242686110153,
      "learning_rate": 0.0002937,
      "loss": 2.5469,
      "step": 1958
    },
    {
      "epoch": 1.2973509933774834,
      "grad_norm": 1.3535454084609597,
      "learning_rate": 0.00029384999999999996,
      "loss": 2.7188,
      "step": 1959
    },
    {
      "epoch": 1.2980132450331126,
      "grad_norm": 39.181124657694724,
      "learning_rate": 0.000294,
      "loss": 3.1875,
      "step": 1960
    },
    {
      "epoch": 1.2986754966887417,
      "grad_norm": 0.9259981151085336,
      "learning_rate": 0.00029414999999999997,
      "loss": 2.4844,
      "step": 1961
    },
    {
      "epoch": 1.2993377483443709,
      "grad_norm": 1.0720675503395423,
      "learning_rate": 0.00029429999999999994,
      "loss": 2.8125,
      "step": 1962
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9875026945065355,
      "learning_rate": 0.00029445,
      "loss": 2.3281,
      "step": 1963
    },
    {
      "epoch": 1.3006622516556292,
      "grad_norm": 1.0922421253948331,
      "learning_rate": 0.00029459999999999995,
      "loss": 2.625,
      "step": 1964
    },
    {
      "epoch": 1.3013245033112582,
      "grad_norm": 1.116007377266788,
      "learning_rate": 0.00029475,
      "loss": 2.4062,
      "step": 1965
    },
    {
      "epoch": 1.3019867549668875,
      "grad_norm": 1.0548331634043413,
      "learning_rate": 0.00029489999999999996,
      "loss": 2.5,
      "step": 1966
    },
    {
      "epoch": 1.3026490066225165,
      "grad_norm": 0.9782879048981354,
      "learning_rate": 0.00029505,
      "loss": 2.4062,
      "step": 1967
    },
    {
      "epoch": 1.3033112582781456,
      "grad_norm": 1.056137184337501,
      "learning_rate": 0.00029519999999999997,
      "loss": 2.5938,
      "step": 1968
    },
    {
      "epoch": 1.3039735099337748,
      "grad_norm": 1.0941836673266787,
      "learning_rate": 0.00029535,
      "loss": 2.6562,
      "step": 1969
    },
    {
      "epoch": 1.304635761589404,
      "grad_norm": 1.1532094349044157,
      "learning_rate": 0.00029549999999999997,
      "loss": 2.6406,
      "step": 1970
    },
    {
      "epoch": 1.3052980132450331,
      "grad_norm": 1.0131475739399138,
      "learning_rate": 0.00029565,
      "loss": 2.1875,
      "step": 1971
    },
    {
      "epoch": 1.3059602649006623,
      "grad_norm": 1.147507884991115,
      "learning_rate": 0.0002958,
      "loss": 2.7188,
      "step": 1972
    },
    {
      "epoch": 1.3066225165562915,
      "grad_norm": 1.1045130325774761,
      "learning_rate": 0.00029595,
      "loss": 2.6094,
      "step": 1973
    },
    {
      "epoch": 1.3072847682119204,
      "grad_norm": 1.0895706669188263,
      "learning_rate": 0.0002961,
      "loss": 2.5938,
      "step": 1974
    },
    {
      "epoch": 1.3079470198675498,
      "grad_norm": 1.0713927376251628,
      "learning_rate": 0.00029624999999999996,
      "loss": 2.6875,
      "step": 1975
    },
    {
      "epoch": 1.3086092715231787,
      "grad_norm": 1.0336689741676732,
      "learning_rate": 0.0002964,
      "loss": 2.4844,
      "step": 1976
    },
    {
      "epoch": 1.309271523178808,
      "grad_norm": 1.0818326941357388,
      "learning_rate": 0.00029654999999999997,
      "loss": 2.2188,
      "step": 1977
    },
    {
      "epoch": 1.309933774834437,
      "grad_norm": 1.1595925386010197,
      "learning_rate": 0.00029669999999999995,
      "loss": 2.4375,
      "step": 1978
    },
    {
      "epoch": 1.3105960264900662,
      "grad_norm": 1.0076444919530783,
      "learning_rate": 0.00029685,
      "loss": 2.5469,
      "step": 1979
    },
    {
      "epoch": 1.3112582781456954,
      "grad_norm": 1.1021443681044725,
      "learning_rate": 0.00029699999999999996,
      "loss": 2.7188,
      "step": 1980
    },
    {
      "epoch": 1.3119205298013246,
      "grad_norm": 1.2742319205681256,
      "learning_rate": 0.00029715,
      "loss": 2.125,
      "step": 1981
    },
    {
      "epoch": 1.3125827814569537,
      "grad_norm": 1.1548730432150276,
      "learning_rate": 0.00029729999999999996,
      "loss": 2.4688,
      "step": 1982
    },
    {
      "epoch": 1.3132450331125827,
      "grad_norm": 1.0808407989829563,
      "learning_rate": 0.00029745,
      "loss": 2.1719,
      "step": 1983
    },
    {
      "epoch": 1.313907284768212,
      "grad_norm": 2.6683188367977073,
      "learning_rate": 0.00029759999999999997,
      "loss": 2.7969,
      "step": 1984
    },
    {
      "epoch": 1.314569536423841,
      "grad_norm": 1.1339643271225937,
      "learning_rate": 0.00029775,
      "loss": 2.4531,
      "step": 1985
    },
    {
      "epoch": 1.3152317880794702,
      "grad_norm": 1.1007110203489667,
      "learning_rate": 0.0002979,
      "loss": 2.4219,
      "step": 1986
    },
    {
      "epoch": 1.3158940397350993,
      "grad_norm": 1.050090083826493,
      "learning_rate": 0.00029805,
      "loss": 2.4688,
      "step": 1987
    },
    {
      "epoch": 1.3165562913907285,
      "grad_norm": 1.0532719772266244,
      "learning_rate": 0.0002982,
      "loss": 2.375,
      "step": 1988
    },
    {
      "epoch": 1.3172185430463577,
      "grad_norm": 1.0148735207987534,
      "learning_rate": 0.00029835,
      "loss": 2.5,
      "step": 1989
    },
    {
      "epoch": 1.3178807947019868,
      "grad_norm": 1.3468871824253617,
      "learning_rate": 0.0002985,
      "loss": 2.5156,
      "step": 1990
    },
    {
      "epoch": 1.318543046357616,
      "grad_norm": 1.0513339303068943,
      "learning_rate": 0.00029864999999999997,
      "loss": 2.4219,
      "step": 1991
    },
    {
      "epoch": 1.319205298013245,
      "grad_norm": 1.1553184355427268,
      "learning_rate": 0.0002988,
      "loss": 2.75,
      "step": 1992
    },
    {
      "epoch": 1.319867549668874,
      "grad_norm": 1.1858219448525618,
      "learning_rate": 0.00029895,
      "loss": 2.7812,
      "step": 1993
    },
    {
      "epoch": 1.3205298013245033,
      "grad_norm": 1.093980672612447,
      "learning_rate": 0.00029909999999999995,
      "loss": 1.9844,
      "step": 1994
    },
    {
      "epoch": 1.3211920529801324,
      "grad_norm": 1.1947924271549226,
      "learning_rate": 0.00029925,
      "loss": 2.5938,
      "step": 1995
    },
    {
      "epoch": 1.3218543046357616,
      "grad_norm": 1.0875267890028453,
      "learning_rate": 0.00029939999999999996,
      "loss": 2.5625,
      "step": 1996
    },
    {
      "epoch": 1.3225165562913908,
      "grad_norm": 1.2949422085707263,
      "learning_rate": 0.00029955,
      "loss": 2.6406,
      "step": 1997
    },
    {
      "epoch": 1.32317880794702,
      "grad_norm": 1.0739303521176613,
      "learning_rate": 0.00029969999999999997,
      "loss": 2.5938,
      "step": 1998
    },
    {
      "epoch": 1.323841059602649,
      "grad_norm": 1.1677531520886069,
      "learning_rate": 0.00029985,
      "loss": 2.4531,
      "step": 1999
    },
    {
      "epoch": 1.3245033112582782,
      "grad_norm": 0.9897328569294896,
      "learning_rate": 0.0003,
      "loss": 2.4688,
      "step": 2000
    },
    {
      "epoch": 1.3251655629139072,
      "grad_norm": 0.9106212163888883,
      "learning_rate": 0.0002999999851491401,
      "loss": 2.3438,
      "step": 2001
    },
    {
      "epoch": 1.3258278145695364,
      "grad_norm": 1.054385263828949,
      "learning_rate": 0.00029999994059656354,
      "loss": 2.6094,
      "step": 2002
    },
    {
      "epoch": 1.3264900662251655,
      "grad_norm": 0.9827580505608354,
      "learning_rate": 0.000299999866342279,
      "loss": 2.5156,
      "step": 2003
    },
    {
      "epoch": 1.3271523178807947,
      "grad_norm": 1.0135124480904556,
      "learning_rate": 0.00029999976238630124,
      "loss": 2.375,
      "step": 2004
    },
    {
      "epoch": 1.3278145695364238,
      "grad_norm": 1.0921632137293598,
      "learning_rate": 0.00029999962872865086,
      "loss": 2.4531,
      "step": 2005
    },
    {
      "epoch": 1.328476821192053,
      "grad_norm": 1.0284942774506889,
      "learning_rate": 0.00029999946536935423,
      "loss": 2.5469,
      "step": 2006
    },
    {
      "epoch": 1.3291390728476822,
      "grad_norm": 0.9660339299106807,
      "learning_rate": 0.0002999992723084438,
      "loss": 2.7188,
      "step": 2007
    },
    {
      "epoch": 1.3298013245033111,
      "grad_norm": 1.1351336909668122,
      "learning_rate": 0.00029999904954595784,
      "loss": 2.7969,
      "step": 2008
    },
    {
      "epoch": 1.3304635761589405,
      "grad_norm": 1.2793949440623427,
      "learning_rate": 0.0002999987970819403,
      "loss": 2.75,
      "step": 2009
    },
    {
      "epoch": 1.3311258278145695,
      "grad_norm": 1.0471921414360634,
      "learning_rate": 0.0002999985149164413,
      "loss": 2.0156,
      "step": 2010
    },
    {
      "epoch": 1.3317880794701986,
      "grad_norm": 0.9722668965549524,
      "learning_rate": 0.00029999820304951665,
      "loss": 2.5781,
      "step": 2011
    },
    {
      "epoch": 1.3324503311258278,
      "grad_norm": 1.1012710706926796,
      "learning_rate": 0.00029999786148122813,
      "loss": 2.5312,
      "step": 2012
    },
    {
      "epoch": 1.333112582781457,
      "grad_norm": 1.0511107087686171,
      "learning_rate": 0.00029999749021164334,
      "loss": 2.5156,
      "step": 2013
    },
    {
      "epoch": 1.333774834437086,
      "grad_norm": 1.0997450493383003,
      "learning_rate": 0.0002999970892408359,
      "loss": 2.3438,
      "step": 2014
    },
    {
      "epoch": 1.3344370860927153,
      "grad_norm": 0.9934301707366604,
      "learning_rate": 0.0002999966585688851,
      "loss": 2.4062,
      "step": 2015
    },
    {
      "epoch": 1.3350993377483444,
      "grad_norm": 1.0130746712865908,
      "learning_rate": 0.0002999961981958762,
      "loss": 2.6562,
      "step": 2016
    },
    {
      "epoch": 1.3357615894039734,
      "grad_norm": 0.979423502669621,
      "learning_rate": 0.0002999957081219005,
      "loss": 2.3906,
      "step": 2017
    },
    {
      "epoch": 1.3364238410596028,
      "grad_norm": 0.977531539824706,
      "learning_rate": 0.0002999951883470548,
      "loss": 2.5625,
      "step": 2018
    },
    {
      "epoch": 1.3370860927152317,
      "grad_norm": 1.0364533471056445,
      "learning_rate": 0.0002999946388714423,
      "loss": 2.4219,
      "step": 2019
    },
    {
      "epoch": 1.3377483443708609,
      "grad_norm": 1.0808251568241538,
      "learning_rate": 0.0002999940596951716,
      "loss": 2.5938,
      "step": 2020
    },
    {
      "epoch": 1.33841059602649,
      "grad_norm": 1.1719723977286878,
      "learning_rate": 0.0002999934508183575,
      "loss": 2.75,
      "step": 2021
    },
    {
      "epoch": 1.3390728476821192,
      "grad_norm": 1.1025455427458621,
      "learning_rate": 0.0002999928122411205,
      "loss": 2.0312,
      "step": 2022
    },
    {
      "epoch": 1.3397350993377484,
      "grad_norm": 0.9910353888871356,
      "learning_rate": 0.00029999214396358705,
      "loss": 2.0156,
      "step": 2023
    },
    {
      "epoch": 1.3403973509933775,
      "grad_norm": 0.9977347497991871,
      "learning_rate": 0.0002999914459858895,
      "loss": 2.5469,
      "step": 2024
    },
    {
      "epoch": 1.3410596026490067,
      "grad_norm": 1.0804623662503163,
      "learning_rate": 0.0002999907183081661,
      "loss": 2.5938,
      "step": 2025
    },
    {
      "epoch": 1.3417218543046356,
      "grad_norm": 1.40944556666985,
      "learning_rate": 0.0002999899609305608,
      "loss": 2.7031,
      "step": 2026
    },
    {
      "epoch": 1.342384105960265,
      "grad_norm": 0.9921557939340224,
      "learning_rate": 0.00029998917385322366,
      "loss": 2.5781,
      "step": 2027
    },
    {
      "epoch": 1.343046357615894,
      "grad_norm": 1.1239899099975554,
      "learning_rate": 0.00029998835707631063,
      "loss": 2.7188,
      "step": 2028
    },
    {
      "epoch": 1.3437086092715231,
      "grad_norm": 1.0518251072319211,
      "learning_rate": 0.00029998751059998323,
      "loss": 2.9062,
      "step": 2029
    },
    {
      "epoch": 1.3443708609271523,
      "grad_norm": 1.0585330315514934,
      "learning_rate": 0.00029998663442440924,
      "loss": 2.3125,
      "step": 2030
    },
    {
      "epoch": 1.3450331125827815,
      "grad_norm": 0.9025844197740192,
      "learning_rate": 0.0002999857285497621,
      "loss": 2.1875,
      "step": 2031
    },
    {
      "epoch": 1.3456953642384106,
      "grad_norm": 1.1210696515198164,
      "learning_rate": 0.00029998479297622114,
      "loss": 2.5625,
      "step": 2032
    },
    {
      "epoch": 1.3463576158940398,
      "grad_norm": 1.0952408376390268,
      "learning_rate": 0.00029998382770397164,
      "loss": 2.625,
      "step": 2033
    },
    {
      "epoch": 1.347019867549669,
      "grad_norm": 1.0601953619614632,
      "learning_rate": 0.0002999828327332048,
      "loss": 2.6562,
      "step": 2034
    },
    {
      "epoch": 1.347682119205298,
      "grad_norm": 1.056257914944039,
      "learning_rate": 0.0002999818080641175,
      "loss": 2.625,
      "step": 2035
    },
    {
      "epoch": 1.3483443708609273,
      "grad_norm": 1.002870581309078,
      "learning_rate": 0.0002999807536969128,
      "loss": 2.5938,
      "step": 2036
    },
    {
      "epoch": 1.3490066225165562,
      "grad_norm": 1.132751865514166,
      "learning_rate": 0.00029997966963179936,
      "loss": 2.5,
      "step": 2037
    },
    {
      "epoch": 1.3496688741721854,
      "grad_norm": 0.9273223497681905,
      "learning_rate": 0.00029997855586899185,
      "loss": 2.2344,
      "step": 2038
    },
    {
      "epoch": 1.3503311258278146,
      "grad_norm": 1.0183091582389598,
      "learning_rate": 0.00029997741240871085,
      "loss": 2.4062,
      "step": 2039
    },
    {
      "epoch": 1.3509933774834437,
      "grad_norm": 0.9648463263310283,
      "learning_rate": 0.0002999762392511828,
      "loss": 2.4688,
      "step": 2040
    },
    {
      "epoch": 1.351655629139073,
      "grad_norm": 1.0817590367134078,
      "learning_rate": 0.0002999750363966399,
      "loss": 2.4375,
      "step": 2041
    },
    {
      "epoch": 1.352317880794702,
      "grad_norm": 1.026529476359781,
      "learning_rate": 0.0002999738038453204,
      "loss": 2.7344,
      "step": 2042
    },
    {
      "epoch": 1.3529801324503312,
      "grad_norm": 0.9650344341618357,
      "learning_rate": 0.0002999725415974684,
      "loss": 2.4688,
      "step": 2043
    },
    {
      "epoch": 1.3536423841059602,
      "grad_norm": 0.9349776003181619,
      "learning_rate": 0.0002999712496533337,
      "loss": 2.375,
      "step": 2044
    },
    {
      "epoch": 1.3543046357615893,
      "grad_norm": 1.122369921650896,
      "learning_rate": 0.0002999699280131723,
      "loss": 2.6406,
      "step": 2045
    },
    {
      "epoch": 1.3549668874172185,
      "grad_norm": 0.9545700947631062,
      "learning_rate": 0.00029996857667724574,
      "loss": 2.3125,
      "step": 2046
    },
    {
      "epoch": 1.3556291390728477,
      "grad_norm": 1.104748855317723,
      "learning_rate": 0.00029996719564582164,
      "loss": 2.7656,
      "step": 2047
    },
    {
      "epoch": 1.3562913907284768,
      "grad_norm": 1.002662558031696,
      "learning_rate": 0.0002999657849191735,
      "loss": 2.375,
      "step": 2048
    },
    {
      "epoch": 1.356953642384106,
      "grad_norm": 1.0260077076481458,
      "learning_rate": 0.0002999643444975807,
      "loss": 2.4219,
      "step": 2049
    },
    {
      "epoch": 1.3576158940397351,
      "grad_norm": 0.9882912082533333,
      "learning_rate": 0.0002999628743813283,
      "loss": 2.1094,
      "step": 2050
    },
    {
      "epoch": 1.3582781456953643,
      "grad_norm": 1.1265108467549234,
      "learning_rate": 0.00029996137457070763,
      "loss": 2.5312,
      "step": 2051
    },
    {
      "epoch": 1.3589403973509935,
      "grad_norm": 1.3069110943498685,
      "learning_rate": 0.00029995984506601546,
      "loss": 2.8438,
      "step": 2052
    },
    {
      "epoch": 1.3596026490066224,
      "grad_norm": 1.1169943828318265,
      "learning_rate": 0.00029995828586755474,
      "loss": 2.5469,
      "step": 2053
    },
    {
      "epoch": 1.3602649006622516,
      "grad_norm": 1.24683348164736,
      "learning_rate": 0.0002999566969756342,
      "loss": 2.3906,
      "step": 2054
    },
    {
      "epoch": 1.3609271523178808,
      "grad_norm": 1.3264826271754002,
      "learning_rate": 0.00029995507839056853,
      "loss": 2.5625,
      "step": 2055
    },
    {
      "epoch": 1.36158940397351,
      "grad_norm": 0.9310865908412236,
      "learning_rate": 0.00029995343011267805,
      "loss": 2.5312,
      "step": 2056
    },
    {
      "epoch": 1.362251655629139,
      "grad_norm": 1.0657039094623508,
      "learning_rate": 0.0002999517521422894,
      "loss": 2.7031,
      "step": 2057
    },
    {
      "epoch": 1.3629139072847682,
      "grad_norm": 1.0238749959680067,
      "learning_rate": 0.00029995004447973457,
      "loss": 2.4375,
      "step": 2058
    },
    {
      "epoch": 1.3635761589403974,
      "grad_norm": 1.156791465048237,
      "learning_rate": 0.00029994830712535176,
      "loss": 2.6719,
      "step": 2059
    },
    {
      "epoch": 1.3642384105960264,
      "grad_norm": 0.9984995218019491,
      "learning_rate": 0.00029994654007948514,
      "loss": 2.3125,
      "step": 2060
    },
    {
      "epoch": 1.3649006622516557,
      "grad_norm": 1.0928268740664213,
      "learning_rate": 0.00029994474334248453,
      "loss": 2.5781,
      "step": 2061
    },
    {
      "epoch": 1.3655629139072847,
      "grad_norm": 1.0765424489806634,
      "learning_rate": 0.0002999429169147056,
      "loss": 2.6875,
      "step": 2062
    },
    {
      "epoch": 1.3662251655629138,
      "grad_norm": 0.9437256024088053,
      "learning_rate": 0.0002999410607965101,
      "loss": 2.4219,
      "step": 2063
    },
    {
      "epoch": 1.366887417218543,
      "grad_norm": 1.0676649742268547,
      "learning_rate": 0.0002999391749882655,
      "loss": 2.2656,
      "step": 2064
    },
    {
      "epoch": 1.3675496688741722,
      "grad_norm": 1.1699558002606203,
      "learning_rate": 0.00029993725949034535,
      "loss": 2.7031,
      "step": 2065
    },
    {
      "epoch": 1.3682119205298013,
      "grad_norm": 1.2963926717576646,
      "learning_rate": 0.0002999353143031288,
      "loss": 2.625,
      "step": 2066
    },
    {
      "epoch": 1.3688741721854305,
      "grad_norm": 0.9424376115189346,
      "learning_rate": 0.00029993333942700105,
      "loss": 2.6094,
      "step": 2067
    },
    {
      "epoch": 1.3695364238410597,
      "grad_norm": 1.0281093363124507,
      "learning_rate": 0.00029993133486235315,
      "loss": 2.6875,
      "step": 2068
    },
    {
      "epoch": 1.3701986754966886,
      "grad_norm": 0.9722275326151685,
      "learning_rate": 0.0002999293006095821,
      "loss": 2.5625,
      "step": 2069
    },
    {
      "epoch": 1.370860927152318,
      "grad_norm": 1.011360433036984,
      "learning_rate": 0.00029992723666909065,
      "loss": 2.75,
      "step": 2070
    },
    {
      "epoch": 1.371523178807947,
      "grad_norm": 1.1275751769331597,
      "learning_rate": 0.0002999251430412874,
      "loss": 2.4375,
      "step": 2071
    },
    {
      "epoch": 1.372185430463576,
      "grad_norm": 1.0648901210761295,
      "learning_rate": 0.000299923019726587,
      "loss": 2.5469,
      "step": 2072
    },
    {
      "epoch": 1.3728476821192053,
      "grad_norm": 1.042670875014918,
      "learning_rate": 0.00029992086672540994,
      "loss": 2.6719,
      "step": 2073
    },
    {
      "epoch": 1.3735099337748344,
      "grad_norm": 1.2108186499996882,
      "learning_rate": 0.0002999186840381825,
      "loss": 2.75,
      "step": 2074
    },
    {
      "epoch": 1.3741721854304636,
      "grad_norm": 1.0190419555836847,
      "learning_rate": 0.0002999164716653368,
      "loss": 2.6562,
      "step": 2075
    },
    {
      "epoch": 1.3748344370860928,
      "grad_norm": 0.9335672354134906,
      "learning_rate": 0.00029991422960731106,
      "loss": 2.375,
      "step": 2076
    },
    {
      "epoch": 1.375496688741722,
      "grad_norm": 0.9950280715301916,
      "learning_rate": 0.00029991195786454905,
      "loss": 2.4844,
      "step": 2077
    },
    {
      "epoch": 1.3761589403973509,
      "grad_norm": 1.0306797800037197,
      "learning_rate": 0.0002999096564375007,
      "loss": 2.625,
      "step": 2078
    },
    {
      "epoch": 1.3768211920529803,
      "grad_norm": 0.9315985494953134,
      "learning_rate": 0.0002999073253266218,
      "loss": 2.5938,
      "step": 2079
    },
    {
      "epoch": 1.3774834437086092,
      "grad_norm": 1.0285396108311404,
      "learning_rate": 0.00029990496453237376,
      "loss": 2.6094,
      "step": 2080
    },
    {
      "epoch": 1.3781456953642384,
      "grad_norm": 1.058219687594806,
      "learning_rate": 0.0002999025740552242,
      "loss": 2.0625,
      "step": 2081
    },
    {
      "epoch": 1.3788079470198675,
      "grad_norm": 1.108192494461978,
      "learning_rate": 0.00029990015389564636,
      "loss": 2.6719,
      "step": 2082
    },
    {
      "epoch": 1.3794701986754967,
      "grad_norm": 0.9928633586645363,
      "learning_rate": 0.00029989770405411956,
      "loss": 2.5625,
      "step": 2083
    },
    {
      "epoch": 1.3801324503311259,
      "grad_norm": 1.026506255561202,
      "learning_rate": 0.00029989522453112877,
      "loss": 2.4219,
      "step": 2084
    },
    {
      "epoch": 1.380794701986755,
      "grad_norm": 1.1097359745458428,
      "learning_rate": 0.000299892715327165,
      "loss": 2.6094,
      "step": 2085
    },
    {
      "epoch": 1.3814569536423842,
      "grad_norm": 1.197333947763301,
      "learning_rate": 0.0002998901764427252,
      "loss": 2.0625,
      "step": 2086
    },
    {
      "epoch": 1.3821192052980131,
      "grad_norm": 0.9703797561813698,
      "learning_rate": 0.00029988760787831207,
      "loss": 2.5,
      "step": 2087
    },
    {
      "epoch": 1.3827814569536425,
      "grad_norm": 1.0829678223300725,
      "learning_rate": 0.0002998850096344341,
      "loss": 2.4844,
      "step": 2088
    },
    {
      "epoch": 1.3834437086092715,
      "grad_norm": 0.9875225963662323,
      "learning_rate": 0.00029988238171160586,
      "loss": 2.5781,
      "step": 2089
    },
    {
      "epoch": 1.3841059602649006,
      "grad_norm": 0.9181551745807098,
      "learning_rate": 0.0002998797241103477,
      "loss": 2.1406,
      "step": 2090
    },
    {
      "epoch": 1.3847682119205298,
      "grad_norm": 1.0793421842263742,
      "learning_rate": 0.0002998770368311859,
      "loss": 2.625,
      "step": 2091
    },
    {
      "epoch": 1.385430463576159,
      "grad_norm": 1.0202697923864872,
      "learning_rate": 0.0002998743198746524,
      "loss": 2.3594,
      "step": 2092
    },
    {
      "epoch": 1.3860927152317881,
      "grad_norm": 1.0085313654720545,
      "learning_rate": 0.00029987157324128544,
      "loss": 2.6719,
      "step": 2093
    },
    {
      "epoch": 1.3867549668874173,
      "grad_norm": 0.9682317825838836,
      "learning_rate": 0.0002998687969316287,
      "loss": 2.5625,
      "step": 2094
    },
    {
      "epoch": 1.3874172185430464,
      "grad_norm": 1.0247628071903343,
      "learning_rate": 0.000299865990946232,
      "loss": 2.6719,
      "step": 2095
    },
    {
      "epoch": 1.3880794701986754,
      "grad_norm": 0.95291044958702,
      "learning_rate": 0.0002998631552856509,
      "loss": 2.4844,
      "step": 2096
    },
    {
      "epoch": 1.3887417218543046,
      "grad_norm": 1.635911671151028,
      "learning_rate": 0.000299860289950447,
      "loss": 2.625,
      "step": 2097
    },
    {
      "epoch": 1.3894039735099337,
      "grad_norm": 1.0380295247735245,
      "learning_rate": 0.0002998573949411875,
      "loss": 2.5,
      "step": 2098
    },
    {
      "epoch": 1.3900662251655629,
      "grad_norm": 1.0723657652729095,
      "learning_rate": 0.0002998544702584458,
      "loss": 2.6406,
      "step": 2099
    },
    {
      "epoch": 1.390728476821192,
      "grad_norm": 1.0909682795165,
      "learning_rate": 0.00029985151590280096,
      "loss": 2.7031,
      "step": 2100
    },
    {
      "epoch": 1.3913907284768212,
      "grad_norm": 1.132269107255355,
      "learning_rate": 0.000299848531874838,
      "loss": 2.5312,
      "step": 2101
    },
    {
      "epoch": 1.3920529801324504,
      "grad_norm": 1.0424865897875386,
      "learning_rate": 0.00029984551817514773,
      "loss": 2.5625,
      "step": 2102
    },
    {
      "epoch": 1.3927152317880795,
      "grad_norm": 1.215998969244023,
      "learning_rate": 0.0002998424748043269,
      "loss": 2.6094,
      "step": 2103
    },
    {
      "epoch": 1.3933774834437087,
      "grad_norm": 0.9827945155801082,
      "learning_rate": 0.0002998394017629783,
      "loss": 2.4375,
      "step": 2104
    },
    {
      "epoch": 1.3940397350993377,
      "grad_norm": 0.9744273230170389,
      "learning_rate": 0.00029983629905171015,
      "loss": 2.5469,
      "step": 2105
    },
    {
      "epoch": 1.3947019867549668,
      "grad_norm": 1.1271484555081328,
      "learning_rate": 0.00029983316667113707,
      "loss": 2.6094,
      "step": 2106
    },
    {
      "epoch": 1.395364238410596,
      "grad_norm": 1.096239640373542,
      "learning_rate": 0.00029983000462187917,
      "loss": 2.8281,
      "step": 2107
    },
    {
      "epoch": 1.3960264900662251,
      "grad_norm": 1.0393770138040104,
      "learning_rate": 0.0002998268129045626,
      "loss": 2.7812,
      "step": 2108
    },
    {
      "epoch": 1.3966887417218543,
      "grad_norm": 1.009495811360415,
      "learning_rate": 0.0002998235915198194,
      "loss": 2.7344,
      "step": 2109
    },
    {
      "epoch": 1.3973509933774835,
      "grad_norm": 1.024976689277755,
      "learning_rate": 0.0002998203404682874,
      "loss": 2.2812,
      "step": 2110
    },
    {
      "epoch": 1.3980132450331126,
      "grad_norm": 1.0709768509277813,
      "learning_rate": 0.00029981705975061035,
      "loss": 2.5625,
      "step": 2111
    },
    {
      "epoch": 1.3986754966887418,
      "grad_norm": 0.9189981798877973,
      "learning_rate": 0.00029981374936743785,
      "loss": 2.375,
      "step": 2112
    },
    {
      "epoch": 1.399337748344371,
      "grad_norm": 1.1019402390564017,
      "learning_rate": 0.00029981040931942543,
      "loss": 2.8125,
      "step": 2113
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9162996994941862,
      "learning_rate": 0.00029980703960723445,
      "loss": 2.3125,
      "step": 2114
    },
    {
      "epoch": 1.400662251655629,
      "grad_norm": 1.0926442260991014,
      "learning_rate": 0.0002998036402315322,
      "loss": 2.8594,
      "step": 2115
    },
    {
      "epoch": 1.4013245033112582,
      "grad_norm": 1.0555369624437276,
      "learning_rate": 0.00029980021119299167,
      "loss": 2.6562,
      "step": 2116
    },
    {
      "epoch": 1.4019867549668874,
      "grad_norm": 0.9092918659556374,
      "learning_rate": 0.0002997967524922919,
      "loss": 2.0,
      "step": 2117
    },
    {
      "epoch": 1.4026490066225166,
      "grad_norm": 1.0665436080451218,
      "learning_rate": 0.00029979326413011786,
      "loss": 2.5781,
      "step": 2118
    },
    {
      "epoch": 1.4033112582781457,
      "grad_norm": 1.011513565088108,
      "learning_rate": 0.0002997897461071601,
      "loss": 2.6562,
      "step": 2119
    },
    {
      "epoch": 1.403973509933775,
      "grad_norm": 1.0477597468168856,
      "learning_rate": 0.00029978619842411543,
      "loss": 2.7969,
      "step": 2120
    },
    {
      "epoch": 1.4046357615894038,
      "grad_norm": 0.88325777130976,
      "learning_rate": 0.00029978262108168615,
      "loss": 2.4219,
      "step": 2121
    },
    {
      "epoch": 1.4052980132450332,
      "grad_norm": 1.0496082092803243,
      "learning_rate": 0.00029977901408058074,
      "loss": 2.5938,
      "step": 2122
    },
    {
      "epoch": 1.4059602649006622,
      "grad_norm": 0.973819815876037,
      "learning_rate": 0.0002997753774215133,
      "loss": 2.7031,
      "step": 2123
    },
    {
      "epoch": 1.4066225165562913,
      "grad_norm": 0.9961371582182782,
      "learning_rate": 0.0002997717111052041,
      "loss": 2.4688,
      "step": 2124
    },
    {
      "epoch": 1.4072847682119205,
      "grad_norm": 0.8927419266461215,
      "learning_rate": 0.000299768015132379,
      "loss": 2.1875,
      "step": 2125
    },
    {
      "epoch": 1.4079470198675497,
      "grad_norm": 0.8993635239288216,
      "learning_rate": 0.00029976428950376986,
      "loss": 2.5156,
      "step": 2126
    },
    {
      "epoch": 1.4086092715231788,
      "grad_norm": 1.1536660343448664,
      "learning_rate": 0.0002997605342201144,
      "loss": 2.5625,
      "step": 2127
    },
    {
      "epoch": 1.409271523178808,
      "grad_norm": 0.9935687797992114,
      "learning_rate": 0.00029975674928215627,
      "loss": 2.4531,
      "step": 2128
    },
    {
      "epoch": 1.4099337748344372,
      "grad_norm": 1.1328804926151956,
      "learning_rate": 0.00029975293469064484,
      "loss": 2.8594,
      "step": 2129
    },
    {
      "epoch": 1.410596026490066,
      "grad_norm": 1.0656844295547943,
      "learning_rate": 0.00029974909044633546,
      "loss": 2.625,
      "step": 2130
    },
    {
      "epoch": 1.4112582781456955,
      "grad_norm": 0.9407947356703492,
      "learning_rate": 0.0002997452165499894,
      "loss": 2.3594,
      "step": 2131
    },
    {
      "epoch": 1.4119205298013244,
      "grad_norm": 0.9536151602500974,
      "learning_rate": 0.00029974131300237366,
      "loss": 2.4844,
      "step": 2132
    },
    {
      "epoch": 1.4125827814569536,
      "grad_norm": 0.9848676103028161,
      "learning_rate": 0.00029973737980426124,
      "loss": 2.3281,
      "step": 2133
    },
    {
      "epoch": 1.4132450331125828,
      "grad_norm": 1.2430168077390826,
      "learning_rate": 0.00029973341695643086,
      "loss": 2.7188,
      "step": 2134
    },
    {
      "epoch": 1.413907284768212,
      "grad_norm": 0.9449159280040796,
      "learning_rate": 0.00029972942445966733,
      "loss": 2.4531,
      "step": 2135
    },
    {
      "epoch": 1.414569536423841,
      "grad_norm": 1.1416375068990001,
      "learning_rate": 0.0002997254023147612,
      "loss": 2.6094,
      "step": 2136
    },
    {
      "epoch": 1.4152317880794703,
      "grad_norm": 1.6639698092751323,
      "learning_rate": 0.00029972135052250883,
      "loss": 2.4531,
      "step": 2137
    },
    {
      "epoch": 1.4158940397350994,
      "grad_norm": 1.0859096917199904,
      "learning_rate": 0.0002997172690837126,
      "loss": 2.2344,
      "step": 2138
    },
    {
      "epoch": 1.4165562913907284,
      "grad_norm": 1.0517869458832836,
      "learning_rate": 0.0002997131579991806,
      "loss": 2.4375,
      "step": 2139
    },
    {
      "epoch": 1.4172185430463577,
      "grad_norm": 0.984439846038069,
      "learning_rate": 0.00029970901726972694,
      "loss": 2.3438,
      "step": 2140
    },
    {
      "epoch": 1.4178807947019867,
      "grad_norm": 1.0841803994312005,
      "learning_rate": 0.00029970484689617144,
      "loss": 2.6719,
      "step": 2141
    },
    {
      "epoch": 1.4185430463576159,
      "grad_norm": 0.9571427054971211,
      "learning_rate": 0.00029970064687934,
      "loss": 2.4062,
      "step": 2142
    },
    {
      "epoch": 1.419205298013245,
      "grad_norm": 0.9968538494134072,
      "learning_rate": 0.00029969641722006423,
      "loss": 2.5312,
      "step": 2143
    },
    {
      "epoch": 1.4198675496688742,
      "grad_norm": 0.9838747613924839,
      "learning_rate": 0.00029969215791918153,
      "loss": 2.2812,
      "step": 2144
    },
    {
      "epoch": 1.4205298013245033,
      "grad_norm": 1.0002782704685769,
      "learning_rate": 0.0002996878689775355,
      "loss": 2.7188,
      "step": 2145
    },
    {
      "epoch": 1.4211920529801325,
      "grad_norm": 1.0772060173754583,
      "learning_rate": 0.0002996835503959753,
      "loss": 2.7031,
      "step": 2146
    },
    {
      "epoch": 1.4218543046357617,
      "grad_norm": 0.9145045826150184,
      "learning_rate": 0.000299679202175356,
      "loss": 2.625,
      "step": 2147
    },
    {
      "epoch": 1.4225165562913906,
      "grad_norm": 1.0217006279292333,
      "learning_rate": 0.0002996748243165387,
      "loss": 2.4219,
      "step": 2148
    },
    {
      "epoch": 1.4231788079470198,
      "grad_norm": 0.9962119284023744,
      "learning_rate": 0.0002996704168203902,
      "loss": 2.5312,
      "step": 2149
    },
    {
      "epoch": 1.423841059602649,
      "grad_norm": 0.9462382168409187,
      "learning_rate": 0.0002996659796877833,
      "loss": 2.6562,
      "step": 2150
    },
    {
      "epoch": 1.4245033112582781,
      "grad_norm": 1.0113280370470705,
      "learning_rate": 0.0002996615129195965,
      "loss": 2.5781,
      "step": 2151
    },
    {
      "epoch": 1.4251655629139073,
      "grad_norm": 0.9215109975379985,
      "learning_rate": 0.0002996570165167143,
      "loss": 2.2656,
      "step": 2152
    },
    {
      "epoch": 1.4258278145695364,
      "grad_norm": 1.12200714133647,
      "learning_rate": 0.0002996524904800271,
      "loss": 2.5312,
      "step": 2153
    },
    {
      "epoch": 1.4264900662251656,
      "grad_norm": 0.9472627848008581,
      "learning_rate": 0.0002996479348104311,
      "loss": 2.625,
      "step": 2154
    },
    {
      "epoch": 1.4271523178807948,
      "grad_norm": 1.0065261669622478,
      "learning_rate": 0.00029964334950882834,
      "loss": 2.5781,
      "step": 2155
    },
    {
      "epoch": 1.427814569536424,
      "grad_norm": 1.0467617191929983,
      "learning_rate": 0.0002996387345761268,
      "loss": 2.0625,
      "step": 2156
    },
    {
      "epoch": 1.4284768211920529,
      "grad_norm": 0.9684265857571045,
      "learning_rate": 0.00029963409001324015,
      "loss": 2.4062,
      "step": 2157
    },
    {
      "epoch": 1.429139072847682,
      "grad_norm": 0.8781180766063381,
      "learning_rate": 0.0002996294158210883,
      "loss": 2.125,
      "step": 2158
    },
    {
      "epoch": 1.4298013245033112,
      "grad_norm": 1.0049911770768698,
      "learning_rate": 0.0002996247120005966,
      "loss": 2.6875,
      "step": 2159
    },
    {
      "epoch": 1.4304635761589404,
      "grad_norm": 0.926129318755258,
      "learning_rate": 0.0002996199785526966,
      "loss": 2.125,
      "step": 2160
    },
    {
      "epoch": 1.4311258278145695,
      "grad_norm": 0.7898323276246039,
      "learning_rate": 0.0002996152154783255,
      "loss": 1.8984,
      "step": 2161
    },
    {
      "epoch": 1.4317880794701987,
      "grad_norm": 0.9626418487070505,
      "learning_rate": 0.0002996104227784264,
      "loss": 2.625,
      "step": 2162
    },
    {
      "epoch": 1.4324503311258279,
      "grad_norm": 0.9225346212743132,
      "learning_rate": 0.0002996056004539483,
      "loss": 2.5625,
      "step": 2163
    },
    {
      "epoch": 1.433112582781457,
      "grad_norm": 1.054888544536959,
      "learning_rate": 0.0002996007485058463,
      "loss": 2.8594,
      "step": 2164
    },
    {
      "epoch": 1.4337748344370862,
      "grad_norm": 0.8805679398921958,
      "learning_rate": 0.0002995958669350809,
      "loss": 2.125,
      "step": 2165
    },
    {
      "epoch": 1.4344370860927151,
      "grad_norm": 0.9921058010312066,
      "learning_rate": 0.0002995909557426188,
      "loss": 2.6094,
      "step": 2166
    },
    {
      "epoch": 1.4350993377483443,
      "grad_norm": 0.9666739059398567,
      "learning_rate": 0.00029958601492943243,
      "loss": 2.2188,
      "step": 2167
    },
    {
      "epoch": 1.4357615894039735,
      "grad_norm": 1.0461201495503443,
      "learning_rate": 0.00029958104449650017,
      "loss": 2.6875,
      "step": 2168
    },
    {
      "epoch": 1.4364238410596026,
      "grad_norm": 2.3315355821867647,
      "learning_rate": 0.00029957604444480625,
      "loss": 2.5,
      "step": 2169
    },
    {
      "epoch": 1.4370860927152318,
      "grad_norm": 0.9654622286173267,
      "learning_rate": 0.00029957101477534067,
      "loss": 2.6719,
      "step": 2170
    },
    {
      "epoch": 1.437748344370861,
      "grad_norm": 1.0580376279345556,
      "learning_rate": 0.0002995659554890993,
      "loss": 2.5469,
      "step": 2171
    },
    {
      "epoch": 1.4384105960264901,
      "grad_norm": 0.9972337421662041,
      "learning_rate": 0.00029956086658708414,
      "loss": 2.3906,
      "step": 2172
    },
    {
      "epoch": 1.439072847682119,
      "grad_norm": 1.1547474700803408,
      "learning_rate": 0.0002995557480703027,
      "loss": 2.5625,
      "step": 2173
    },
    {
      "epoch": 1.4397350993377485,
      "grad_norm": 1.0690611728683848,
      "learning_rate": 0.00029955059993976854,
      "loss": 2.2969,
      "step": 2174
    },
    {
      "epoch": 1.4403973509933774,
      "grad_norm": 1.0262170423749843,
      "learning_rate": 0.00029954542219650103,
      "loss": 2.7344,
      "step": 2175
    },
    {
      "epoch": 1.4410596026490066,
      "grad_norm": 1.0424042469909958,
      "learning_rate": 0.0002995402148415255,
      "loss": 2.2344,
      "step": 2176
    },
    {
      "epoch": 1.4417218543046357,
      "grad_norm": 0.9853496474894284,
      "learning_rate": 0.00029953497787587295,
      "loss": 2.4844,
      "step": 2177
    },
    {
      "epoch": 1.442384105960265,
      "grad_norm": 1.050554866514043,
      "learning_rate": 0.00029952971130058043,
      "loss": 2.6094,
      "step": 2178
    },
    {
      "epoch": 1.443046357615894,
      "grad_norm": 0.9362634108301414,
      "learning_rate": 0.0002995244151166908,
      "loss": 2.1875,
      "step": 2179
    },
    {
      "epoch": 1.4437086092715232,
      "grad_norm": 0.8963364612115888,
      "learning_rate": 0.0002995190893252527,
      "loss": 2.2812,
      "step": 2180
    },
    {
      "epoch": 1.4443708609271524,
      "grad_norm": 0.9531468696034419,
      "learning_rate": 0.0002995137339273208,
      "loss": 2.5781,
      "step": 2181
    },
    {
      "epoch": 1.4450331125827813,
      "grad_norm": 1.030754419163838,
      "learning_rate": 0.00029950834892395535,
      "loss": 2.5625,
      "step": 2182
    },
    {
      "epoch": 1.4456953642384107,
      "grad_norm": 1.0200434739291868,
      "learning_rate": 0.00029950293431622285,
      "loss": 2.4844,
      "step": 2183
    },
    {
      "epoch": 1.4463576158940397,
      "grad_norm": 0.9745848250975345,
      "learning_rate": 0.00029949749010519533,
      "loss": 2.4531,
      "step": 2184
    },
    {
      "epoch": 1.4470198675496688,
      "grad_norm": 0.9992595477958197,
      "learning_rate": 0.0002994920162919508,
      "loss": 2.2812,
      "step": 2185
    },
    {
      "epoch": 1.447682119205298,
      "grad_norm": 1.0011650865411004,
      "learning_rate": 0.00029948651287757327,
      "loss": 2.1094,
      "step": 2186
    },
    {
      "epoch": 1.4483443708609272,
      "grad_norm": 1.0626348387906883,
      "learning_rate": 0.00029948097986315234,
      "loss": 2.5781,
      "step": 2187
    },
    {
      "epoch": 1.4490066225165563,
      "grad_norm": 1.0959646330787183,
      "learning_rate": 0.0002994754172497836,
      "loss": 2.4688,
      "step": 2188
    },
    {
      "epoch": 1.4496688741721855,
      "grad_norm": 0.9306188761457541,
      "learning_rate": 0.00029946982503856857,
      "loss": 2.6719,
      "step": 2189
    },
    {
      "epoch": 1.4503311258278146,
      "grad_norm": 0.9073547574130615,
      "learning_rate": 0.0002994642032306146,
      "loss": 2.3594,
      "step": 2190
    },
    {
      "epoch": 1.4509933774834436,
      "grad_norm": 0.9797658272770757,
      "learning_rate": 0.00029945855182703486,
      "loss": 2.7188,
      "step": 2191
    },
    {
      "epoch": 1.451655629139073,
      "grad_norm": 0.9212378264447347,
      "learning_rate": 0.00029945287082894836,
      "loss": 2.5938,
      "step": 2192
    },
    {
      "epoch": 1.452317880794702,
      "grad_norm": 1.0151534078778723,
      "learning_rate": 0.00029944716023748,
      "loss": 2.4531,
      "step": 2193
    },
    {
      "epoch": 1.452980132450331,
      "grad_norm": 1.0209988612900802,
      "learning_rate": 0.00029944142005376055,
      "loss": 2.4062,
      "step": 2194
    },
    {
      "epoch": 1.4536423841059603,
      "grad_norm": 1.0569536280321712,
      "learning_rate": 0.00029943565027892665,
      "loss": 2.4688,
      "step": 2195
    },
    {
      "epoch": 1.4543046357615894,
      "grad_norm": 0.973073077517115,
      "learning_rate": 0.0002994298509141208,
      "loss": 2.625,
      "step": 2196
    },
    {
      "epoch": 1.4549668874172186,
      "grad_norm": 0.8472445519570411,
      "learning_rate": 0.00029942402196049127,
      "loss": 2.3125,
      "step": 2197
    },
    {
      "epoch": 1.4556291390728477,
      "grad_norm": 0.9250924754416165,
      "learning_rate": 0.0002994181634191923,
      "loss": 2.4062,
      "step": 2198
    },
    {
      "epoch": 1.456291390728477,
      "grad_norm": 0.9972108791238056,
      "learning_rate": 0.00029941227529138397,
      "loss": 2.75,
      "step": 2199
    },
    {
      "epoch": 1.4569536423841059,
      "grad_norm": 1.1005013481880201,
      "learning_rate": 0.00029940635757823216,
      "loss": 2.5156,
      "step": 2200
    },
    {
      "epoch": 1.457615894039735,
      "grad_norm": 0.90795962645883,
      "learning_rate": 0.0002994004102809087,
      "loss": 2.5781,
      "step": 2201
    },
    {
      "epoch": 1.4582781456953642,
      "grad_norm": 1.1582081424160675,
      "learning_rate": 0.00029939443340059116,
      "loss": 2.5469,
      "step": 2202
    },
    {
      "epoch": 1.4589403973509933,
      "grad_norm": 0.914456740134419,
      "learning_rate": 0.00029938842693846303,
      "loss": 2.3906,
      "step": 2203
    },
    {
      "epoch": 1.4596026490066225,
      "grad_norm": 0.9160385265915898,
      "learning_rate": 0.0002993823908957137,
      "loss": 2.0,
      "step": 2204
    },
    {
      "epoch": 1.4602649006622517,
      "grad_norm": 1.1101265255685904,
      "learning_rate": 0.00029937632527353837,
      "loss": 2.4688,
      "step": 2205
    },
    {
      "epoch": 1.4609271523178808,
      "grad_norm": 1.1247032209702812,
      "learning_rate": 0.0002993702300731381,
      "loss": 2.8906,
      "step": 2206
    },
    {
      "epoch": 1.46158940397351,
      "grad_norm": 0.9182031063560402,
      "learning_rate": 0.0002993641052957198,
      "loss": 2.8906,
      "step": 2207
    },
    {
      "epoch": 1.4622516556291392,
      "grad_norm": 0.9884498179904752,
      "learning_rate": 0.00029935795094249623,
      "loss": 2.3281,
      "step": 2208
    },
    {
      "epoch": 1.4629139072847681,
      "grad_norm": 0.9904464978923531,
      "learning_rate": 0.00029935176701468603,
      "loss": 2.4531,
      "step": 2209
    },
    {
      "epoch": 1.4635761589403973,
      "grad_norm": 0.8934308664586316,
      "learning_rate": 0.0002993455535135137,
      "loss": 2.2656,
      "step": 2210
    },
    {
      "epoch": 1.4642384105960264,
      "grad_norm": 0.9311145650141537,
      "learning_rate": 0.0002993393104402096,
      "loss": 2.4844,
      "step": 2211
    },
    {
      "epoch": 1.4649006622516556,
      "grad_norm": 0.89757965722654,
      "learning_rate": 0.0002993330377960099,
      "loss": 2.4375,
      "step": 2212
    },
    {
      "epoch": 1.4655629139072848,
      "grad_norm": 0.8820427296521648,
      "learning_rate": 0.0002993267355821567,
      "loss": 2.3125,
      "step": 2213
    },
    {
      "epoch": 1.466225165562914,
      "grad_norm": 1.0339670667228942,
      "learning_rate": 0.00029932040379989787,
      "loss": 2.7344,
      "step": 2214
    },
    {
      "epoch": 1.466887417218543,
      "grad_norm": 1.0508568588045355,
      "learning_rate": 0.00029931404245048715,
      "loss": 1.875,
      "step": 2215
    },
    {
      "epoch": 1.4675496688741723,
      "grad_norm": 0.9309277007384041,
      "learning_rate": 0.0002993076515351842,
      "loss": 2.6875,
      "step": 2216
    },
    {
      "epoch": 1.4682119205298014,
      "grad_norm": 0.8326487364228112,
      "learning_rate": 0.00029930123105525455,
      "loss": 2.4062,
      "step": 2217
    },
    {
      "epoch": 1.4688741721854304,
      "grad_norm": 0.9733888966128189,
      "learning_rate": 0.0002992947810119694,
      "loss": 2.6562,
      "step": 2218
    },
    {
      "epoch": 1.4695364238410595,
      "grad_norm": 0.9247161309578887,
      "learning_rate": 0.00029928830140660604,
      "loss": 2.4531,
      "step": 2219
    },
    {
      "epoch": 1.4701986754966887,
      "grad_norm": 0.9310946953839669,
      "learning_rate": 0.0002992817922404475,
      "loss": 2.5625,
      "step": 2220
    },
    {
      "epoch": 1.4708609271523179,
      "grad_norm": 1.0249183840394278,
      "learning_rate": 0.00029927525351478257,
      "loss": 2.4844,
      "step": 2221
    },
    {
      "epoch": 1.471523178807947,
      "grad_norm": 0.9056492402580734,
      "learning_rate": 0.0002992686852309061,
      "loss": 2.4531,
      "step": 2222
    },
    {
      "epoch": 1.4721854304635762,
      "grad_norm": 1.0069941009306147,
      "learning_rate": 0.0002992620873901186,
      "loss": 2.125,
      "step": 2223
    },
    {
      "epoch": 1.4728476821192054,
      "grad_norm": 0.9080274376842974,
      "learning_rate": 0.0002992554599937266,
      "loss": 2.375,
      "step": 2224
    },
    {
      "epoch": 1.4735099337748343,
      "grad_norm": 0.9574691269621646,
      "learning_rate": 0.0002992488030430424,
      "loss": 2.4844,
      "step": 2225
    },
    {
      "epoch": 1.4741721854304637,
      "grad_norm": 0.8857828960475895,
      "learning_rate": 0.000299242116539384,
      "loss": 2.0469,
      "step": 2226
    },
    {
      "epoch": 1.4748344370860926,
      "grad_norm": 0.9392572199148592,
      "learning_rate": 0.0002992354004840756,
      "loss": 2.4844,
      "step": 2227
    },
    {
      "epoch": 1.4754966887417218,
      "grad_norm": 0.9645189684553113,
      "learning_rate": 0.000299228654878447,
      "loss": 2.6562,
      "step": 2228
    },
    {
      "epoch": 1.476158940397351,
      "grad_norm": 1.0249253285767028,
      "learning_rate": 0.00029922187972383383,
      "loss": 2.625,
      "step": 2229
    },
    {
      "epoch": 1.4768211920529801,
      "grad_norm": 0.8895718191214521,
      "learning_rate": 0.00029921507502157774,
      "loss": 2.5312,
      "step": 2230
    },
    {
      "epoch": 1.4774834437086093,
      "grad_norm": 0.9697110142340752,
      "learning_rate": 0.00029920824077302605,
      "loss": 2.3906,
      "step": 2231
    },
    {
      "epoch": 1.4781456953642385,
      "grad_norm": 0.945257059297628,
      "learning_rate": 0.00029920137697953214,
      "loss": 2.4375,
      "step": 2232
    },
    {
      "epoch": 1.4788079470198676,
      "grad_norm": 0.9839439850869712,
      "learning_rate": 0.000299194483642455,
      "loss": 2.6094,
      "step": 2233
    },
    {
      "epoch": 1.4794701986754966,
      "grad_norm": 1.026555664998305,
      "learning_rate": 0.0002991875607631596,
      "loss": 2.5312,
      "step": 2234
    },
    {
      "epoch": 1.480132450331126,
      "grad_norm": 1.0400850075677455,
      "learning_rate": 0.0002991806083430169,
      "loss": 2.8594,
      "step": 2235
    },
    {
      "epoch": 1.480794701986755,
      "grad_norm": 1.6812644678260764,
      "learning_rate": 0.0002991736263834034,
      "loss": 2.4531,
      "step": 2236
    },
    {
      "epoch": 1.481456953642384,
      "grad_norm": 1.0481246874714307,
      "learning_rate": 0.0002991666148857016,
      "loss": 2.7031,
      "step": 2237
    },
    {
      "epoch": 1.4821192052980132,
      "grad_norm": 1.0080732245522337,
      "learning_rate": 0.0002991595738513,
      "loss": 2.3594,
      "step": 2238
    },
    {
      "epoch": 1.4827814569536424,
      "grad_norm": 0.9414461677030689,
      "learning_rate": 0.00029915250328159263,
      "loss": 2.5312,
      "step": 2239
    },
    {
      "epoch": 1.4834437086092715,
      "grad_norm": 0.9626123170038507,
      "learning_rate": 0.00029914540317797973,
      "loss": 2.5469,
      "step": 2240
    },
    {
      "epoch": 1.4841059602649007,
      "grad_norm": 0.96942463610998,
      "learning_rate": 0.000299138273541867,
      "loss": 1.9297,
      "step": 2241
    },
    {
      "epoch": 1.4847682119205299,
      "grad_norm": 1.040747242865704,
      "learning_rate": 0.0002991311143746664,
      "loss": 2.5781,
      "step": 2242
    },
    {
      "epoch": 1.4854304635761588,
      "grad_norm": 1.1787401948161933,
      "learning_rate": 0.00029912392567779536,
      "loss": 2.875,
      "step": 2243
    },
    {
      "epoch": 1.4860927152317882,
      "grad_norm": 0.9710683383118753,
      "learning_rate": 0.0002991167074526774,
      "loss": 2.5156,
      "step": 2244
    },
    {
      "epoch": 1.4867549668874172,
      "grad_norm": 0.8931900777410339,
      "learning_rate": 0.00029910945970074176,
      "loss": 2.4219,
      "step": 2245
    },
    {
      "epoch": 1.4874172185430463,
      "grad_norm": 1.032200369726091,
      "learning_rate": 0.0002991021824234236,
      "loss": 2.4844,
      "step": 2246
    },
    {
      "epoch": 1.4880794701986755,
      "grad_norm": 1.0916729537974976,
      "learning_rate": 0.000299094875622164,
      "loss": 2.5156,
      "step": 2247
    },
    {
      "epoch": 1.4887417218543046,
      "grad_norm": 0.900077158911175,
      "learning_rate": 0.0002990875392984097,
      "loss": 2.0312,
      "step": 2248
    },
    {
      "epoch": 1.4894039735099338,
      "grad_norm": 0.89775513470307,
      "learning_rate": 0.00029908017345361336,
      "loss": 2.3906,
      "step": 2249
    },
    {
      "epoch": 1.490066225165563,
      "grad_norm": 0.9965560385657952,
      "learning_rate": 0.00029907277808923356,
      "loss": 2.4688,
      "step": 2250
    },
    {
      "epoch": 1.4907284768211921,
      "grad_norm": 1.109496676069234,
      "learning_rate": 0.0002990653532067346,
      "loss": 2.8125,
      "step": 2251
    },
    {
      "epoch": 1.491390728476821,
      "grad_norm": 0.9651585987199373,
      "learning_rate": 0.0002990578988075867,
      "loss": 2.6562,
      "step": 2252
    },
    {
      "epoch": 1.4920529801324505,
      "grad_norm": 0.9516042390966882,
      "learning_rate": 0.000299050414893266,
      "loss": 2.125,
      "step": 2253
    },
    {
      "epoch": 1.4927152317880794,
      "grad_norm": 1.0959892438052803,
      "learning_rate": 0.0002990429014652543,
      "loss": 2.4844,
      "step": 2254
    },
    {
      "epoch": 1.4933774834437086,
      "grad_norm": 0.9295136835936418,
      "learning_rate": 0.0002990353585250394,
      "loss": 2.7031,
      "step": 2255
    },
    {
      "epoch": 1.4940397350993377,
      "grad_norm": 0.9764504972134843,
      "learning_rate": 0.0002990277860741149,
      "loss": 2.7969,
      "step": 2256
    },
    {
      "epoch": 1.494701986754967,
      "grad_norm": 0.9694103062956547,
      "learning_rate": 0.0002990201841139802,
      "loss": 2.5625,
      "step": 2257
    },
    {
      "epoch": 1.495364238410596,
      "grad_norm": 1.0119943119577604,
      "learning_rate": 0.0002990125526461405,
      "loss": 2.5312,
      "step": 2258
    },
    {
      "epoch": 1.4960264900662252,
      "grad_norm": 0.9893357839726827,
      "learning_rate": 0.00029900489167210715,
      "loss": 2.0312,
      "step": 2259
    },
    {
      "epoch": 1.4966887417218544,
      "grad_norm": 0.8753220294517673,
      "learning_rate": 0.00029899720119339686,
      "loss": 1.9844,
      "step": 2260
    },
    {
      "epoch": 1.4973509933774833,
      "grad_norm": 1.066527330356536,
      "learning_rate": 0.0002989894812115326,
      "loss": 2.7031,
      "step": 2261
    },
    {
      "epoch": 1.4980132450331125,
      "grad_norm": 0.9629555378366176,
      "learning_rate": 0.00029898173172804293,
      "loss": 2.6094,
      "step": 2262
    },
    {
      "epoch": 1.4986754966887417,
      "grad_norm": 0.8913995764898816,
      "learning_rate": 0.0002989739527444624,
      "loss": 2.1719,
      "step": 2263
    },
    {
      "epoch": 1.4993377483443708,
      "grad_norm": 0.9658175403720054,
      "learning_rate": 0.0002989661442623312,
      "loss": 2.4844,
      "step": 2264
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9734340932988587,
      "learning_rate": 0.00029895830628319563,
      "loss": 2.4844,
      "step": 2265
    },
    {
      "epoch": 1.5006622516556292,
      "grad_norm": 0.9816524994313781,
      "learning_rate": 0.0002989504388086077,
      "loss": 2.5625,
      "step": 2266
    },
    {
      "epoch": 1.5013245033112583,
      "grad_norm": 0.9929745950744212,
      "learning_rate": 0.0002989425418401252,
      "loss": 2.6875,
      "step": 2267
    },
    {
      "epoch": 1.5019867549668873,
      "grad_norm": 0.9783998967602694,
      "learning_rate": 0.00029893461537931185,
      "loss": 2.7188,
      "step": 2268
    },
    {
      "epoch": 1.5026490066225167,
      "grad_norm": 0.9472676338288144,
      "learning_rate": 0.00029892665942773716,
      "loss": 2.4531,
      "step": 2269
    },
    {
      "epoch": 1.5033112582781456,
      "grad_norm": 1.0038027271668695,
      "learning_rate": 0.0002989186739869765,
      "loss": 2.8438,
      "step": 2270
    },
    {
      "epoch": 1.503973509933775,
      "grad_norm": 0.880262084566376,
      "learning_rate": 0.0002989106590586111,
      "loss": 2.5938,
      "step": 2271
    },
    {
      "epoch": 1.504635761589404,
      "grad_norm": 0.9546595433612308,
      "learning_rate": 0.00029890261464422807,
      "loss": 2.2031,
      "step": 2272
    },
    {
      "epoch": 1.505298013245033,
      "grad_norm": 1.0304215914545412,
      "learning_rate": 0.0002988945407454202,
      "loss": 2.0469,
      "step": 2273
    },
    {
      "epoch": 1.5059602649006623,
      "grad_norm": 1.0082173313359755,
      "learning_rate": 0.0002988864373637862,
      "loss": 2.4375,
      "step": 2274
    },
    {
      "epoch": 1.5066225165562914,
      "grad_norm": 0.9174620352802345,
      "learning_rate": 0.0002988783045009307,
      "loss": 2.4219,
      "step": 2275
    },
    {
      "epoch": 1.5072847682119206,
      "grad_norm": 0.9889198017995099,
      "learning_rate": 0.00029887014215846405,
      "loss": 2.625,
      "step": 2276
    },
    {
      "epoch": 1.5079470198675495,
      "grad_norm": 0.9571846479884542,
      "learning_rate": 0.00029886195033800256,
      "loss": 2.4062,
      "step": 2277
    },
    {
      "epoch": 1.508609271523179,
      "grad_norm": 0.9545811897127711,
      "learning_rate": 0.0002988537290411682,
      "loss": 2.2969,
      "step": 2278
    },
    {
      "epoch": 1.5092715231788079,
      "grad_norm": 1.0962799388965638,
      "learning_rate": 0.000298845478269589,
      "loss": 2.6094,
      "step": 2279
    },
    {
      "epoch": 1.5099337748344372,
      "grad_norm": 1.0133772617585435,
      "learning_rate": 0.0002988371980248986,
      "loss": 2.5781,
      "step": 2280
    },
    {
      "epoch": 1.5105960264900662,
      "grad_norm": 0.9128192630021276,
      "learning_rate": 0.00029882888830873663,
      "loss": 2.5469,
      "step": 2281
    },
    {
      "epoch": 1.5112582781456954,
      "grad_norm": 0.8833398705599705,
      "learning_rate": 0.0002988205491227485,
      "loss": 2.5156,
      "step": 2282
    },
    {
      "epoch": 1.5119205298013245,
      "grad_norm": 0.8314339604795769,
      "learning_rate": 0.0002988121804685855,
      "loss": 2.4062,
      "step": 2283
    },
    {
      "epoch": 1.5125827814569537,
      "grad_norm": 0.9287077315177573,
      "learning_rate": 0.0002988037823479047,
      "loss": 2.1406,
      "step": 2284
    },
    {
      "epoch": 1.5132450331125828,
      "grad_norm": 0.9627672760179488,
      "learning_rate": 0.00029879535476236903,
      "loss": 2.6094,
      "step": 2285
    },
    {
      "epoch": 1.5139072847682118,
      "grad_norm": 0.8508930409776032,
      "learning_rate": 0.0002987868977136472,
      "loss": 2.4375,
      "step": 2286
    },
    {
      "epoch": 1.5145695364238412,
      "grad_norm": 0.8712854269686106,
      "learning_rate": 0.0002987784112034138,
      "loss": 2.5469,
      "step": 2287
    },
    {
      "epoch": 1.5152317880794701,
      "grad_norm": 1.0499644764796048,
      "learning_rate": 0.00029876989523334936,
      "loss": 2.7969,
      "step": 2288
    },
    {
      "epoch": 1.5158940397350993,
      "grad_norm": 0.940808725635407,
      "learning_rate": 0.00029876134980514,
      "loss": 2.2031,
      "step": 2289
    },
    {
      "epoch": 1.5165562913907285,
      "grad_norm": 0.7610166684769152,
      "learning_rate": 0.00029875277492047794,
      "loss": 2.125,
      "step": 2290
    },
    {
      "epoch": 1.5172185430463576,
      "grad_norm": 0.8540399164426568,
      "learning_rate": 0.00029874417058106103,
      "loss": 2.375,
      "step": 2291
    },
    {
      "epoch": 1.5178807947019868,
      "grad_norm": 0.9202743303114995,
      "learning_rate": 0.00029873553678859307,
      "loss": 2.625,
      "step": 2292
    },
    {
      "epoch": 1.518543046357616,
      "grad_norm": 1.003214411076348,
      "learning_rate": 0.00029872687354478357,
      "loss": 2.4219,
      "step": 2293
    },
    {
      "epoch": 1.519205298013245,
      "grad_norm": 0.8953183449789073,
      "learning_rate": 0.0002987181808513481,
      "loss": 2.4531,
      "step": 2294
    },
    {
      "epoch": 1.519867549668874,
      "grad_norm": 0.8983376888905227,
      "learning_rate": 0.0002987094587100078,
      "loss": 2.3438,
      "step": 2295
    },
    {
      "epoch": 1.5205298013245034,
      "grad_norm": 0.8689615262058188,
      "learning_rate": 0.00029870070712248974,
      "loss": 1.9688,
      "step": 2296
    },
    {
      "epoch": 1.5211920529801324,
      "grad_norm": 0.8401751857227879,
      "learning_rate": 0.0002986919260905269,
      "loss": 2.375,
      "step": 2297
    },
    {
      "epoch": 1.5218543046357615,
      "grad_norm": 1.0394392811345194,
      "learning_rate": 0.000298683115615858,
      "loss": 2.5625,
      "step": 2298
    },
    {
      "epoch": 1.5225165562913907,
      "grad_norm": 1.0124725877938936,
      "learning_rate": 0.0002986742757002276,
      "loss": 2.5938,
      "step": 2299
    },
    {
      "epoch": 1.5231788079470199,
      "grad_norm": 1.0393730476410066,
      "learning_rate": 0.0002986654063453861,
      "loss": 2.4062,
      "step": 2300
    },
    {
      "epoch": 1.523841059602649,
      "grad_norm": 0.9114001830138886,
      "learning_rate": 0.00029865650755308973,
      "loss": 2.5781,
      "step": 2301
    },
    {
      "epoch": 1.524503311258278,
      "grad_norm": 0.9910707700753598,
      "learning_rate": 0.0002986475793251006,
      "loss": 2.4844,
      "step": 2302
    },
    {
      "epoch": 1.5251655629139074,
      "grad_norm": 0.9974585673477396,
      "learning_rate": 0.0002986386216631866,
      "loss": 2.8125,
      "step": 2303
    },
    {
      "epoch": 1.5258278145695363,
      "grad_norm": 0.9410577666642724,
      "learning_rate": 0.0002986296345691214,
      "loss": 2.1406,
      "step": 2304
    },
    {
      "epoch": 1.5264900662251657,
      "grad_norm": 0.8831950557754421,
      "learning_rate": 0.0002986206180446846,
      "loss": 2.0469,
      "step": 2305
    },
    {
      "epoch": 1.5271523178807946,
      "grad_norm": 0.8589849832315861,
      "learning_rate": 0.00029861157209166154,
      "loss": 2.4219,
      "step": 2306
    },
    {
      "epoch": 1.5278145695364238,
      "grad_norm": 0.8912659034075949,
      "learning_rate": 0.0002986024967118434,
      "loss": 2.5781,
      "step": 2307
    },
    {
      "epoch": 1.528476821192053,
      "grad_norm": 1.1690949523673406,
      "learning_rate": 0.0002985933919070272,
      "loss": 2.4688,
      "step": 2308
    },
    {
      "epoch": 1.5291390728476821,
      "grad_norm": 1.0646105778195296,
      "learning_rate": 0.00029858425767901594,
      "loss": 2.4688,
      "step": 2309
    },
    {
      "epoch": 1.5298013245033113,
      "grad_norm": 0.8374765050660513,
      "learning_rate": 0.0002985750940296181,
      "loss": 2.3594,
      "step": 2310
    },
    {
      "epoch": 1.5304635761589402,
      "grad_norm": 0.9532256370755269,
      "learning_rate": 0.00029856590096064836,
      "loss": 2.5312,
      "step": 2311
    },
    {
      "epoch": 1.5311258278145696,
      "grad_norm": 0.8775748988864893,
      "learning_rate": 0.00029855667847392695,
      "loss": 2.2969,
      "step": 2312
    },
    {
      "epoch": 1.5317880794701986,
      "grad_norm": 0.9907184930039408,
      "learning_rate": 0.00029854742657128006,
      "loss": 2.4219,
      "step": 2313
    },
    {
      "epoch": 1.532450331125828,
      "grad_norm": 0.956373136861473,
      "learning_rate": 0.00029853814525453967,
      "loss": 2.3906,
      "step": 2314
    },
    {
      "epoch": 1.533112582781457,
      "grad_norm": 0.9685820840848379,
      "learning_rate": 0.00029852883452554356,
      "loss": 2.5625,
      "step": 2315
    },
    {
      "epoch": 1.533774834437086,
      "grad_norm": 0.932313468229487,
      "learning_rate": 0.0002985194943861354,
      "loss": 2.5312,
      "step": 2316
    },
    {
      "epoch": 1.5344370860927152,
      "grad_norm": 1.0244475744412036,
      "learning_rate": 0.0002985101248381646,
      "loss": 2.8125,
      "step": 2317
    },
    {
      "epoch": 1.5350993377483444,
      "grad_norm": 0.9502117724836681,
      "learning_rate": 0.0002985007258834865,
      "loss": 2.4688,
      "step": 2318
    },
    {
      "epoch": 1.5357615894039736,
      "grad_norm": 1.0646374754798642,
      "learning_rate": 0.00029849129752396224,
      "loss": 2.4531,
      "step": 2319
    },
    {
      "epoch": 1.5364238410596025,
      "grad_norm": 0.994271909670904,
      "learning_rate": 0.0002984818397614586,
      "loss": 2.5312,
      "step": 2320
    },
    {
      "epoch": 1.5370860927152319,
      "grad_norm": 1.3409726468281091,
      "learning_rate": 0.00029847235259784836,
      "loss": 2.6094,
      "step": 2321
    },
    {
      "epoch": 1.5377483443708608,
      "grad_norm": 0.9340548772315902,
      "learning_rate": 0.0002984628360350102,
      "loss": 2.3438,
      "step": 2322
    },
    {
      "epoch": 1.5384105960264902,
      "grad_norm": 0.97459115939099,
      "learning_rate": 0.00029845329007482837,
      "loss": 2.625,
      "step": 2323
    },
    {
      "epoch": 1.5390728476821192,
      "grad_norm": 0.9239066205749943,
      "learning_rate": 0.0002984437147191932,
      "loss": 2.4062,
      "step": 2324
    },
    {
      "epoch": 1.5397350993377483,
      "grad_norm": 0.9450340327131521,
      "learning_rate": 0.0002984341099700006,
      "loss": 2.2969,
      "step": 2325
    },
    {
      "epoch": 1.5403973509933775,
      "grad_norm": 0.9880422459754131,
      "learning_rate": 0.00029842447582915254,
      "loss": 2.6094,
      "step": 2326
    },
    {
      "epoch": 1.5410596026490067,
      "grad_norm": 0.9353465623155267,
      "learning_rate": 0.00029841481229855664,
      "loss": 2.6094,
      "step": 2327
    },
    {
      "epoch": 1.5417218543046358,
      "grad_norm": 0.9505862196628337,
      "learning_rate": 0.00029840511938012635,
      "loss": 2.7656,
      "step": 2328
    },
    {
      "epoch": 1.5423841059602648,
      "grad_norm": 0.96840165754237,
      "learning_rate": 0.000298395397075781,
      "loss": 2.5469,
      "step": 2329
    },
    {
      "epoch": 1.5430463576158941,
      "grad_norm": 0.9471417742023834,
      "learning_rate": 0.0002983856453874458,
      "loss": 2.5781,
      "step": 2330
    },
    {
      "epoch": 1.543708609271523,
      "grad_norm": 1.0361705525948484,
      "learning_rate": 0.0002983758643170516,
      "loss": 2.3438,
      "step": 2331
    },
    {
      "epoch": 1.5443708609271525,
      "grad_norm": 0.9711235007772678,
      "learning_rate": 0.00029836605386653514,
      "loss": 2.1406,
      "step": 2332
    },
    {
      "epoch": 1.5450331125827814,
      "grad_norm": 0.8907324349231668,
      "learning_rate": 0.0002983562140378391,
      "loss": 2.3594,
      "step": 2333
    },
    {
      "epoch": 1.5456953642384106,
      "grad_norm": 1.0357534061322606,
      "learning_rate": 0.0002983463448329118,
      "loss": 2.4375,
      "step": 2334
    },
    {
      "epoch": 1.5463576158940397,
      "grad_norm": 0.9708299997851642,
      "learning_rate": 0.00029833644625370755,
      "loss": 2.2969,
      "step": 2335
    },
    {
      "epoch": 1.547019867549669,
      "grad_norm": 0.858879059028281,
      "learning_rate": 0.0002983265183021863,
      "loss": 2.3906,
      "step": 2336
    },
    {
      "epoch": 1.547682119205298,
      "grad_norm": 0.9739831016371527,
      "learning_rate": 0.00029831656098031387,
      "loss": 2.5469,
      "step": 2337
    },
    {
      "epoch": 1.548344370860927,
      "grad_norm": 0.9697674657167231,
      "learning_rate": 0.00029830657429006204,
      "loss": 2.4844,
      "step": 2338
    },
    {
      "epoch": 1.5490066225165564,
      "grad_norm": 0.9673812037816254,
      "learning_rate": 0.00029829655823340815,
      "loss": 2.4062,
      "step": 2339
    },
    {
      "epoch": 1.5496688741721854,
      "grad_norm": 0.9158567071596156,
      "learning_rate": 0.0002982865128123356,
      "loss": 2.5625,
      "step": 2340
    },
    {
      "epoch": 1.5503311258278145,
      "grad_norm": 0.9920372217043345,
      "learning_rate": 0.00029827643802883346,
      "loss": 2.75,
      "step": 2341
    },
    {
      "epoch": 1.5509933774834437,
      "grad_norm": 0.9817012933590009,
      "learning_rate": 0.00029826633388489667,
      "loss": 2.8281,
      "step": 2342
    },
    {
      "epoch": 1.5516556291390728,
      "grad_norm": 0.9503736626809571,
      "learning_rate": 0.0002982562003825259,
      "loss": 2.5625,
      "step": 2343
    },
    {
      "epoch": 1.552317880794702,
      "grad_norm": 1.0030852686766019,
      "learning_rate": 0.00029824603752372784,
      "loss": 2.4375,
      "step": 2344
    },
    {
      "epoch": 1.5529801324503312,
      "grad_norm": 0.9344704733189892,
      "learning_rate": 0.0002982358453105147,
      "loss": 2.4375,
      "step": 2345
    },
    {
      "epoch": 1.5536423841059603,
      "grad_norm": 0.9332399013724976,
      "learning_rate": 0.00029822562374490474,
      "loss": 2.0469,
      "step": 2346
    },
    {
      "epoch": 1.5543046357615893,
      "grad_norm": 1.0400771581708228,
      "learning_rate": 0.0002982153728289219,
      "loss": 2.6406,
      "step": 2347
    },
    {
      "epoch": 1.5549668874172187,
      "grad_norm": 0.9299631193745552,
      "learning_rate": 0.0002982050925645961,
      "loss": 2.5625,
      "step": 2348
    },
    {
      "epoch": 1.5556291390728476,
      "grad_norm": 1.0543976538834634,
      "learning_rate": 0.00029819478295396273,
      "loss": 2.5625,
      "step": 2349
    },
    {
      "epoch": 1.5562913907284768,
      "grad_norm": 0.8048431251298995,
      "learning_rate": 0.00029818444399906343,
      "loss": 2.4062,
      "step": 2350
    },
    {
      "epoch": 1.556953642384106,
      "grad_norm": 1.0602584321964825,
      "learning_rate": 0.00029817407570194526,
      "loss": 2.3906,
      "step": 2351
    },
    {
      "epoch": 1.557615894039735,
      "grad_norm": 0.9291421931039523,
      "learning_rate": 0.00029816367806466144,
      "loss": 2.5312,
      "step": 2352
    },
    {
      "epoch": 1.5582781456953643,
      "grad_norm": 0.992995606800549,
      "learning_rate": 0.00029815325108927063,
      "loss": 2.375,
      "step": 2353
    },
    {
      "epoch": 1.5589403973509932,
      "grad_norm": 0.9741620788160839,
      "learning_rate": 0.0002981427947778376,
      "loss": 2.625,
      "step": 2354
    },
    {
      "epoch": 1.5596026490066226,
      "grad_norm": 0.8615299422277266,
      "learning_rate": 0.00029813230913243287,
      "loss": 2.0781,
      "step": 2355
    },
    {
      "epoch": 1.5602649006622515,
      "grad_norm": 0.8798929319777776,
      "learning_rate": 0.0002981217941551326,
      "loss": 2.5156,
      "step": 2356
    },
    {
      "epoch": 1.560927152317881,
      "grad_norm": 0.8279006252632788,
      "learning_rate": 0.0002981112498480189,
      "loss": 2.4688,
      "step": 2357
    },
    {
      "epoch": 1.5615894039735099,
      "grad_norm": 1.002257797790237,
      "learning_rate": 0.0002981006762131797,
      "loss": 2.5,
      "step": 2358
    },
    {
      "epoch": 1.562251655629139,
      "grad_norm": 0.9171342633827861,
      "learning_rate": 0.00029809007325270874,
      "loss": 2.5625,
      "step": 2359
    },
    {
      "epoch": 1.5629139072847682,
      "grad_norm": 0.941713190640774,
      "learning_rate": 0.00029807944096870543,
      "loss": 2.5312,
      "step": 2360
    },
    {
      "epoch": 1.5635761589403974,
      "grad_norm": 1.0245051698378338,
      "learning_rate": 0.00029806877936327516,
      "loss": 2.5625,
      "step": 2361
    },
    {
      "epoch": 1.5642384105960265,
      "grad_norm": 0.9750012959043385,
      "learning_rate": 0.000298058088438529,
      "loss": 2.5938,
      "step": 2362
    },
    {
      "epoch": 1.5649006622516555,
      "grad_norm": 0.9564782676316433,
      "learning_rate": 0.0002980473681965839,
      "loss": 2.3438,
      "step": 2363
    },
    {
      "epoch": 1.5655629139072849,
      "grad_norm": 1.0372406776313476,
      "learning_rate": 0.0002980366186395626,
      "loss": 1.9844,
      "step": 2364
    },
    {
      "epoch": 1.5662251655629138,
      "grad_norm": 0.8622284588621415,
      "learning_rate": 0.00029802583976959365,
      "loss": 2.3438,
      "step": 2365
    },
    {
      "epoch": 1.5668874172185432,
      "grad_norm": 1.0890346987948818,
      "learning_rate": 0.0002980150315888113,
      "loss": 2.6719,
      "step": 2366
    },
    {
      "epoch": 1.5675496688741721,
      "grad_norm": 0.9009961808051479,
      "learning_rate": 0.0002980041940993558,
      "loss": 2.4062,
      "step": 2367
    },
    {
      "epoch": 1.5682119205298013,
      "grad_norm": 1.0873686512594303,
      "learning_rate": 0.0002979933273033731,
      "loss": 2.7344,
      "step": 2368
    },
    {
      "epoch": 1.5688741721854305,
      "grad_norm": 0.9468863401231004,
      "learning_rate": 0.00029798243120301483,
      "loss": 2.6406,
      "step": 2369
    },
    {
      "epoch": 1.5695364238410596,
      "grad_norm": 0.9154868169855724,
      "learning_rate": 0.00029797150580043865,
      "loss": 2.4062,
      "step": 2370
    },
    {
      "epoch": 1.5701986754966888,
      "grad_norm": 0.9990317802731059,
      "learning_rate": 0.0002979605510978079,
      "loss": 2.5156,
      "step": 2371
    },
    {
      "epoch": 1.5708609271523177,
      "grad_norm": 0.9572739789961483,
      "learning_rate": 0.0002979495670972917,
      "loss": 2.6562,
      "step": 2372
    },
    {
      "epoch": 1.5715231788079471,
      "grad_norm": 1.0905066247282007,
      "learning_rate": 0.00029793855380106504,
      "loss": 2.7031,
      "step": 2373
    },
    {
      "epoch": 1.572185430463576,
      "grad_norm": 0.8891378730523425,
      "learning_rate": 0.00029792751121130866,
      "loss": 1.9062,
      "step": 2374
    },
    {
      "epoch": 1.5728476821192054,
      "grad_norm": 0.9267373929256784,
      "learning_rate": 0.0002979164393302091,
      "loss": 2.375,
      "step": 2375
    },
    {
      "epoch": 1.5735099337748344,
      "grad_norm": 1.012734501282315,
      "learning_rate": 0.0002979053381599588,
      "loss": 2.7344,
      "step": 2376
    },
    {
      "epoch": 1.5741721854304636,
      "grad_norm": 0.8562492228571245,
      "learning_rate": 0.0002978942077027558,
      "loss": 2.5312,
      "step": 2377
    },
    {
      "epoch": 1.5748344370860927,
      "grad_norm": 0.8732027729215307,
      "learning_rate": 0.0002978830479608042,
      "loss": 2.5156,
      "step": 2378
    },
    {
      "epoch": 1.5754966887417219,
      "grad_norm": 0.8641806821922278,
      "learning_rate": 0.00029787185893631363,
      "loss": 2.3906,
      "step": 2379
    },
    {
      "epoch": 1.576158940397351,
      "grad_norm": 0.8640118854531463,
      "learning_rate": 0.00029786064063149967,
      "loss": 2.4375,
      "step": 2380
    },
    {
      "epoch": 1.57682119205298,
      "grad_norm": 0.9513393906552332,
      "learning_rate": 0.00029784939304858374,
      "loss": 2.5312,
      "step": 2381
    },
    {
      "epoch": 1.5774834437086094,
      "grad_norm": 1.020123235678892,
      "learning_rate": 0.00029783811618979295,
      "loss": 2.6875,
      "step": 2382
    },
    {
      "epoch": 1.5781456953642383,
      "grad_norm": 0.9272453328144052,
      "learning_rate": 0.0002978268100573602,
      "loss": 2.5469,
      "step": 2383
    },
    {
      "epoch": 1.5788079470198677,
      "grad_norm": 0.9623133678816124,
      "learning_rate": 0.0002978154746535244,
      "loss": 2.4375,
      "step": 2384
    },
    {
      "epoch": 1.5794701986754967,
      "grad_norm": 0.9659305387567694,
      "learning_rate": 0.0002978041099805299,
      "loss": 2.3594,
      "step": 2385
    },
    {
      "epoch": 1.5801324503311258,
      "grad_norm": 1.0379222943849518,
      "learning_rate": 0.00029779271604062703,
      "loss": 2.4531,
      "step": 2386
    },
    {
      "epoch": 1.580794701986755,
      "grad_norm": 0.8737672293738501,
      "learning_rate": 0.00029778129283607207,
      "loss": 2.4844,
      "step": 2387
    },
    {
      "epoch": 1.5814569536423841,
      "grad_norm": 0.8325813173900654,
      "learning_rate": 0.00029776984036912685,
      "loss": 2.2031,
      "step": 2388
    },
    {
      "epoch": 1.5821192052980133,
      "grad_norm": 0.8755006523790766,
      "learning_rate": 0.0002977583586420591,
      "loss": 2.375,
      "step": 2389
    },
    {
      "epoch": 1.5827814569536423,
      "grad_norm": 1.0033656813564527,
      "learning_rate": 0.0002977468476571424,
      "loss": 2.5781,
      "step": 2390
    },
    {
      "epoch": 1.5834437086092716,
      "grad_norm": 0.9334592130908209,
      "learning_rate": 0.0002977353074166559,
      "loss": 2.4062,
      "step": 2391
    },
    {
      "epoch": 1.5841059602649006,
      "grad_norm": 0.8954430600009667,
      "learning_rate": 0.0002977237379228849,
      "loss": 2.6719,
      "step": 2392
    },
    {
      "epoch": 1.58476821192053,
      "grad_norm": 0.8210776148421538,
      "learning_rate": 0.00029771213917812014,
      "loss": 2.4531,
      "step": 2393
    },
    {
      "epoch": 1.585430463576159,
      "grad_norm": 0.9484909865010892,
      "learning_rate": 0.00029770051118465843,
      "loss": 2.3906,
      "step": 2394
    },
    {
      "epoch": 1.586092715231788,
      "grad_norm": 0.9987138274395734,
      "learning_rate": 0.0002976888539448021,
      "loss": 2.6719,
      "step": 2395
    },
    {
      "epoch": 1.5867549668874172,
      "grad_norm": 0.8930728065901855,
      "learning_rate": 0.0002976771674608595,
      "loss": 2.4688,
      "step": 2396
    },
    {
      "epoch": 1.5874172185430464,
      "grad_norm": 0.8237134434385337,
      "learning_rate": 0.0002976654517351447,
      "loss": 2.5156,
      "step": 2397
    },
    {
      "epoch": 1.5880794701986756,
      "grad_norm": 0.8619875144250574,
      "learning_rate": 0.00029765370676997755,
      "loss": 2.4531,
      "step": 2398
    },
    {
      "epoch": 1.5887417218543045,
      "grad_norm": 0.9294276874983126,
      "learning_rate": 0.00029764193256768363,
      "loss": 2.4531,
      "step": 2399
    },
    {
      "epoch": 1.589403973509934,
      "grad_norm": 0.8831402325195679,
      "learning_rate": 0.0002976301291305944,
      "loss": 2.6406,
      "step": 2400
    },
    {
      "epoch": 1.5900662251655628,
      "grad_norm": 1.0230412118662695,
      "learning_rate": 0.00029761829646104705,
      "loss": 2.4375,
      "step": 2401
    },
    {
      "epoch": 1.590728476821192,
      "grad_norm": 0.8894562258071816,
      "learning_rate": 0.00029760643456138466,
      "loss": 2.2656,
      "step": 2402
    },
    {
      "epoch": 1.5913907284768212,
      "grad_norm": 0.8078294510479886,
      "learning_rate": 0.00029759454343395597,
      "loss": 2.25,
      "step": 2403
    },
    {
      "epoch": 1.5920529801324503,
      "grad_norm": 1.0209886475711418,
      "learning_rate": 0.0002975826230811156,
      "loss": 2.5469,
      "step": 2404
    },
    {
      "epoch": 1.5927152317880795,
      "grad_norm": 1.0808786751764354,
      "learning_rate": 0.00029757067350522374,
      "loss": 2.7031,
      "step": 2405
    },
    {
      "epoch": 1.5933774834437087,
      "grad_norm": 0.8428675741475674,
      "learning_rate": 0.00029755869470864685,
      "loss": 2.4062,
      "step": 2406
    },
    {
      "epoch": 1.5940397350993378,
      "grad_norm": 1.0010016669089024,
      "learning_rate": 0.0002975466866937566,
      "loss": 2.7812,
      "step": 2407
    },
    {
      "epoch": 1.5947019867549668,
      "grad_norm": 0.9569858602738587,
      "learning_rate": 0.0002975346494629308,
      "loss": 2.4531,
      "step": 2408
    },
    {
      "epoch": 1.5953642384105962,
      "grad_norm": 1.0268137863826086,
      "learning_rate": 0.0002975225830185531,
      "loss": 2.5469,
      "step": 2409
    },
    {
      "epoch": 1.596026490066225,
      "grad_norm": 0.9053607000346373,
      "learning_rate": 0.0002975104873630126,
      "loss": 2.4375,
      "step": 2410
    },
    {
      "epoch": 1.5966887417218543,
      "grad_norm": 1.0184609758465082,
      "learning_rate": 0.00029749836249870444,
      "loss": 2.5,
      "step": 2411
    },
    {
      "epoch": 1.5973509933774834,
      "grad_norm": 0.9042872727381593,
      "learning_rate": 0.00029748620842802946,
      "loss": 2.0781,
      "step": 2412
    },
    {
      "epoch": 1.5980132450331126,
      "grad_norm": 0.8591852312806941,
      "learning_rate": 0.0002974740251533944,
      "loss": 2.5,
      "step": 2413
    },
    {
      "epoch": 1.5986754966887418,
      "grad_norm": 0.934462510995766,
      "learning_rate": 0.0002974618126772116,
      "loss": 2.7656,
      "step": 2414
    },
    {
      "epoch": 1.5993377483443707,
      "grad_norm": 0.8730359068258811,
      "learning_rate": 0.0002974495710018993,
      "loss": 2.4375,
      "step": 2415
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9872681835620981,
      "learning_rate": 0.0002974373001298815,
      "loss": 2.4844,
      "step": 2416
    },
    {
      "epoch": 1.600662251655629,
      "grad_norm": 0.8919496174672944,
      "learning_rate": 0.00029742500006358795,
      "loss": 2.5,
      "step": 2417
    },
    {
      "epoch": 1.6013245033112584,
      "grad_norm": 0.9487790343811981,
      "learning_rate": 0.0002974126708054542,
      "loss": 2.3281,
      "step": 2418
    },
    {
      "epoch": 1.6019867549668874,
      "grad_norm": 0.8546058049536142,
      "learning_rate": 0.0002974003123579216,
      "loss": 2.5,
      "step": 2419
    },
    {
      "epoch": 1.6026490066225165,
      "grad_norm": 0.859949829192009,
      "learning_rate": 0.0002973879247234373,
      "loss": 2.5625,
      "step": 2420
    },
    {
      "epoch": 1.6033112582781457,
      "grad_norm": 0.7985351918342742,
      "learning_rate": 0.00029737550790445407,
      "loss": 2.0625,
      "step": 2421
    },
    {
      "epoch": 1.6039735099337749,
      "grad_norm": 0.8857349917786531,
      "learning_rate": 0.0002973630619034307,
      "loss": 2.3438,
      "step": 2422
    },
    {
      "epoch": 1.604635761589404,
      "grad_norm": 0.8285877575852945,
      "learning_rate": 0.00029735058672283166,
      "loss": 2.4844,
      "step": 2423
    },
    {
      "epoch": 1.605298013245033,
      "grad_norm": 0.9846045595173922,
      "learning_rate": 0.00029733808236512716,
      "loss": 2.5781,
      "step": 2424
    },
    {
      "epoch": 1.6059602649006623,
      "grad_norm": 0.9801182541446352,
      "learning_rate": 0.0002973255488327931,
      "loss": 2.4844,
      "step": 2425
    },
    {
      "epoch": 1.6066225165562913,
      "grad_norm": 0.9868258291042384,
      "learning_rate": 0.0002973129861283114,
      "loss": 2.6562,
      "step": 2426
    },
    {
      "epoch": 1.6072847682119207,
      "grad_norm": 0.8602606548908338,
      "learning_rate": 0.0002973003942541695,
      "loss": 2.5156,
      "step": 2427
    },
    {
      "epoch": 1.6079470198675496,
      "grad_norm": 0.924995734651,
      "learning_rate": 0.0002972877732128608,
      "loss": 2.4375,
      "step": 2428
    },
    {
      "epoch": 1.6086092715231788,
      "grad_norm": 0.9904999208179349,
      "learning_rate": 0.0002972751230068844,
      "loss": 2.5156,
      "step": 2429
    },
    {
      "epoch": 1.609271523178808,
      "grad_norm": 0.9204157229376017,
      "learning_rate": 0.0002972624436387452,
      "loss": 2.5938,
      "step": 2430
    },
    {
      "epoch": 1.6099337748344371,
      "grad_norm": 0.9287160542684919,
      "learning_rate": 0.0002972497351109539,
      "loss": 2.4531,
      "step": 2431
    },
    {
      "epoch": 1.6105960264900663,
      "grad_norm": 0.9651980681652861,
      "learning_rate": 0.00029723699742602685,
      "loss": 2.7031,
      "step": 2432
    },
    {
      "epoch": 1.6112582781456952,
      "grad_norm": 0.9202741001296313,
      "learning_rate": 0.0002972242305864863,
      "loss": 2.1094,
      "step": 2433
    },
    {
      "epoch": 1.6119205298013246,
      "grad_norm": 0.9875269864386703,
      "learning_rate": 0.0002972114345948602,
      "loss": 2.6562,
      "step": 2434
    },
    {
      "epoch": 1.6125827814569536,
      "grad_norm": 0.9186820809571713,
      "learning_rate": 0.00029719860945368233,
      "loss": 2.5781,
      "step": 2435
    },
    {
      "epoch": 1.613245033112583,
      "grad_norm": 0.8523000304138898,
      "learning_rate": 0.0002971857551654922,
      "loss": 2.2344,
      "step": 2436
    },
    {
      "epoch": 1.6139072847682119,
      "grad_norm": 0.956453137167526,
      "learning_rate": 0.0002971728717328352,
      "loss": 2.6406,
      "step": 2437
    },
    {
      "epoch": 1.614569536423841,
      "grad_norm": 0.9050015882918971,
      "learning_rate": 0.0002971599591582622,
      "loss": 2.7969,
      "step": 2438
    },
    {
      "epoch": 1.6152317880794702,
      "grad_norm": 0.8637140999603633,
      "learning_rate": 0.0002971470174443302,
      "loss": 2.4219,
      "step": 2439
    },
    {
      "epoch": 1.6158940397350994,
      "grad_norm": 0.9863673427417546,
      "learning_rate": 0.00029713404659360176,
      "loss": 2.4531,
      "step": 2440
    },
    {
      "epoch": 1.6165562913907285,
      "grad_norm": 0.8955056107363167,
      "learning_rate": 0.0002971210466086453,
      "loss": 2.5,
      "step": 2441
    },
    {
      "epoch": 1.6172185430463575,
      "grad_norm": 0.8879588175380839,
      "learning_rate": 0.0002971080174920348,
      "loss": 2.2188,
      "step": 2442
    },
    {
      "epoch": 1.6178807947019869,
      "grad_norm": 1.028000790993832,
      "learning_rate": 0.00029709495924635045,
      "loss": 2.4844,
      "step": 2443
    },
    {
      "epoch": 1.6185430463576158,
      "grad_norm": 0.8637714928622464,
      "learning_rate": 0.0002970818718741777,
      "loss": 2.3125,
      "step": 2444
    },
    {
      "epoch": 1.6192052980132452,
      "grad_norm": 0.8635138525369889,
      "learning_rate": 0.0002970687553781081,
      "loss": 2.0,
      "step": 2445
    },
    {
      "epoch": 1.6198675496688741,
      "grad_norm": 0.9575566593010487,
      "learning_rate": 0.0002970556097607388,
      "loss": 2.8125,
      "step": 2446
    },
    {
      "epoch": 1.6205298013245033,
      "grad_norm": 0.8954183814866251,
      "learning_rate": 0.00029704243502467293,
      "loss": 2.5,
      "step": 2447
    },
    {
      "epoch": 1.6211920529801325,
      "grad_norm": 0.9744444378487812,
      "learning_rate": 0.0002970292311725191,
      "loss": 2.7031,
      "step": 2448
    },
    {
      "epoch": 1.6218543046357616,
      "grad_norm": 0.8441910576985993,
      "learning_rate": 0.0002970159982068918,
      "loss": 2.3125,
      "step": 2449
    },
    {
      "epoch": 1.6225165562913908,
      "grad_norm": 0.8350917558959503,
      "learning_rate": 0.0002970027361304114,
      "loss": 2.1094,
      "step": 2450
    },
    {
      "epoch": 1.6231788079470197,
      "grad_norm": 0.9557612191133796,
      "learning_rate": 0.0002969894449457039,
      "loss": 2.5781,
      "step": 2451
    },
    {
      "epoch": 1.6238410596026491,
      "grad_norm": 0.8798993955553475,
      "learning_rate": 0.0002969761246554012,
      "loss": 2.4688,
      "step": 2452
    },
    {
      "epoch": 1.624503311258278,
      "grad_norm": 0.9249936981912198,
      "learning_rate": 0.00029696277526214073,
      "loss": 2.5625,
      "step": 2453
    },
    {
      "epoch": 1.6251655629139072,
      "grad_norm": 0.9011911893529734,
      "learning_rate": 0.0002969493967685659,
      "loss": 2.5156,
      "step": 2454
    },
    {
      "epoch": 1.6258278145695364,
      "grad_norm": 0.8126041197100745,
      "learning_rate": 0.0002969359891773258,
      "loss": 2.0469,
      "step": 2455
    },
    {
      "epoch": 1.6264900662251656,
      "grad_norm": 1.0140357982383312,
      "learning_rate": 0.00029692255249107527,
      "loss": 2.5781,
      "step": 2456
    },
    {
      "epoch": 1.6271523178807947,
      "grad_norm": 0.8957509478966947,
      "learning_rate": 0.0002969090867124749,
      "loss": 2.4688,
      "step": 2457
    },
    {
      "epoch": 1.627814569536424,
      "grad_norm": 0.8271319658991183,
      "learning_rate": 0.00029689559184419115,
      "loss": 2.3906,
      "step": 2458
    },
    {
      "epoch": 1.628476821192053,
      "grad_norm": 0.884732262071767,
      "learning_rate": 0.0002968820678888961,
      "loss": 2.2969,
      "step": 2459
    },
    {
      "epoch": 1.629139072847682,
      "grad_norm": 0.8943569482564087,
      "learning_rate": 0.00029686851484926765,
      "loss": 2.5312,
      "step": 2460
    },
    {
      "epoch": 1.6298013245033114,
      "grad_norm": 0.925214913149083,
      "learning_rate": 0.00029685493272798945,
      "loss": 2.4688,
      "step": 2461
    },
    {
      "epoch": 1.6304635761589403,
      "grad_norm": 0.9227541809368459,
      "learning_rate": 0.000296841321527751,
      "loss": 2.2812,
      "step": 2462
    },
    {
      "epoch": 1.6311258278145695,
      "grad_norm": 0.8909742394212812,
      "learning_rate": 0.00029682768125124736,
      "loss": 2.625,
      "step": 2463
    },
    {
      "epoch": 1.6317880794701987,
      "grad_norm": 0.8858356984496005,
      "learning_rate": 0.0002968140119011795,
      "loss": 2.2656,
      "step": 2464
    },
    {
      "epoch": 1.6324503311258278,
      "grad_norm": 1.006105203066916,
      "learning_rate": 0.00029680031348025413,
      "loss": 2.4688,
      "step": 2465
    },
    {
      "epoch": 1.633112582781457,
      "grad_norm": 0.8962316846275129,
      "learning_rate": 0.0002967865859911837,
      "loss": 2.3906,
      "step": 2466
    },
    {
      "epoch": 1.633774834437086,
      "grad_norm": 0.8756714418443138,
      "learning_rate": 0.0002967728294366863,
      "loss": 1.9844,
      "step": 2467
    },
    {
      "epoch": 1.6344370860927153,
      "grad_norm": 0.882409504515992,
      "learning_rate": 0.0002967590438194861,
      "loss": 2.3438,
      "step": 2468
    },
    {
      "epoch": 1.6350993377483443,
      "grad_norm": 1.0424492258171814,
      "learning_rate": 0.0002967452291423126,
      "loss": 2.4062,
      "step": 2469
    },
    {
      "epoch": 1.6357615894039736,
      "grad_norm": 0.9904991551817648,
      "learning_rate": 0.0002967313854079014,
      "loss": 2.6875,
      "step": 2470
    },
    {
      "epoch": 1.6364238410596026,
      "grad_norm": 0.8835217164067358,
      "learning_rate": 0.00029671751261899364,
      "loss": 2.4062,
      "step": 2471
    },
    {
      "epoch": 1.6370860927152318,
      "grad_norm": 0.8926008196475873,
      "learning_rate": 0.0002967036107783363,
      "loss": 2.2188,
      "step": 2472
    },
    {
      "epoch": 1.637748344370861,
      "grad_norm": 0.8575907785237484,
      "learning_rate": 0.00029668967988868213,
      "loss": 2.4062,
      "step": 2473
    },
    {
      "epoch": 1.63841059602649,
      "grad_norm": 0.9965891846674804,
      "learning_rate": 0.00029667571995278965,
      "loss": 2.2656,
      "step": 2474
    },
    {
      "epoch": 1.6390728476821192,
      "grad_norm": 0.9171955389767135,
      "learning_rate": 0.000296661730973423,
      "loss": 2.3281,
      "step": 2475
    },
    {
      "epoch": 1.6397350993377482,
      "grad_norm": 0.8940498241238249,
      "learning_rate": 0.0002966477129533522,
      "loss": 2.625,
      "step": 2476
    },
    {
      "epoch": 1.6403973509933776,
      "grad_norm": 1.0076348590343014,
      "learning_rate": 0.000296633665895353,
      "loss": 2.7031,
      "step": 2477
    },
    {
      "epoch": 1.6410596026490065,
      "grad_norm": 0.9670535478522838,
      "learning_rate": 0.00029661958980220683,
      "loss": 2.5781,
      "step": 2478
    },
    {
      "epoch": 1.641721854304636,
      "grad_norm": 0.9169111806738729,
      "learning_rate": 0.000296605484676701,
      "loss": 2.4844,
      "step": 2479
    },
    {
      "epoch": 1.6423841059602649,
      "grad_norm": 0.9061594441033957,
      "learning_rate": 0.0002965913505216283,
      "loss": 2.4844,
      "step": 2480
    },
    {
      "epoch": 1.643046357615894,
      "grad_norm": 0.8807957931037431,
      "learning_rate": 0.00029657718733978763,
      "loss": 2.7344,
      "step": 2481
    },
    {
      "epoch": 1.6437086092715232,
      "grad_norm": 0.9516746326153281,
      "learning_rate": 0.00029656299513398346,
      "loss": 2.4375,
      "step": 2482
    },
    {
      "epoch": 1.6443708609271523,
      "grad_norm": 0.885504056530061,
      "learning_rate": 0.0002965487739070259,
      "loss": 2.6562,
      "step": 2483
    },
    {
      "epoch": 1.6450331125827815,
      "grad_norm": 0.9207784940011139,
      "learning_rate": 0.00029653452366173105,
      "loss": 2.5,
      "step": 2484
    },
    {
      "epoch": 1.6456953642384105,
      "grad_norm": 0.8724305614966632,
      "learning_rate": 0.00029652024440092055,
      "loss": 2.4062,
      "step": 2485
    },
    {
      "epoch": 1.6463576158940398,
      "grad_norm": 0.9588223423779472,
      "learning_rate": 0.00029650593612742175,
      "loss": 2.0781,
      "step": 2486
    },
    {
      "epoch": 1.6470198675496688,
      "grad_norm": 0.7856607897252336,
      "learning_rate": 0.00029649159884406805,
      "loss": 2.0312,
      "step": 2487
    },
    {
      "epoch": 1.6476821192052982,
      "grad_norm": 0.9393493811563344,
      "learning_rate": 0.0002964772325536983,
      "loss": 2.5938,
      "step": 2488
    },
    {
      "epoch": 1.648344370860927,
      "grad_norm": 0.8342894854101043,
      "learning_rate": 0.00029646283725915724,
      "loss": 2.4531,
      "step": 2489
    },
    {
      "epoch": 1.6490066225165563,
      "grad_norm": 0.875487361416563,
      "learning_rate": 0.00029644841296329517,
      "loss": 2.2188,
      "step": 2490
    },
    {
      "epoch": 1.6496688741721854,
      "grad_norm": 0.8719427346322601,
      "learning_rate": 0.0002964339596689684,
      "loss": 2.3125,
      "step": 2491
    },
    {
      "epoch": 1.6503311258278146,
      "grad_norm": 0.7819528258197547,
      "learning_rate": 0.00029641947737903883,
      "loss": 2.2812,
      "step": 2492
    },
    {
      "epoch": 1.6509933774834438,
      "grad_norm": 0.9539859238537944,
      "learning_rate": 0.0002964049660963741,
      "loss": 2.5938,
      "step": 2493
    },
    {
      "epoch": 1.6516556291390727,
      "grad_norm": 0.8887884747205,
      "learning_rate": 0.00029639042582384754,
      "loss": 2.5938,
      "step": 2494
    },
    {
      "epoch": 1.652317880794702,
      "grad_norm": 0.7918565196093121,
      "learning_rate": 0.0002963758565643384,
      "loss": 2.1562,
      "step": 2495
    },
    {
      "epoch": 1.652980132450331,
      "grad_norm": 0.8455026625817008,
      "learning_rate": 0.0002963612583207315,
      "loss": 2.0156,
      "step": 2496
    },
    {
      "epoch": 1.6536423841059604,
      "grad_norm": 0.9137528803305601,
      "learning_rate": 0.00029634663109591745,
      "loss": 2.375,
      "step": 2497
    },
    {
      "epoch": 1.6543046357615894,
      "grad_norm": 1.0820947024216907,
      "learning_rate": 0.0002963319748927927,
      "loss": 2.3906,
      "step": 2498
    },
    {
      "epoch": 1.6549668874172185,
      "grad_norm": 0.8570299091131258,
      "learning_rate": 0.00029631728971425924,
      "loss": 2.3906,
      "step": 2499
    },
    {
      "epoch": 1.6556291390728477,
      "grad_norm": 0.9557455477475454,
      "learning_rate": 0.0002963025755632249,
      "loss": 2.5938,
      "step": 2500
    },
    {
      "epoch": 1.6562913907284769,
      "grad_norm": 0.962988600969181,
      "learning_rate": 0.0002962878324426034,
      "loss": 2.5938,
      "step": 2501
    },
    {
      "epoch": 1.656953642384106,
      "grad_norm": 0.9142053875905668,
      "learning_rate": 0.0002962730603553139,
      "loss": 2.3594,
      "step": 2502
    },
    {
      "epoch": 1.657615894039735,
      "grad_norm": 0.8832372421562507,
      "learning_rate": 0.0002962582593042814,
      "loss": 2.4688,
      "step": 2503
    },
    {
      "epoch": 1.6582781456953644,
      "grad_norm": 1.0021957273667554,
      "learning_rate": 0.00029624342929243685,
      "loss": 2.3594,
      "step": 2504
    },
    {
      "epoch": 1.6589403973509933,
      "grad_norm": 1.0082572135692691,
      "learning_rate": 0.0002962285703227167,
      "loss": 2.6094,
      "step": 2505
    },
    {
      "epoch": 1.6596026490066225,
      "grad_norm": 0.9189166003149819,
      "learning_rate": 0.00029621368239806306,
      "loss": 2.5938,
      "step": 2506
    },
    {
      "epoch": 1.6602649006622516,
      "grad_norm": 1.4262735074300428,
      "learning_rate": 0.0002961987655214241,
      "loss": 2.2344,
      "step": 2507
    },
    {
      "epoch": 1.6609271523178808,
      "grad_norm": 0.9321857571152382,
      "learning_rate": 0.00029618381969575343,
      "loss": 2.8125,
      "step": 2508
    },
    {
      "epoch": 1.66158940397351,
      "grad_norm": 0.9378131613914461,
      "learning_rate": 0.00029616884492401056,
      "loss": 2.6562,
      "step": 2509
    },
    {
      "epoch": 1.6622516556291391,
      "grad_norm": 0.9207033235213784,
      "learning_rate": 0.00029615384120916055,
      "loss": 2.6562,
      "step": 2510
    },
    {
      "epoch": 1.6629139072847683,
      "grad_norm": 0.9891487030342996,
      "learning_rate": 0.00029613880855417445,
      "loss": 2.4219,
      "step": 2511
    },
    {
      "epoch": 1.6635761589403972,
      "grad_norm": 0.9565369105191688,
      "learning_rate": 0.00029612374696202877,
      "loss": 2.5781,
      "step": 2512
    },
    {
      "epoch": 1.6642384105960266,
      "grad_norm": 1.0412300250695123,
      "learning_rate": 0.000296108656435706,
      "loss": 2.4219,
      "step": 2513
    },
    {
      "epoch": 1.6649006622516556,
      "grad_norm": 0.812507220229497,
      "learning_rate": 0.00029609353697819415,
      "loss": 2.25,
      "step": 2514
    },
    {
      "epoch": 1.6655629139072847,
      "grad_norm": 0.9664473875187014,
      "learning_rate": 0.00029607838859248704,
      "loss": 2.5469,
      "step": 2515
    },
    {
      "epoch": 1.666225165562914,
      "grad_norm": 0.9288071014102777,
      "learning_rate": 0.00029606321128158424,
      "loss": 2.4531,
      "step": 2516
    },
    {
      "epoch": 1.666887417218543,
      "grad_norm": 0.8402553885271143,
      "learning_rate": 0.00029604800504849107,
      "loss": 2.3438,
      "step": 2517
    },
    {
      "epoch": 1.6675496688741722,
      "grad_norm": 0.9270230842027883,
      "learning_rate": 0.00029603276989621856,
      "loss": 2.5469,
      "step": 2518
    },
    {
      "epoch": 1.6682119205298012,
      "grad_norm": 0.7845918121844384,
      "learning_rate": 0.0002960175058277834,
      "loss": 2.25,
      "step": 2519
    },
    {
      "epoch": 1.6688741721854305,
      "grad_norm": 1.1314153668884515,
      "learning_rate": 0.00029600221284620796,
      "loss": 2.5,
      "step": 2520
    },
    {
      "epoch": 1.6695364238410595,
      "grad_norm": 0.9142033066283387,
      "learning_rate": 0.0002959868909545206,
      "loss": 2.3906,
      "step": 2521
    },
    {
      "epoch": 1.6701986754966889,
      "grad_norm": 0.7461381558359854,
      "learning_rate": 0.0002959715401557551,
      "loss": 2.3906,
      "step": 2522
    },
    {
      "epoch": 1.6708609271523178,
      "grad_norm": 0.8737013975995181,
      "learning_rate": 0.00029595616045295116,
      "loss": 2.4531,
      "step": 2523
    },
    {
      "epoch": 1.671523178807947,
      "grad_norm": 1.0407656323481111,
      "learning_rate": 0.00029594075184915413,
      "loss": 2.7344,
      "step": 2524
    },
    {
      "epoch": 1.6721854304635762,
      "grad_norm": 0.9087627064755572,
      "learning_rate": 0.00029592531434741507,
      "loss": 2.3281,
      "step": 2525
    },
    {
      "epoch": 1.6728476821192053,
      "grad_norm": 0.9507041330774911,
      "learning_rate": 0.00029590984795079076,
      "loss": 2.7031,
      "step": 2526
    },
    {
      "epoch": 1.6735099337748345,
      "grad_norm": 0.8800413602214542,
      "learning_rate": 0.00029589435266234377,
      "loss": 2.3281,
      "step": 2527
    },
    {
      "epoch": 1.6741721854304634,
      "grad_norm": 0.9529742091132241,
      "learning_rate": 0.0002958788284851424,
      "loss": 2.4062,
      "step": 2528
    },
    {
      "epoch": 1.6748344370860928,
      "grad_norm": 0.8583137509860099,
      "learning_rate": 0.00029586327542226046,
      "loss": 2.6406,
      "step": 2529
    },
    {
      "epoch": 1.6754966887417218,
      "grad_norm": 0.8543336615175251,
      "learning_rate": 0.0002958476934767777,
      "loss": 2.4531,
      "step": 2530
    },
    {
      "epoch": 1.6761589403973511,
      "grad_norm": 0.8105857849733314,
      "learning_rate": 0.0002958320826517796,
      "loss": 2.4844,
      "step": 2531
    },
    {
      "epoch": 1.67682119205298,
      "grad_norm": 0.8728057328419205,
      "learning_rate": 0.0002958164429503572,
      "loss": 2.375,
      "step": 2532
    },
    {
      "epoch": 1.6774834437086092,
      "grad_norm": 0.9172841279358025,
      "learning_rate": 0.00029580077437560735,
      "loss": 2.5781,
      "step": 2533
    },
    {
      "epoch": 1.6781456953642384,
      "grad_norm": 0.9290784498546605,
      "learning_rate": 0.00029578507693063267,
      "loss": 2.375,
      "step": 2534
    },
    {
      "epoch": 1.6788079470198676,
      "grad_norm": 0.828707997520692,
      "learning_rate": 0.00029576935061854134,
      "loss": 2.1094,
      "step": 2535
    },
    {
      "epoch": 1.6794701986754967,
      "grad_norm": 0.8279893603198488,
      "learning_rate": 0.0002957535954424474,
      "loss": 2.5469,
      "step": 2536
    },
    {
      "epoch": 1.6801324503311257,
      "grad_norm": 0.9090676639471604,
      "learning_rate": 0.00029573781140547055,
      "loss": 2.375,
      "step": 2537
    },
    {
      "epoch": 1.680794701986755,
      "grad_norm": 1.0400049925891237,
      "learning_rate": 0.00029572199851073626,
      "loss": 2.4062,
      "step": 2538
    },
    {
      "epoch": 1.681456953642384,
      "grad_norm": 0.8678711591981842,
      "learning_rate": 0.00029570615676137555,
      "loss": 2.5312,
      "step": 2539
    },
    {
      "epoch": 1.6821192052980134,
      "grad_norm": 0.8957696337410335,
      "learning_rate": 0.0002956902861605254,
      "loss": 2.4219,
      "step": 2540
    },
    {
      "epoch": 1.6827814569536423,
      "grad_norm": 0.8155419036151653,
      "learning_rate": 0.00029567438671132827,
      "loss": 2.2031,
      "step": 2541
    },
    {
      "epoch": 1.6834437086092715,
      "grad_norm": 0.887347998499952,
      "learning_rate": 0.0002956584584169325,
      "loss": 2.5312,
      "step": 2542
    },
    {
      "epoch": 1.6841059602649007,
      "grad_norm": 0.8282048220679504,
      "learning_rate": 0.000295642501280492,
      "loss": 2.3594,
      "step": 2543
    },
    {
      "epoch": 1.6847682119205298,
      "grad_norm": 2.3942236716724477,
      "learning_rate": 0.00029562651530516654,
      "loss": 2.3906,
      "step": 2544
    },
    {
      "epoch": 1.685430463576159,
      "grad_norm": 0.7996151923135151,
      "learning_rate": 0.0002956105004941215,
      "loss": 2.4062,
      "step": 2545
    },
    {
      "epoch": 1.686092715231788,
      "grad_norm": 0.8299602241433127,
      "learning_rate": 0.00029559445685052794,
      "loss": 2.4531,
      "step": 2546
    },
    {
      "epoch": 1.6867549668874173,
      "grad_norm": 1.4081350342925683,
      "learning_rate": 0.0002955783843775628,
      "loss": 2.4531,
      "step": 2547
    },
    {
      "epoch": 1.6874172185430463,
      "grad_norm": 0.9081043676704833,
      "learning_rate": 0.0002955622830784085,
      "loss": 2.5312,
      "step": 2548
    },
    {
      "epoch": 1.6880794701986757,
      "grad_norm": 0.9831379797706636,
      "learning_rate": 0.00029554615295625334,
      "loss": 2.6719,
      "step": 2549
    },
    {
      "epoch": 1.6887417218543046,
      "grad_norm": 0.8479943026568344,
      "learning_rate": 0.0002955299940142913,
      "loss": 2.5156,
      "step": 2550
    },
    {
      "epoch": 1.6894039735099338,
      "grad_norm": 0.9874858672005594,
      "learning_rate": 0.000295513806255722,
      "loss": 2.5781,
      "step": 2551
    },
    {
      "epoch": 1.690066225165563,
      "grad_norm": 0.9233420361431135,
      "learning_rate": 0.00029549758968375076,
      "loss": 2.6406,
      "step": 2552
    },
    {
      "epoch": 1.690728476821192,
      "grad_norm": 0.8932505653663981,
      "learning_rate": 0.0002954813443015887,
      "loss": 2.6406,
      "step": 2553
    },
    {
      "epoch": 1.6913907284768213,
      "grad_norm": 0.8876311546012673,
      "learning_rate": 0.0002954650701124526,
      "loss": 2.3906,
      "step": 2554
    },
    {
      "epoch": 1.6920529801324502,
      "grad_norm": 0.852746015599343,
      "learning_rate": 0.0002954487671195649,
      "loss": 2.5312,
      "step": 2555
    },
    {
      "epoch": 1.6927152317880796,
      "grad_norm": 0.9468494986205876,
      "learning_rate": 0.0002954324353261538,
      "loss": 2.375,
      "step": 2556
    },
    {
      "epoch": 1.6933774834437085,
      "grad_norm": 0.813245238450774,
      "learning_rate": 0.00029541607473545313,
      "loss": 2.5156,
      "step": 2557
    },
    {
      "epoch": 1.6940397350993377,
      "grad_norm": 0.8858025214658736,
      "learning_rate": 0.0002953996853507026,
      "loss": 2.4688,
      "step": 2558
    },
    {
      "epoch": 1.6947019867549669,
      "grad_norm": 0.8959918089730343,
      "learning_rate": 0.0002953832671751474,
      "loss": 2.2812,
      "step": 2559
    },
    {
      "epoch": 1.695364238410596,
      "grad_norm": 0.8531950629710732,
      "learning_rate": 0.00029536682021203846,
      "loss": 2.375,
      "step": 2560
    },
    {
      "epoch": 1.6960264900662252,
      "grad_norm": 0.8729907720406065,
      "learning_rate": 0.00029535034446463265,
      "loss": 2.2031,
      "step": 2561
    },
    {
      "epoch": 1.6966887417218544,
      "grad_norm": 0.8239848429446279,
      "learning_rate": 0.00029533383993619216,
      "loss": 2.2031,
      "step": 2562
    },
    {
      "epoch": 1.6973509933774835,
      "grad_norm": 0.8489352105649699,
      "learning_rate": 0.0002953173066299852,
      "loss": 1.9062,
      "step": 2563
    },
    {
      "epoch": 1.6980132450331125,
      "grad_norm": 1.0663965885204487,
      "learning_rate": 0.00029530074454928553,
      "loss": 2.6094,
      "step": 2564
    },
    {
      "epoch": 1.6986754966887418,
      "grad_norm": 0.8208280592398177,
      "learning_rate": 0.00029528415369737263,
      "loss": 2.5156,
      "step": 2565
    },
    {
      "epoch": 1.6993377483443708,
      "grad_norm": 0.9004009129093324,
      "learning_rate": 0.0002952675340775316,
      "loss": 2.5469,
      "step": 2566
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9865877868444369,
      "learning_rate": 0.00029525088569305345,
      "loss": 2.3906,
      "step": 2567
    },
    {
      "epoch": 1.7006622516556291,
      "grad_norm": 0.9815568120361647,
      "learning_rate": 0.00029523420854723467,
      "loss": 2.5781,
      "step": 2568
    },
    {
      "epoch": 1.7013245033112583,
      "grad_norm": 0.8749133725247288,
      "learning_rate": 0.0002952175026433775,
      "loss": 2.1094,
      "step": 2569
    },
    {
      "epoch": 1.7019867549668874,
      "grad_norm": 0.8185752254637615,
      "learning_rate": 0.00029520076798479004,
      "loss": 2.3906,
      "step": 2570
    },
    {
      "epoch": 1.7026490066225164,
      "grad_norm": 0.8725437517479909,
      "learning_rate": 0.0002951840045747858,
      "loss": 2.7031,
      "step": 2571
    },
    {
      "epoch": 1.7033112582781458,
      "grad_norm": 1.0448924896854326,
      "learning_rate": 0.0002951672124166841,
      "loss": 2.5469,
      "step": 2572
    },
    {
      "epoch": 1.7039735099337747,
      "grad_norm": 0.9038091239346047,
      "learning_rate": 0.00029515039151381013,
      "loss": 2.4375,
      "step": 2573
    },
    {
      "epoch": 1.704635761589404,
      "grad_norm": 0.8547174134056884,
      "learning_rate": 0.0002951335418694946,
      "loss": 2.3594,
      "step": 2574
    },
    {
      "epoch": 1.705298013245033,
      "grad_norm": 0.9438213909168077,
      "learning_rate": 0.00029511666348707377,
      "loss": 2.625,
      "step": 2575
    },
    {
      "epoch": 1.7059602649006622,
      "grad_norm": 0.9022703357036153,
      "learning_rate": 0.0002950997563698899,
      "loss": 2.6719,
      "step": 2576
    },
    {
      "epoch": 1.7066225165562914,
      "grad_norm": 0.8758259625307955,
      "learning_rate": 0.0002950828205212908,
      "loss": 2.4688,
      "step": 2577
    },
    {
      "epoch": 1.7072847682119205,
      "grad_norm": 0.7597435384810322,
      "learning_rate": 0.00029506585594462985,
      "loss": 2.0,
      "step": 2578
    },
    {
      "epoch": 1.7079470198675497,
      "grad_norm": 0.9595066863042475,
      "learning_rate": 0.00029504886264326635,
      "loss": 2.3438,
      "step": 2579
    },
    {
      "epoch": 1.7086092715231787,
      "grad_norm": 0.8278902823425804,
      "learning_rate": 0.00029503184062056503,
      "loss": 2.4375,
      "step": 2580
    },
    {
      "epoch": 1.709271523178808,
      "grad_norm": 0.9532073703761917,
      "learning_rate": 0.0002950147898798966,
      "loss": 2.4844,
      "step": 2581
    },
    {
      "epoch": 1.709933774834437,
      "grad_norm": 0.9130402306698728,
      "learning_rate": 0.0002949977104246372,
      "loss": 2.4375,
      "step": 2582
    },
    {
      "epoch": 1.7105960264900664,
      "grad_norm": 0.8643657542637402,
      "learning_rate": 0.0002949806022581688,
      "loss": 2.5,
      "step": 2583
    },
    {
      "epoch": 1.7112582781456953,
      "grad_norm": 0.869608509106683,
      "learning_rate": 0.00029496346538387903,
      "loss": 2.125,
      "step": 2584
    },
    {
      "epoch": 1.7119205298013245,
      "grad_norm": 0.9011635701158911,
      "learning_rate": 0.00029494629980516114,
      "loss": 2.7344,
      "step": 2585
    },
    {
      "epoch": 1.7125827814569536,
      "grad_norm": 0.8917203130369887,
      "learning_rate": 0.0002949291055254142,
      "loss": 2.1406,
      "step": 2586
    },
    {
      "epoch": 1.7132450331125828,
      "grad_norm": 0.815564886098106,
      "learning_rate": 0.0002949118825480427,
      "loss": 2.3125,
      "step": 2587
    },
    {
      "epoch": 1.713907284768212,
      "grad_norm": 0.8388413079792099,
      "learning_rate": 0.00029489463087645717,
      "loss": 2.0938,
      "step": 2588
    },
    {
      "epoch": 1.714569536423841,
      "grad_norm": 0.8938948816351111,
      "learning_rate": 0.0002948773505140735,
      "loss": 2.5156,
      "step": 2589
    },
    {
      "epoch": 1.7152317880794703,
      "grad_norm": 1.4641219330235307,
      "learning_rate": 0.0002948600414643135,
      "loss": 2.5469,
      "step": 2590
    },
    {
      "epoch": 1.7158940397350992,
      "grad_norm": 0.9428062940583675,
      "learning_rate": 0.00029484270373060457,
      "loss": 2.5156,
      "step": 2591
    },
    {
      "epoch": 1.7165562913907286,
      "grad_norm": 1.0349241727680316,
      "learning_rate": 0.00029482533731637966,
      "loss": 2.5469,
      "step": 2592
    },
    {
      "epoch": 1.7172185430463576,
      "grad_norm": 0.924445677813016,
      "learning_rate": 0.0002948079422250776,
      "loss": 2.4531,
      "step": 2593
    },
    {
      "epoch": 1.7178807947019867,
      "grad_norm": 0.8739211016556457,
      "learning_rate": 0.0002947905184601429,
      "loss": 2.625,
      "step": 2594
    },
    {
      "epoch": 1.718543046357616,
      "grad_norm": 0.905093205886452,
      "learning_rate": 0.00029477306602502547,
      "loss": 2.3125,
      "step": 2595
    },
    {
      "epoch": 1.719205298013245,
      "grad_norm": 1.1800893862975632,
      "learning_rate": 0.0002947555849231813,
      "loss": 2.5625,
      "step": 2596
    },
    {
      "epoch": 1.7198675496688742,
      "grad_norm": 0.876992072911533,
      "learning_rate": 0.0002947380751580716,
      "loss": 2.6875,
      "step": 2597
    },
    {
      "epoch": 1.7205298013245032,
      "grad_norm": 0.9045128789329269,
      "learning_rate": 0.00029472053673316377,
      "loss": 2.4531,
      "step": 2598
    },
    {
      "epoch": 1.7211920529801326,
      "grad_norm": 1.0076668952204506,
      "learning_rate": 0.00029470296965193046,
      "loss": 2.5312,
      "step": 2599
    },
    {
      "epoch": 1.7218543046357615,
      "grad_norm": 0.8658080755026943,
      "learning_rate": 0.00029468537391785023,
      "loss": 2.1094,
      "step": 2600
    },
    {
      "epoch": 1.7225165562913909,
      "grad_norm": 0.9112029606928851,
      "learning_rate": 0.00029466774953440717,
      "loss": 2.2656,
      "step": 2601
    },
    {
      "epoch": 1.7231788079470198,
      "grad_norm": 0.9508769476165537,
      "learning_rate": 0.0002946500965050911,
      "loss": 2.5469,
      "step": 2602
    },
    {
      "epoch": 1.723841059602649,
      "grad_norm": 0.8880323273655687,
      "learning_rate": 0.00029463241483339763,
      "loss": 2.4844,
      "step": 2603
    },
    {
      "epoch": 1.7245033112582782,
      "grad_norm": 1.0959616783922201,
      "learning_rate": 0.0002946147045228278,
      "loss": 2.5312,
      "step": 2604
    },
    {
      "epoch": 1.7251655629139073,
      "grad_norm": 0.8380873258396921,
      "learning_rate": 0.00029459696557688857,
      "loss": 2.2969,
      "step": 2605
    },
    {
      "epoch": 1.7258278145695365,
      "grad_norm": 0.9496119050145493,
      "learning_rate": 0.00029457919799909244,
      "loss": 2.5625,
      "step": 2606
    },
    {
      "epoch": 1.7264900662251654,
      "grad_norm": 0.8997829824758405,
      "learning_rate": 0.0002945614017929575,
      "loss": 2.5469,
      "step": 2607
    },
    {
      "epoch": 1.7271523178807948,
      "grad_norm": 0.8013575242660411,
      "learning_rate": 0.00029454357696200766,
      "loss": 2.5156,
      "step": 2608
    },
    {
      "epoch": 1.7278145695364238,
      "grad_norm": 0.7601464676044001,
      "learning_rate": 0.00029452572350977246,
      "loss": 2.2812,
      "step": 2609
    },
    {
      "epoch": 1.728476821192053,
      "grad_norm": 0.8357018195397167,
      "learning_rate": 0.00029450784143978705,
      "loss": 2.2812,
      "step": 2610
    },
    {
      "epoch": 1.729139072847682,
      "grad_norm": 0.8779253565084463,
      "learning_rate": 0.0002944899307555924,
      "loss": 2.4219,
      "step": 2611
    },
    {
      "epoch": 1.7298013245033113,
      "grad_norm": 0.8128595901308762,
      "learning_rate": 0.00029447199146073483,
      "loss": 2.2344,
      "step": 2612
    },
    {
      "epoch": 1.7304635761589404,
      "grad_norm": 0.8914357203209446,
      "learning_rate": 0.0002944540235587667,
      "loss": 2.5,
      "step": 2613
    },
    {
      "epoch": 1.7311258278145696,
      "grad_norm": 0.7938264710319881,
      "learning_rate": 0.00029443602705324574,
      "loss": 2.0781,
      "step": 2614
    },
    {
      "epoch": 1.7317880794701987,
      "grad_norm": 0.9561369119063774,
      "learning_rate": 0.00029441800194773555,
      "loss": 2.7031,
      "step": 2615
    },
    {
      "epoch": 1.7324503311258277,
      "grad_norm": 0.7912879176153828,
      "learning_rate": 0.00029439994824580527,
      "loss": 2.0469,
      "step": 2616
    },
    {
      "epoch": 1.733112582781457,
      "grad_norm": 0.9485230747633093,
      "learning_rate": 0.0002943818659510297,
      "loss": 2.375,
      "step": 2617
    },
    {
      "epoch": 1.733774834437086,
      "grad_norm": 0.8938223942657839,
      "learning_rate": 0.00029436375506698947,
      "loss": 2.5625,
      "step": 2618
    },
    {
      "epoch": 1.7344370860927152,
      "grad_norm": 0.8014321293308624,
      "learning_rate": 0.0002943456155972706,
      "loss": 2.5,
      "step": 2619
    },
    {
      "epoch": 1.7350993377483444,
      "grad_norm": 0.776173143016004,
      "learning_rate": 0.000294327447545465,
      "loss": 2.4844,
      "step": 2620
    },
    {
      "epoch": 1.7357615894039735,
      "grad_norm": 0.8492281761739533,
      "learning_rate": 0.00029430925091517013,
      "loss": 2.5781,
      "step": 2621
    },
    {
      "epoch": 1.7364238410596027,
      "grad_norm": 0.9099304754853278,
      "learning_rate": 0.0002942910257099891,
      "loss": 2.125,
      "step": 2622
    },
    {
      "epoch": 1.7370860927152316,
      "grad_norm": 0.8052824648123044,
      "learning_rate": 0.00029427277193353076,
      "loss": 2.4688,
      "step": 2623
    },
    {
      "epoch": 1.737748344370861,
      "grad_norm": 0.851870556251229,
      "learning_rate": 0.0002942544895894096,
      "loss": 2.5469,
      "step": 2624
    },
    {
      "epoch": 1.73841059602649,
      "grad_norm": 0.913991095015386,
      "learning_rate": 0.0002942361786812456,
      "loss": 2.4219,
      "step": 2625
    },
    {
      "epoch": 1.7390728476821193,
      "grad_norm": 0.9821603424791842,
      "learning_rate": 0.0002942178392126646,
      "loss": 2.5938,
      "step": 2626
    },
    {
      "epoch": 1.7397350993377483,
      "grad_norm": 0.7811066994666566,
      "learning_rate": 0.00029419947118729803,
      "loss": 2.3125,
      "step": 2627
    },
    {
      "epoch": 1.7403973509933774,
      "grad_norm": 0.8660721352089412,
      "learning_rate": 0.000294181074608783,
      "loss": 2.4844,
      "step": 2628
    },
    {
      "epoch": 1.7410596026490066,
      "grad_norm": 0.8990310938208633,
      "learning_rate": 0.0002941626494807622,
      "loss": 2.4844,
      "step": 2629
    },
    {
      "epoch": 1.7417218543046358,
      "grad_norm": 0.8567746170510693,
      "learning_rate": 0.0002941441958068841,
      "loss": 2.6562,
      "step": 2630
    },
    {
      "epoch": 1.742384105960265,
      "grad_norm": 0.8568844723559527,
      "learning_rate": 0.00029412571359080256,
      "loss": 2.3594,
      "step": 2631
    },
    {
      "epoch": 1.7430463576158939,
      "grad_norm": 0.8932848395155568,
      "learning_rate": 0.0002941072028361775,
      "loss": 2.4219,
      "step": 2632
    },
    {
      "epoch": 1.7437086092715233,
      "grad_norm": 0.9534666479369729,
      "learning_rate": 0.00029408866354667406,
      "loss": 2.6875,
      "step": 2633
    },
    {
      "epoch": 1.7443708609271522,
      "grad_norm": 0.8953637745281912,
      "learning_rate": 0.0002940700957259633,
      "loss": 2.5312,
      "step": 2634
    },
    {
      "epoch": 1.7450331125827816,
      "grad_norm": 0.8451000041389838,
      "learning_rate": 0.00029405149937772193,
      "loss": 2.5156,
      "step": 2635
    },
    {
      "epoch": 1.7456953642384105,
      "grad_norm": 0.8940103418471532,
      "learning_rate": 0.00029403287450563214,
      "loss": 2.5469,
      "step": 2636
    },
    {
      "epoch": 1.7463576158940397,
      "grad_norm": 0.9426672271862322,
      "learning_rate": 0.000294014221113382,
      "loss": 2.4531,
      "step": 2637
    },
    {
      "epoch": 1.7470198675496689,
      "grad_norm": 0.9385047914825806,
      "learning_rate": 0.00029399553920466493,
      "loss": 2.6406,
      "step": 2638
    },
    {
      "epoch": 1.747682119205298,
      "grad_norm": 0.9116378443931781,
      "learning_rate": 0.00029397682878318026,
      "loss": 2.4062,
      "step": 2639
    },
    {
      "epoch": 1.7483443708609272,
      "grad_norm": 0.8516199524262862,
      "learning_rate": 0.00029395808985263286,
      "loss": 2.1719,
      "step": 2640
    },
    {
      "epoch": 1.7490066225165561,
      "grad_norm": 0.8570796980818687,
      "learning_rate": 0.00029393932241673324,
      "loss": 2.6406,
      "step": 2641
    },
    {
      "epoch": 1.7496688741721855,
      "grad_norm": 0.8343818904281359,
      "learning_rate": 0.00029392052647919756,
      "loss": 2.2969,
      "step": 2642
    },
    {
      "epoch": 1.7503311258278145,
      "grad_norm": 0.8471568027021015,
      "learning_rate": 0.00029390170204374763,
      "loss": 2.2812,
      "step": 2643
    },
    {
      "epoch": 1.7509933774834439,
      "grad_norm": 0.7638677553252173,
      "learning_rate": 0.000293882849114111,
      "loss": 1.9766,
      "step": 2644
    },
    {
      "epoch": 1.7516556291390728,
      "grad_norm": 0.9641437644247945,
      "learning_rate": 0.00029386396769402057,
      "loss": 2.3906,
      "step": 2645
    },
    {
      "epoch": 1.752317880794702,
      "grad_norm": 0.7798963545236832,
      "learning_rate": 0.0002938450577872153,
      "loss": 1.9219,
      "step": 2646
    },
    {
      "epoch": 1.7529801324503311,
      "grad_norm": 0.8140688030377,
      "learning_rate": 0.0002938261193974394,
      "loss": 2.3594,
      "step": 2647
    },
    {
      "epoch": 1.7536423841059603,
      "grad_norm": 0.8572115492263485,
      "learning_rate": 0.00029380715252844293,
      "loss": 2.5625,
      "step": 2648
    },
    {
      "epoch": 1.7543046357615895,
      "grad_norm": 0.9073779247739722,
      "learning_rate": 0.00029378815718398156,
      "loss": 2.5625,
      "step": 2649
    },
    {
      "epoch": 1.7549668874172184,
      "grad_norm": 0.8386030844272973,
      "learning_rate": 0.00029376913336781667,
      "loss": 2.6562,
      "step": 2650
    },
    {
      "epoch": 1.7556291390728478,
      "grad_norm": 0.7525949989711908,
      "learning_rate": 0.000293750081083715,
      "loss": 2.3281,
      "step": 2651
    },
    {
      "epoch": 1.7562913907284767,
      "grad_norm": 0.8514203681062066,
      "learning_rate": 0.0002937310003354493,
      "loss": 2.4062,
      "step": 2652
    },
    {
      "epoch": 1.7569536423841061,
      "grad_norm": 0.8720789337643873,
      "learning_rate": 0.00029371189112679767,
      "loss": 2.4062,
      "step": 2653
    },
    {
      "epoch": 1.757615894039735,
      "grad_norm": 0.8443915840001723,
      "learning_rate": 0.0002936927534615441,
      "loss": 2.4219,
      "step": 2654
    },
    {
      "epoch": 1.7582781456953642,
      "grad_norm": 0.9770159526010178,
      "learning_rate": 0.00029367358734347783,
      "loss": 2.6719,
      "step": 2655
    },
    {
      "epoch": 1.7589403973509934,
      "grad_norm": 0.9070314453618926,
      "learning_rate": 0.00029365439277639416,
      "loss": 2.4219,
      "step": 2656
    },
    {
      "epoch": 1.7596026490066226,
      "grad_norm": 0.8394463706631946,
      "learning_rate": 0.0002936351697640938,
      "loss": 2.0469,
      "step": 2657
    },
    {
      "epoch": 1.7602649006622517,
      "grad_norm": 0.8333935428662533,
      "learning_rate": 0.0002936159183103831,
      "loss": 2.4531,
      "step": 2658
    },
    {
      "epoch": 1.7609271523178807,
      "grad_norm": 0.8649218055828674,
      "learning_rate": 0.00029359663841907403,
      "loss": 2.1406,
      "step": 2659
    },
    {
      "epoch": 1.76158940397351,
      "grad_norm": 0.8941759011656215,
      "learning_rate": 0.00029357733009398426,
      "loss": 2.4062,
      "step": 2660
    },
    {
      "epoch": 1.762251655629139,
      "grad_norm": 1.0074338571514996,
      "learning_rate": 0.0002935579933389371,
      "loss": 2.5,
      "step": 2661
    },
    {
      "epoch": 1.7629139072847684,
      "grad_norm": 0.9041812272556902,
      "learning_rate": 0.00029353862815776146,
      "loss": 2.1094,
      "step": 2662
    },
    {
      "epoch": 1.7635761589403973,
      "grad_norm": 0.8879486594626026,
      "learning_rate": 0.0002935192345542918,
      "loss": 2.3594,
      "step": 2663
    },
    {
      "epoch": 1.7642384105960265,
      "grad_norm": 0.868098083827403,
      "learning_rate": 0.0002934998125323683,
      "loss": 2.3594,
      "step": 2664
    },
    {
      "epoch": 1.7649006622516556,
      "grad_norm": 0.9695870541360891,
      "learning_rate": 0.00029348036209583676,
      "loss": 2.5781,
      "step": 2665
    },
    {
      "epoch": 1.7655629139072848,
      "grad_norm": 0.8948715250795777,
      "learning_rate": 0.0002934608832485485,
      "loss": 2.4531,
      "step": 2666
    },
    {
      "epoch": 1.766225165562914,
      "grad_norm": 0.8598146439782913,
      "learning_rate": 0.00029344137599436067,
      "loss": 2.7188,
      "step": 2667
    },
    {
      "epoch": 1.766887417218543,
      "grad_norm": 0.8241243856675451,
      "learning_rate": 0.00029342184033713596,
      "loss": 2.4375,
      "step": 2668
    },
    {
      "epoch": 1.7675496688741723,
      "grad_norm": 0.8831663029574784,
      "learning_rate": 0.00029340227628074254,
      "loss": 2.5625,
      "step": 2669
    },
    {
      "epoch": 1.7682119205298013,
      "grad_norm": 0.853652526509957,
      "learning_rate": 0.0002933826838290543,
      "loss": 2.25,
      "step": 2670
    },
    {
      "epoch": 1.7688741721854304,
      "grad_norm": 0.8797026123943295,
      "learning_rate": 0.0002933630629859509,
      "loss": 2.7188,
      "step": 2671
    },
    {
      "epoch": 1.7695364238410596,
      "grad_norm": 0.8237761136968011,
      "learning_rate": 0.00029334341375531744,
      "loss": 2.5469,
      "step": 2672
    },
    {
      "epoch": 1.7701986754966887,
      "grad_norm": 0.9808214391095679,
      "learning_rate": 0.00029332373614104467,
      "loss": 2.3125,
      "step": 2673
    },
    {
      "epoch": 1.770860927152318,
      "grad_norm": 0.8866948571482459,
      "learning_rate": 0.00029330403014702895,
      "loss": 2.2969,
      "step": 2674
    },
    {
      "epoch": 1.771523178807947,
      "grad_norm": 0.8469823823469748,
      "learning_rate": 0.0002932842957771724,
      "loss": 2.375,
      "step": 2675
    },
    {
      "epoch": 1.7721854304635762,
      "grad_norm": 0.7621893699495792,
      "learning_rate": 0.0002932645330353825,
      "loss": 1.8047,
      "step": 2676
    },
    {
      "epoch": 1.7728476821192052,
      "grad_norm": 0.8964418884166597,
      "learning_rate": 0.00029324474192557264,
      "loss": 2.5156,
      "step": 2677
    },
    {
      "epoch": 1.7735099337748346,
      "grad_norm": 0.9008802138384604,
      "learning_rate": 0.00029322492245166163,
      "loss": 2.4219,
      "step": 2678
    },
    {
      "epoch": 1.7741721854304635,
      "grad_norm": 0.9406056613369347,
      "learning_rate": 0.00029320507461757394,
      "loss": 2.5312,
      "step": 2679
    },
    {
      "epoch": 1.7748344370860927,
      "grad_norm": 0.8856409432493761,
      "learning_rate": 0.0002931851984272397,
      "loss": 2.4375,
      "step": 2680
    },
    {
      "epoch": 1.7754966887417218,
      "grad_norm": 0.7864502570816322,
      "learning_rate": 0.0002931652938845946,
      "loss": 1.9922,
      "step": 2681
    },
    {
      "epoch": 1.776158940397351,
      "grad_norm": 0.8648185757329448,
      "learning_rate": 0.00029314536099357994,
      "loss": 2.5625,
      "step": 2682
    },
    {
      "epoch": 1.7768211920529802,
      "grad_norm": 0.8730847341985265,
      "learning_rate": 0.00029312539975814275,
      "loss": 2.4688,
      "step": 2683
    },
    {
      "epoch": 1.7774834437086091,
      "grad_norm": 0.9300643598384762,
      "learning_rate": 0.0002931054101822355,
      "loss": 2.2969,
      "step": 2684
    },
    {
      "epoch": 1.7781456953642385,
      "grad_norm": 1.0270446961806259,
      "learning_rate": 0.0002930853922698164,
      "loss": 2.625,
      "step": 2685
    },
    {
      "epoch": 1.7788079470198674,
      "grad_norm": 0.8484477793427707,
      "learning_rate": 0.0002930653460248492,
      "loss": 2.4688,
      "step": 2686
    },
    {
      "epoch": 1.7794701986754968,
      "grad_norm": 0.8728264532974931,
      "learning_rate": 0.0002930452714513033,
      "loss": 2.5312,
      "step": 2687
    },
    {
      "epoch": 1.7801324503311258,
      "grad_norm": 0.896251269940751,
      "learning_rate": 0.0002930251685531537,
      "loss": 2.6719,
      "step": 2688
    },
    {
      "epoch": 1.780794701986755,
      "grad_norm": 0.8964313104850266,
      "learning_rate": 0.00029300503733438097,
      "loss": 2.6875,
      "step": 2689
    },
    {
      "epoch": 1.781456953642384,
      "grad_norm": 0.923668422300788,
      "learning_rate": 0.00029298487779897135,
      "loss": 2.3438,
      "step": 2690
    },
    {
      "epoch": 1.7821192052980133,
      "grad_norm": 0.9455761584563044,
      "learning_rate": 0.00029296468995091665,
      "loss": 2.7969,
      "step": 2691
    },
    {
      "epoch": 1.7827814569536424,
      "grad_norm": 0.7411616804629328,
      "learning_rate": 0.0002929444737942143,
      "loss": 2.375,
      "step": 2692
    },
    {
      "epoch": 1.7834437086092714,
      "grad_norm": 0.8047181363746244,
      "learning_rate": 0.0002929242293328674,
      "loss": 1.9844,
      "step": 2693
    },
    {
      "epoch": 1.7841059602649008,
      "grad_norm": 0.8629123621638246,
      "learning_rate": 0.00029290395657088447,
      "loss": 2.3906,
      "step": 2694
    },
    {
      "epoch": 1.7847682119205297,
      "grad_norm": 1.0018639065643204,
      "learning_rate": 0.0002928836555122798,
      "loss": 2.375,
      "step": 2695
    },
    {
      "epoch": 1.785430463576159,
      "grad_norm": 0.8634908173056695,
      "learning_rate": 0.00029286332616107317,
      "loss": 2.5312,
      "step": 2696
    },
    {
      "epoch": 1.786092715231788,
      "grad_norm": 0.7548402617398009,
      "learning_rate": 0.0002928429685212902,
      "loss": 1.9453,
      "step": 2697
    },
    {
      "epoch": 1.7867549668874172,
      "grad_norm": 0.8374726248547408,
      "learning_rate": 0.0002928225825969618,
      "loss": 2.4062,
      "step": 2698
    },
    {
      "epoch": 1.7874172185430464,
      "grad_norm": 0.9762234618447262,
      "learning_rate": 0.00029280216839212456,
      "loss": 2.6562,
      "step": 2699
    },
    {
      "epoch": 1.7880794701986755,
      "grad_norm": 0.8263970942745864,
      "learning_rate": 0.00029278172591082086,
      "loss": 2.4219,
      "step": 2700
    },
    {
      "epoch": 1.7887417218543047,
      "grad_norm": 0.8033080088515382,
      "learning_rate": 0.0002927612551570985,
      "loss": 2.4844,
      "step": 2701
    },
    {
      "epoch": 1.7894039735099336,
      "grad_norm": 0.8995662993787829,
      "learning_rate": 0.00029274075613501093,
      "loss": 2.5312,
      "step": 2702
    },
    {
      "epoch": 1.790066225165563,
      "grad_norm": 0.7901400735653719,
      "learning_rate": 0.0002927202288486171,
      "loss": 1.9688,
      "step": 2703
    },
    {
      "epoch": 1.790728476821192,
      "grad_norm": 0.8241622085419643,
      "learning_rate": 0.00029269967330198177,
      "loss": 2.6875,
      "step": 2704
    },
    {
      "epoch": 1.7913907284768213,
      "grad_norm": 0.841941526994542,
      "learning_rate": 0.00029267908949917517,
      "loss": 2.5625,
      "step": 2705
    },
    {
      "epoch": 1.7920529801324503,
      "grad_norm": 0.9325690770359905,
      "learning_rate": 0.00029265847744427303,
      "loss": 2.0938,
      "step": 2706
    },
    {
      "epoch": 1.7927152317880795,
      "grad_norm": 0.9036236428832487,
      "learning_rate": 0.0002926378371413568,
      "loss": 2.5312,
      "step": 2707
    },
    {
      "epoch": 1.7933774834437086,
      "grad_norm": 0.8048091259406812,
      "learning_rate": 0.0002926171685945136,
      "loss": 2.2188,
      "step": 2708
    },
    {
      "epoch": 1.7940397350993378,
      "grad_norm": 0.8784277433348321,
      "learning_rate": 0.0002925964718078359,
      "loss": 2.6875,
      "step": 2709
    },
    {
      "epoch": 1.794701986754967,
      "grad_norm": 0.7946439670008398,
      "learning_rate": 0.000292575746785422,
      "loss": 2.3438,
      "step": 2710
    },
    {
      "epoch": 1.795364238410596,
      "grad_norm": 0.9717756382285718,
      "learning_rate": 0.0002925549935313757,
      "loss": 2.7344,
      "step": 2711
    },
    {
      "epoch": 1.7960264900662253,
      "grad_norm": 0.8252984118908527,
      "learning_rate": 0.0002925342120498063,
      "loss": 2.6562,
      "step": 2712
    },
    {
      "epoch": 1.7966887417218542,
      "grad_norm": 0.7426479975844986,
      "learning_rate": 0.0002925134023448288,
      "loss": 2.2344,
      "step": 2713
    },
    {
      "epoch": 1.7973509933774836,
      "grad_norm": 0.8283589163709337,
      "learning_rate": 0.0002924925644205638,
      "loss": 2.6094,
      "step": 2714
    },
    {
      "epoch": 1.7980132450331126,
      "grad_norm": 0.9166237132175525,
      "learning_rate": 0.0002924716982811374,
      "loss": 2.2812,
      "step": 2715
    },
    {
      "epoch": 1.7986754966887417,
      "grad_norm": 0.7898337290824426,
      "learning_rate": 0.0002924508039306814,
      "loss": 2.4375,
      "step": 2716
    },
    {
      "epoch": 1.7993377483443709,
      "grad_norm": 0.816796393783639,
      "learning_rate": 0.000292429881373333,
      "loss": 2.4219,
      "step": 2717
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8725703709079965,
      "learning_rate": 0.00029240893061323526,
      "loss": 2.6094,
      "step": 2718
    },
    {
      "epoch": 1.8006622516556292,
      "grad_norm": 0.8162546934214674,
      "learning_rate": 0.00029238795165453655,
      "loss": 2.6406,
      "step": 2719
    },
    {
      "epoch": 1.8013245033112582,
      "grad_norm": 0.8564171661760277,
      "learning_rate": 0.000292366944501391,
      "loss": 2.375,
      "step": 2720
    },
    {
      "epoch": 1.8019867549668875,
      "grad_norm": 0.7311155492659066,
      "learning_rate": 0.00029234590915795823,
      "loss": 2.3125,
      "step": 2721
    },
    {
      "epoch": 1.8026490066225165,
      "grad_norm": 0.9270211585795599,
      "learning_rate": 0.00029232484562840357,
      "loss": 2.5625,
      "step": 2722
    },
    {
      "epoch": 1.8033112582781456,
      "grad_norm": 0.822207187609188,
      "learning_rate": 0.0002923037539168977,
      "loss": 2.4688,
      "step": 2723
    },
    {
      "epoch": 1.8039735099337748,
      "grad_norm": 0.8312581112364031,
      "learning_rate": 0.00029228263402761713,
      "loss": 2.4531,
      "step": 2724
    },
    {
      "epoch": 1.804635761589404,
      "grad_norm": 0.8295052516776861,
      "learning_rate": 0.0002922614859647438,
      "loss": 2.2812,
      "step": 2725
    },
    {
      "epoch": 1.8052980132450331,
      "grad_norm": 0.8563373302409606,
      "learning_rate": 0.00029224030973246524,
      "loss": 2.4844,
      "step": 2726
    },
    {
      "epoch": 1.8059602649006623,
      "grad_norm": 0.9192641508105985,
      "learning_rate": 0.00029221910533497466,
      "loss": 2.2656,
      "step": 2727
    },
    {
      "epoch": 1.8066225165562915,
      "grad_norm": 0.8400966643882242,
      "learning_rate": 0.00029219787277647073,
      "loss": 2.3281,
      "step": 2728
    },
    {
      "epoch": 1.8072847682119204,
      "grad_norm": 0.8111631747747882,
      "learning_rate": 0.0002921766120611578,
      "loss": 2.4062,
      "step": 2729
    },
    {
      "epoch": 1.8079470198675498,
      "grad_norm": 0.8757351986689719,
      "learning_rate": 0.00029215532319324555,
      "loss": 1.9844,
      "step": 2730
    },
    {
      "epoch": 1.8086092715231787,
      "grad_norm": 0.8410628802959517,
      "learning_rate": 0.0002921340061769497,
      "loss": 2.5156,
      "step": 2731
    },
    {
      "epoch": 1.809271523178808,
      "grad_norm": 0.774696143741158,
      "learning_rate": 0.000292112661016491,
      "loss": 1.9688,
      "step": 2732
    },
    {
      "epoch": 1.809933774834437,
      "grad_norm": 0.8145399891170766,
      "learning_rate": 0.0002920912877160962,
      "loss": 2.375,
      "step": 2733
    },
    {
      "epoch": 1.8105960264900662,
      "grad_norm": 0.7943576265657718,
      "learning_rate": 0.00029206988627999736,
      "loss": 2.3906,
      "step": 2734
    },
    {
      "epoch": 1.8112582781456954,
      "grad_norm": 0.9874660056277332,
      "learning_rate": 0.0002920484567124323,
      "loss": 2.5,
      "step": 2735
    },
    {
      "epoch": 1.8119205298013243,
      "grad_norm": 0.7744255413369399,
      "learning_rate": 0.0002920269990176443,
      "loss": 2.2812,
      "step": 2736
    },
    {
      "epoch": 1.8125827814569537,
      "grad_norm": 0.8994562161872767,
      "learning_rate": 0.00029200551319988213,
      "loss": 2.2188,
      "step": 2737
    },
    {
      "epoch": 1.8132450331125827,
      "grad_norm": 0.7936106952254149,
      "learning_rate": 0.0002919839992634004,
      "loss": 2.2031,
      "step": 2738
    },
    {
      "epoch": 1.813907284768212,
      "grad_norm": 0.9217683813124031,
      "learning_rate": 0.00029196245721245897,
      "loss": 2.625,
      "step": 2739
    },
    {
      "epoch": 1.814569536423841,
      "grad_norm": 0.7938510823013709,
      "learning_rate": 0.0002919408870513235,
      "loss": 2.2812,
      "step": 2740
    },
    {
      "epoch": 1.8152317880794702,
      "grad_norm": 0.7857139120627535,
      "learning_rate": 0.0002919192887842651,
      "loss": 2.3125,
      "step": 2741
    },
    {
      "epoch": 1.8158940397350993,
      "grad_norm": 0.8964090478965278,
      "learning_rate": 0.0002918976624155604,
      "loss": 2.4219,
      "step": 2742
    },
    {
      "epoch": 1.8165562913907285,
      "grad_norm": 0.8341573851550185,
      "learning_rate": 0.00029187600794949175,
      "loss": 1.9062,
      "step": 2743
    },
    {
      "epoch": 1.8172185430463577,
      "grad_norm": 0.7623273457320935,
      "learning_rate": 0.000291854325390347,
      "loss": 2.0781,
      "step": 2744
    },
    {
      "epoch": 1.8178807947019866,
      "grad_norm": 0.8184260042359358,
      "learning_rate": 0.0002918326147424195,
      "loss": 2.3906,
      "step": 2745
    },
    {
      "epoch": 1.818543046357616,
      "grad_norm": 0.8538983463350875,
      "learning_rate": 0.0002918108760100083,
      "loss": 2.2812,
      "step": 2746
    },
    {
      "epoch": 1.819205298013245,
      "grad_norm": 0.9150232774235978,
      "learning_rate": 0.00029178910919741774,
      "loss": 2.2031,
      "step": 2747
    },
    {
      "epoch": 1.8198675496688743,
      "grad_norm": 0.9499602497781707,
      "learning_rate": 0.00029176731430895805,
      "loss": 2.6406,
      "step": 2748
    },
    {
      "epoch": 1.8205298013245033,
      "grad_norm": 0.8641877906862377,
      "learning_rate": 0.0002917454913489448,
      "loss": 2.3594,
      "step": 2749
    },
    {
      "epoch": 1.8211920529801324,
      "grad_norm": 0.9318545964926954,
      "learning_rate": 0.0002917236403216992,
      "loss": 2.4375,
      "step": 2750
    },
    {
      "epoch": 1.8218543046357616,
      "grad_norm": 0.9360453402692742,
      "learning_rate": 0.000291701761231548,
      "loss": 2.4062,
      "step": 2751
    },
    {
      "epoch": 1.8225165562913908,
      "grad_norm": 0.8732893664657312,
      "learning_rate": 0.0002916798540828236,
      "loss": 2.2656,
      "step": 2752
    },
    {
      "epoch": 1.82317880794702,
      "grad_norm": 0.9136496531802694,
      "learning_rate": 0.00029165791887986365,
      "loss": 2.4062,
      "step": 2753
    },
    {
      "epoch": 1.8238410596026489,
      "grad_norm": 0.9638747997005405,
      "learning_rate": 0.0002916359556270118,
      "loss": 2.5,
      "step": 2754
    },
    {
      "epoch": 1.8245033112582782,
      "grad_norm": 0.8489194087304295,
      "learning_rate": 0.00029161396432861693,
      "loss": 2.6406,
      "step": 2755
    },
    {
      "epoch": 1.8251655629139072,
      "grad_norm": 0.7840490609945583,
      "learning_rate": 0.0002915919449890336,
      "loss": 2.25,
      "step": 2756
    },
    {
      "epoch": 1.8258278145695366,
      "grad_norm": 0.8587547721510731,
      "learning_rate": 0.0002915698976126218,
      "loss": 2.2969,
      "step": 2757
    },
    {
      "epoch": 1.8264900662251655,
      "grad_norm": 0.8312472589776129,
      "learning_rate": 0.0002915478222037472,
      "loss": 2.5781,
      "step": 2758
    },
    {
      "epoch": 1.8271523178807947,
      "grad_norm": 0.8353897774162931,
      "learning_rate": 0.00029152571876678106,
      "loss": 1.9531,
      "step": 2759
    },
    {
      "epoch": 1.8278145695364238,
      "grad_norm": 0.8586402708690111,
      "learning_rate": 0.0002915035873061001,
      "loss": 2.375,
      "step": 2760
    },
    {
      "epoch": 1.828476821192053,
      "grad_norm": 0.8391416286542711,
      "learning_rate": 0.00029148142782608647,
      "loss": 2.5938,
      "step": 2761
    },
    {
      "epoch": 1.8291390728476822,
      "grad_norm": 0.8248537601060804,
      "learning_rate": 0.0002914592403311282,
      "loss": 2.3125,
      "step": 2762
    },
    {
      "epoch": 1.8298013245033111,
      "grad_norm": 0.8553108737350404,
      "learning_rate": 0.0002914370248256185,
      "loss": 2.3281,
      "step": 2763
    },
    {
      "epoch": 1.8304635761589405,
      "grad_norm": 0.8439215284704801,
      "learning_rate": 0.0002914147813139564,
      "loss": 2.4688,
      "step": 2764
    },
    {
      "epoch": 1.8311258278145695,
      "grad_norm": 0.8965129175885984,
      "learning_rate": 0.0002913925098005463,
      "loss": 1.9609,
      "step": 2765
    },
    {
      "epoch": 1.8317880794701988,
      "grad_norm": 0.8446869236957127,
      "learning_rate": 0.00029137021028979827,
      "loss": 2.3125,
      "step": 2766
    },
    {
      "epoch": 1.8324503311258278,
      "grad_norm": 0.8929966131445781,
      "learning_rate": 0.00029134788278612785,
      "loss": 2.6406,
      "step": 2767
    },
    {
      "epoch": 1.833112582781457,
      "grad_norm": 0.8111008079530692,
      "learning_rate": 0.0002913255272939561,
      "loss": 2.4062,
      "step": 2768
    },
    {
      "epoch": 1.833774834437086,
      "grad_norm": 0.7794080271198303,
      "learning_rate": 0.0002913031438177098,
      "loss": 2.2656,
      "step": 2769
    },
    {
      "epoch": 1.8344370860927153,
      "grad_norm": 0.7882965872198856,
      "learning_rate": 0.0002912807323618209,
      "loss": 2.1562,
      "step": 2770
    },
    {
      "epoch": 1.8350993377483444,
      "grad_norm": 0.8327533274924267,
      "learning_rate": 0.00029125829293072733,
      "loss": 2.3906,
      "step": 2771
    },
    {
      "epoch": 1.8357615894039734,
      "grad_norm": 0.8071806382650956,
      "learning_rate": 0.0002912358255288723,
      "loss": 2.4844,
      "step": 2772
    },
    {
      "epoch": 1.8364238410596028,
      "grad_norm": 0.8507806116229872,
      "learning_rate": 0.00029121333016070457,
      "loss": 2.8125,
      "step": 2773
    },
    {
      "epoch": 1.8370860927152317,
      "grad_norm": 0.866661720911409,
      "learning_rate": 0.00029119080683067853,
      "loss": 2.3906,
      "step": 2774
    },
    {
      "epoch": 1.8377483443708609,
      "grad_norm": 0.857101530741126,
      "learning_rate": 0.000291168255543254,
      "loss": 2.3594,
      "step": 2775
    },
    {
      "epoch": 1.83841059602649,
      "grad_norm": 0.8738788626617696,
      "learning_rate": 0.0002911456763028965,
      "loss": 2.5312,
      "step": 2776
    },
    {
      "epoch": 1.8390728476821192,
      "grad_norm": 0.7948734270727188,
      "learning_rate": 0.0002911230691140769,
      "loss": 2.5625,
      "step": 2777
    },
    {
      "epoch": 1.8397350993377484,
      "grad_norm": 0.9133355256492385,
      "learning_rate": 0.00029110043398127163,
      "loss": 2.4531,
      "step": 2778
    },
    {
      "epoch": 1.8403973509933775,
      "grad_norm": 0.8398510864845652,
      "learning_rate": 0.0002910777709089628,
      "loss": 2.2812,
      "step": 2779
    },
    {
      "epoch": 1.8410596026490067,
      "grad_norm": 0.805385989057586,
      "learning_rate": 0.00029105507990163796,
      "loss": 2.375,
      "step": 2780
    },
    {
      "epoch": 1.8417218543046356,
      "grad_norm": 0.900340262713152,
      "learning_rate": 0.0002910323609637901,
      "loss": 2.5156,
      "step": 2781
    },
    {
      "epoch": 1.842384105960265,
      "grad_norm": 0.8896297431090129,
      "learning_rate": 0.0002910096140999179,
      "loss": 2.3906,
      "step": 2782
    },
    {
      "epoch": 1.843046357615894,
      "grad_norm": 0.7898760945016374,
      "learning_rate": 0.0002909868393145255,
      "loss": 2.4219,
      "step": 2783
    },
    {
      "epoch": 1.8437086092715231,
      "grad_norm": 0.8305570326198095,
      "learning_rate": 0.0002909640366121225,
      "loss": 2.5312,
      "step": 2784
    },
    {
      "epoch": 1.8443708609271523,
      "grad_norm": 0.753343651461002,
      "learning_rate": 0.00029094120599722423,
      "loss": 2.4219,
      "step": 2785
    },
    {
      "epoch": 1.8450331125827815,
      "grad_norm": 0.7727513400864662,
      "learning_rate": 0.0002909183474743513,
      "loss": 2.625,
      "step": 2786
    },
    {
      "epoch": 1.8456953642384106,
      "grad_norm": 0.8796627379549826,
      "learning_rate": 0.00029089546104803005,
      "loss": 2.5469,
      "step": 2787
    },
    {
      "epoch": 1.8463576158940396,
      "grad_norm": 0.7529846631125316,
      "learning_rate": 0.0002908725467227921,
      "loss": 2.375,
      "step": 2788
    },
    {
      "epoch": 1.847019867549669,
      "grad_norm": 0.7336721636830027,
      "learning_rate": 0.00029084960450317487,
      "loss": 2.4219,
      "step": 2789
    },
    {
      "epoch": 1.847682119205298,
      "grad_norm": 0.8028387550156288,
      "learning_rate": 0.00029082663439372125,
      "loss": 2.4062,
      "step": 2790
    },
    {
      "epoch": 1.8483443708609273,
      "grad_norm": 0.8345093726243478,
      "learning_rate": 0.00029080363639897937,
      "loss": 2.3125,
      "step": 2791
    },
    {
      "epoch": 1.8490066225165562,
      "grad_norm": 0.7227866765451151,
      "learning_rate": 0.00029078061052350337,
      "loss": 1.9297,
      "step": 2792
    },
    {
      "epoch": 1.8496688741721854,
      "grad_norm": 0.8835945420530996,
      "learning_rate": 0.0002907575567718523,
      "loss": 2.5469,
      "step": 2793
    },
    {
      "epoch": 1.8503311258278146,
      "grad_norm": 0.7774418813521291,
      "learning_rate": 0.00029073447514859144,
      "loss": 2.3281,
      "step": 2794
    },
    {
      "epoch": 1.8509933774834437,
      "grad_norm": 0.797442563911051,
      "learning_rate": 0.0002907113656582909,
      "loss": 2.3438,
      "step": 2795
    },
    {
      "epoch": 1.851655629139073,
      "grad_norm": 0.7637663790956345,
      "learning_rate": 0.0002906882283055268,
      "loss": 2.5156,
      "step": 2796
    },
    {
      "epoch": 1.8523178807947018,
      "grad_norm": 0.8338452761091721,
      "learning_rate": 0.00029066506309488057,
      "loss": 2.3281,
      "step": 2797
    },
    {
      "epoch": 1.8529801324503312,
      "grad_norm": 0.8432857484155026,
      "learning_rate": 0.00029064187003093923,
      "loss": 2.4375,
      "step": 2798
    },
    {
      "epoch": 1.8536423841059602,
      "grad_norm": 0.7904930686909447,
      "learning_rate": 0.0002906186491182951,
      "loss": 2.4688,
      "step": 2799
    },
    {
      "epoch": 1.8543046357615895,
      "grad_norm": 0.7115944270237506,
      "learning_rate": 0.00029059540036154637,
      "loss": 2.2344,
      "step": 2800
    },
    {
      "epoch": 1.8549668874172185,
      "grad_norm": 0.8457780209832122,
      "learning_rate": 0.0002905721237652965,
      "loss": 2.6094,
      "step": 2801
    },
    {
      "epoch": 1.8556291390728477,
      "grad_norm": 0.8419772186843811,
      "learning_rate": 0.0002905488193341545,
      "loss": 2.0312,
      "step": 2802
    },
    {
      "epoch": 1.8562913907284768,
      "grad_norm": 0.8015757635982855,
      "learning_rate": 0.00029052548707273493,
      "loss": 2.4219,
      "step": 2803
    },
    {
      "epoch": 1.856953642384106,
      "grad_norm": 0.829282490360113,
      "learning_rate": 0.0002905021269856579,
      "loss": 2.3594,
      "step": 2804
    },
    {
      "epoch": 1.8576158940397351,
      "grad_norm": 0.8012979939392302,
      "learning_rate": 0.00029047873907754883,
      "loss": 2.3438,
      "step": 2805
    },
    {
      "epoch": 1.858278145695364,
      "grad_norm": 0.8157340250427647,
      "learning_rate": 0.00029045532335303896,
      "loss": 2.375,
      "step": 2806
    },
    {
      "epoch": 1.8589403973509935,
      "grad_norm": 0.7695781997447553,
      "learning_rate": 0.00029043187981676476,
      "loss": 2.4375,
      "step": 2807
    },
    {
      "epoch": 1.8596026490066224,
      "grad_norm": 0.9213249435618963,
      "learning_rate": 0.00029040840847336833,
      "loss": 2.6094,
      "step": 2808
    },
    {
      "epoch": 1.8602649006622518,
      "grad_norm": 0.8969288540602731,
      "learning_rate": 0.00029038490932749734,
      "loss": 2.0,
      "step": 2809
    },
    {
      "epoch": 1.8609271523178808,
      "grad_norm": 0.8788859216430591,
      "learning_rate": 0.00029036138238380484,
      "loss": 2.6719,
      "step": 2810
    },
    {
      "epoch": 1.86158940397351,
      "grad_norm": 0.8576044046619875,
      "learning_rate": 0.00029033782764694945,
      "loss": 2.5469,
      "step": 2811
    },
    {
      "epoch": 1.862251655629139,
      "grad_norm": 0.7872100352690468,
      "learning_rate": 0.00029031424512159523,
      "loss": 2.4219,
      "step": 2812
    },
    {
      "epoch": 1.8629139072847682,
      "grad_norm": 0.8703829713620044,
      "learning_rate": 0.0002902906348124119,
      "loss": 2.5469,
      "step": 2813
    },
    {
      "epoch": 1.8635761589403974,
      "grad_norm": 0.8403564396988256,
      "learning_rate": 0.0002902669967240744,
      "loss": 2.3281,
      "step": 2814
    },
    {
      "epoch": 1.8642384105960264,
      "grad_norm": 0.8660958068289178,
      "learning_rate": 0.00029024333086126344,
      "loss": 2.8438,
      "step": 2815
    },
    {
      "epoch": 1.8649006622516557,
      "grad_norm": 0.786049768033528,
      "learning_rate": 0.0002902196372286652,
      "loss": 2.375,
      "step": 2816
    },
    {
      "epoch": 1.8655629139072847,
      "grad_norm": 0.7491479307835897,
      "learning_rate": 0.0002901959158309712,
      "loss": 2.3281,
      "step": 2817
    },
    {
      "epoch": 1.866225165562914,
      "grad_norm": 0.836145636419037,
      "learning_rate": 0.00029017216667287857,
      "loss": 2.5938,
      "step": 2818
    },
    {
      "epoch": 1.866887417218543,
      "grad_norm": 0.8387925072746757,
      "learning_rate": 0.0002901483897590899,
      "loss": 2.3125,
      "step": 2819
    },
    {
      "epoch": 1.8675496688741722,
      "grad_norm": 0.917939447487194,
      "learning_rate": 0.00029012458509431333,
      "loss": 2.4844,
      "step": 2820
    },
    {
      "epoch": 1.8682119205298013,
      "grad_norm": 0.7844220582433,
      "learning_rate": 0.00029010075268326246,
      "loss": 2.3125,
      "step": 2821
    },
    {
      "epoch": 1.8688741721854305,
      "grad_norm": 0.8885516729095283,
      "learning_rate": 0.00029007689253065627,
      "loss": 2.5469,
      "step": 2822
    },
    {
      "epoch": 1.8695364238410597,
      "grad_norm": 0.8341378403258463,
      "learning_rate": 0.0002900530046412195,
      "loss": 2.3438,
      "step": 2823
    },
    {
      "epoch": 1.8701986754966886,
      "grad_norm": 0.7612362035962441,
      "learning_rate": 0.00029002908901968217,
      "loss": 2.3906,
      "step": 2824
    },
    {
      "epoch": 1.870860927152318,
      "grad_norm": 0.821570814231991,
      "learning_rate": 0.00029000514567077977,
      "loss": 2.4531,
      "step": 2825
    },
    {
      "epoch": 1.871523178807947,
      "grad_norm": 0.8354198413224346,
      "learning_rate": 0.0002899811745992535,
      "loss": 2.375,
      "step": 2826
    },
    {
      "epoch": 1.872185430463576,
      "grad_norm": 0.8935152760523608,
      "learning_rate": 0.0002899571758098497,
      "loss": 2.2656,
      "step": 2827
    },
    {
      "epoch": 1.8728476821192053,
      "grad_norm": 0.787576569676333,
      "learning_rate": 0.00028993314930732066,
      "loss": 2.3438,
      "step": 2828
    },
    {
      "epoch": 1.8735099337748344,
      "grad_norm": 0.7720794518137264,
      "learning_rate": 0.00028990909509642377,
      "loss": 2.5156,
      "step": 2829
    },
    {
      "epoch": 1.8741721854304636,
      "grad_norm": 0.904941294970232,
      "learning_rate": 0.00028988501318192195,
      "loss": 2.375,
      "step": 2830
    },
    {
      "epoch": 1.8748344370860928,
      "grad_norm": 0.8154539387482661,
      "learning_rate": 0.00028986090356858385,
      "loss": 2.4375,
      "step": 2831
    },
    {
      "epoch": 1.875496688741722,
      "grad_norm": 0.8130472182366187,
      "learning_rate": 0.0002898367662611834,
      "loss": 2.4844,
      "step": 2832
    },
    {
      "epoch": 1.8761589403973509,
      "grad_norm": 0.8462639729876643,
      "learning_rate": 0.0002898126012645,
      "loss": 2.4688,
      "step": 2833
    },
    {
      "epoch": 1.8768211920529803,
      "grad_norm": 0.9254018964305748,
      "learning_rate": 0.00028978840858331873,
      "loss": 2.5469,
      "step": 2834
    },
    {
      "epoch": 1.8774834437086092,
      "grad_norm": 0.8757856654895679,
      "learning_rate": 0.0002897641882224299,
      "loss": 2.5,
      "step": 2835
    },
    {
      "epoch": 1.8781456953642384,
      "grad_norm": 0.8511256546954842,
      "learning_rate": 0.00028973994018662946,
      "loss": 2.5156,
      "step": 2836
    },
    {
      "epoch": 1.8788079470198675,
      "grad_norm": 0.8519454121459769,
      "learning_rate": 0.0002897156644807188,
      "loss": 2.5469,
      "step": 2837
    },
    {
      "epoch": 1.8794701986754967,
      "grad_norm": 0.8098819660902985,
      "learning_rate": 0.0002896913611095047,
      "loss": 2.6094,
      "step": 2838
    },
    {
      "epoch": 1.8801324503311259,
      "grad_norm": 0.909954702721027,
      "learning_rate": 0.00028966703007779975,
      "loss": 2.5625,
      "step": 2839
    },
    {
      "epoch": 1.8807947019867548,
      "grad_norm": 0.865144621946133,
      "learning_rate": 0.0002896426713904215,
      "loss": 2.4219,
      "step": 2840
    },
    {
      "epoch": 1.8814569536423842,
      "grad_norm": 0.8324759325972422,
      "learning_rate": 0.00028961828505219344,
      "loss": 2.4531,
      "step": 2841
    },
    {
      "epoch": 1.8821192052980131,
      "grad_norm": 0.8400855033350236,
      "learning_rate": 0.00028959387106794416,
      "loss": 1.8984,
      "step": 2842
    },
    {
      "epoch": 1.8827814569536425,
      "grad_norm": 0.9236805260648048,
      "learning_rate": 0.00028956942944250813,
      "loss": 2.6406,
      "step": 2843
    },
    {
      "epoch": 1.8834437086092715,
      "grad_norm": 0.8034672820521745,
      "learning_rate": 0.0002895449601807249,
      "loss": 2.5156,
      "step": 2844
    },
    {
      "epoch": 1.8841059602649006,
      "grad_norm": 0.768152749955267,
      "learning_rate": 0.00028952046328743975,
      "loss": 2.0625,
      "step": 2845
    },
    {
      "epoch": 1.8847682119205298,
      "grad_norm": 0.815051255854314,
      "learning_rate": 0.0002894959387675033,
      "loss": 2.5156,
      "step": 2846
    },
    {
      "epoch": 1.885430463576159,
      "grad_norm": 0.9056361646954801,
      "learning_rate": 0.0002894713866257717,
      "loss": 2.4844,
      "step": 2847
    },
    {
      "epoch": 1.8860927152317881,
      "grad_norm": 0.7612968134501811,
      "learning_rate": 0.00028944680686710656,
      "loss": 2.3281,
      "step": 2848
    },
    {
      "epoch": 1.886754966887417,
      "grad_norm": 0.8240153449282337,
      "learning_rate": 0.000289422199496375,
      "loss": 2.4844,
      "step": 2849
    },
    {
      "epoch": 1.8874172185430464,
      "grad_norm": 0.803792293248741,
      "learning_rate": 0.00028939756451844945,
      "loss": 2.375,
      "step": 2850
    },
    {
      "epoch": 1.8880794701986754,
      "grad_norm": 0.8728967799364742,
      "learning_rate": 0.000289372901938208,
      "loss": 2.6094,
      "step": 2851
    },
    {
      "epoch": 1.8887417218543048,
      "grad_norm": 0.816174957864966,
      "learning_rate": 0.00028934821176053417,
      "loss": 2.5,
      "step": 2852
    },
    {
      "epoch": 1.8894039735099337,
      "grad_norm": 0.8149326571798456,
      "learning_rate": 0.00028932349399031677,
      "loss": 2.4219,
      "step": 2853
    },
    {
      "epoch": 1.8900662251655629,
      "grad_norm": 0.8505409004253704,
      "learning_rate": 0.0002892987486324503,
      "loss": 2.3281,
      "step": 2854
    },
    {
      "epoch": 1.890728476821192,
      "grad_norm": 0.8331251102621244,
      "learning_rate": 0.00028927397569183456,
      "loss": 2.3438,
      "step": 2855
    },
    {
      "epoch": 1.8913907284768212,
      "grad_norm": 0.811574080937041,
      "learning_rate": 0.00028924917517337493,
      "loss": 2.0156,
      "step": 2856
    },
    {
      "epoch": 1.8920529801324504,
      "grad_norm": 0.8767633493848389,
      "learning_rate": 0.00028922434708198217,
      "loss": 2.4844,
      "step": 2857
    },
    {
      "epoch": 1.8927152317880793,
      "grad_norm": 1.0272918206818542,
      "learning_rate": 0.0002891994914225725,
      "loss": 2.4688,
      "step": 2858
    },
    {
      "epoch": 1.8933774834437087,
      "grad_norm": 0.8361052106068869,
      "learning_rate": 0.0002891746082000678,
      "loss": 2.3281,
      "step": 2859
    },
    {
      "epoch": 1.8940397350993377,
      "grad_norm": 0.8234535062647853,
      "learning_rate": 0.0002891496974193949,
      "loss": 2.2344,
      "step": 2860
    },
    {
      "epoch": 1.894701986754967,
      "grad_norm": 0.8175272727413886,
      "learning_rate": 0.0002891247590854867,
      "loss": 2.0938,
      "step": 2861
    },
    {
      "epoch": 1.895364238410596,
      "grad_norm": 0.8563851871664755,
      "learning_rate": 0.00028909979320328123,
      "loss": 2.3594,
      "step": 2862
    },
    {
      "epoch": 1.8960264900662251,
      "grad_norm": 1.0408293894653124,
      "learning_rate": 0.0002890747997777219,
      "loss": 2.5156,
      "step": 2863
    },
    {
      "epoch": 1.8966887417218543,
      "grad_norm": 0.890811526080028,
      "learning_rate": 0.0002890497788137579,
      "loss": 2.3125,
      "step": 2864
    },
    {
      "epoch": 1.8973509933774835,
      "grad_norm": 0.774309852144894,
      "learning_rate": 0.0002890247303163434,
      "loss": 2.3438,
      "step": 2865
    },
    {
      "epoch": 1.8980132450331126,
      "grad_norm": 0.7604358930556171,
      "learning_rate": 0.00028899965429043847,
      "loss": 1.8516,
      "step": 2866
    },
    {
      "epoch": 1.8986754966887416,
      "grad_norm": 0.764006807882833,
      "learning_rate": 0.0002889745507410084,
      "loss": 2.3281,
      "step": 2867
    },
    {
      "epoch": 1.899337748344371,
      "grad_norm": 0.7948723753423214,
      "learning_rate": 0.0002889494196730241,
      "loss": 2.375,
      "step": 2868
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8105348976309922,
      "learning_rate": 0.0002889242610914615,
      "loss": 2.2812,
      "step": 2869
    },
    {
      "epoch": 1.9006622516556293,
      "grad_norm": 0.8412456084894522,
      "learning_rate": 0.0002888990750013027,
      "loss": 2.4688,
      "step": 2870
    },
    {
      "epoch": 1.9013245033112582,
      "grad_norm": 0.8386192721830606,
      "learning_rate": 0.0002888738614075345,
      "loss": 2.3906,
      "step": 2871
    },
    {
      "epoch": 1.9019867549668874,
      "grad_norm": 0.8824798288314648,
      "learning_rate": 0.00028884862031514964,
      "loss": 2.3125,
      "step": 2872
    },
    {
      "epoch": 1.9026490066225166,
      "grad_norm": 0.9463314200858433,
      "learning_rate": 0.0002888233517291461,
      "loss": 2.25,
      "step": 2873
    },
    {
      "epoch": 1.9033112582781457,
      "grad_norm": 0.7923747005374712,
      "learning_rate": 0.00028879805565452737,
      "loss": 2.4219,
      "step": 2874
    },
    {
      "epoch": 1.903973509933775,
      "grad_norm": 0.8535817109477793,
      "learning_rate": 0.00028877273209630233,
      "loss": 2.6875,
      "step": 2875
    },
    {
      "epoch": 1.9046357615894038,
      "grad_norm": 0.7949300347093511,
      "learning_rate": 0.0002887473810594854,
      "loss": 2.4375,
      "step": 2876
    },
    {
      "epoch": 1.9052980132450332,
      "grad_norm": 0.9116518927897763,
      "learning_rate": 0.0002887220025490963,
      "loss": 2.4375,
      "step": 2877
    },
    {
      "epoch": 1.9059602649006622,
      "grad_norm": 0.7514631649350063,
      "learning_rate": 0.0002886965965701603,
      "loss": 2.375,
      "step": 2878
    },
    {
      "epoch": 1.9066225165562913,
      "grad_norm": 0.8610009397095162,
      "learning_rate": 0.0002886711631277081,
      "loss": 2.4688,
      "step": 2879
    },
    {
      "epoch": 1.9072847682119205,
      "grad_norm": 0.7808018642883163,
      "learning_rate": 0.0002886457022267758,
      "loss": 2.3438,
      "step": 2880
    },
    {
      "epoch": 1.9079470198675497,
      "grad_norm": 0.8880085559979206,
      "learning_rate": 0.0002886202138724049,
      "loss": 2.5938,
      "step": 2881
    },
    {
      "epoch": 1.9086092715231788,
      "grad_norm": 0.7366438862000244,
      "learning_rate": 0.00028859469806964244,
      "loss": 2.0156,
      "step": 2882
    },
    {
      "epoch": 1.909271523178808,
      "grad_norm": 0.8104614526582509,
      "learning_rate": 0.00028856915482354073,
      "loss": 2.2969,
      "step": 2883
    },
    {
      "epoch": 1.9099337748344372,
      "grad_norm": 0.7776768693878011,
      "learning_rate": 0.0002885435841391578,
      "loss": 2.4219,
      "step": 2884
    },
    {
      "epoch": 1.910596026490066,
      "grad_norm": 0.7814699169651934,
      "learning_rate": 0.0002885179860215569,
      "loss": 2.3281,
      "step": 2885
    },
    {
      "epoch": 1.9112582781456955,
      "grad_norm": 0.7608477787307424,
      "learning_rate": 0.0002884923604758067,
      "loss": 2.3281,
      "step": 2886
    },
    {
      "epoch": 1.9119205298013244,
      "grad_norm": 0.8600599481103779,
      "learning_rate": 0.00028846670750698136,
      "loss": 2.4531,
      "step": 2887
    },
    {
      "epoch": 1.9125827814569536,
      "grad_norm": 0.8704771604695623,
      "learning_rate": 0.00028844102712016046,
      "loss": 2.5,
      "step": 2888
    },
    {
      "epoch": 1.9132450331125828,
      "grad_norm": 0.9179862719575775,
      "learning_rate": 0.00028841531932042897,
      "loss": 2.4844,
      "step": 2889
    },
    {
      "epoch": 1.913907284768212,
      "grad_norm": 0.7515594568735481,
      "learning_rate": 0.00028838958411287743,
      "loss": 2.0312,
      "step": 2890
    },
    {
      "epoch": 1.914569536423841,
      "grad_norm": 0.8444074129764717,
      "learning_rate": 0.0002883638215026016,
      "loss": 2.3594,
      "step": 2891
    },
    {
      "epoch": 1.91523178807947,
      "grad_norm": 1.8578889189855137,
      "learning_rate": 0.00028833803149470286,
      "loss": 2.3906,
      "step": 2892
    },
    {
      "epoch": 1.9158940397350994,
      "grad_norm": 0.754080134399696,
      "learning_rate": 0.0002883122140942879,
      "loss": 2.375,
      "step": 2893
    },
    {
      "epoch": 1.9165562913907284,
      "grad_norm": 0.833267882887863,
      "learning_rate": 0.0002882863693064688,
      "loss": 2.4531,
      "step": 2894
    },
    {
      "epoch": 1.9172185430463577,
      "grad_norm": 0.809603732360834,
      "learning_rate": 0.0002882604971363632,
      "loss": 2.3594,
      "step": 2895
    },
    {
      "epoch": 1.9178807947019867,
      "grad_norm": 0.8152211413011785,
      "learning_rate": 0.0002882345975890941,
      "loss": 2.4062,
      "step": 2896
    },
    {
      "epoch": 1.9185430463576159,
      "grad_norm": 0.8859369407881263,
      "learning_rate": 0.0002882086706697898,
      "loss": 2.4219,
      "step": 2897
    },
    {
      "epoch": 1.919205298013245,
      "grad_norm": 0.8529278961926933,
      "learning_rate": 0.0002881827163835843,
      "loss": 2.5312,
      "step": 2898
    },
    {
      "epoch": 1.9198675496688742,
      "grad_norm": 0.72315158333853,
      "learning_rate": 0.00028815673473561666,
      "loss": 2.125,
      "step": 2899
    },
    {
      "epoch": 1.9205298013245033,
      "grad_norm": 0.7726036155339066,
      "learning_rate": 0.00028813072573103163,
      "loss": 2.4688,
      "step": 2900
    },
    {
      "epoch": 1.9211920529801323,
      "grad_norm": 0.773743587470732,
      "learning_rate": 0.0002881046893749793,
      "loss": 2.2812,
      "step": 2901
    },
    {
      "epoch": 1.9218543046357617,
      "grad_norm": 0.8666089123457624,
      "learning_rate": 0.0002880786256726152,
      "loss": 2.5156,
      "step": 2902
    },
    {
      "epoch": 1.9225165562913906,
      "grad_norm": 0.8189555334416532,
      "learning_rate": 0.00028805253462910017,
      "loss": 2.6719,
      "step": 2903
    },
    {
      "epoch": 1.92317880794702,
      "grad_norm": 0.7428493848005827,
      "learning_rate": 0.0002880264162496005,
      "loss": 2.0469,
      "step": 2904
    },
    {
      "epoch": 1.923841059602649,
      "grad_norm": 0.7791561407549328,
      "learning_rate": 0.0002880002705392881,
      "loss": 2.0312,
      "step": 2905
    },
    {
      "epoch": 1.9245033112582781,
      "grad_norm": 0.791653762097023,
      "learning_rate": 0.0002879740975033399,
      "loss": 2.625,
      "step": 2906
    },
    {
      "epoch": 1.9251655629139073,
      "grad_norm": 0.8372867842182472,
      "learning_rate": 0.00028794789714693865,
      "loss": 2.625,
      "step": 2907
    },
    {
      "epoch": 1.9258278145695364,
      "grad_norm": 0.7281281423461923,
      "learning_rate": 0.0002879216694752722,
      "loss": 2.2344,
      "step": 2908
    },
    {
      "epoch": 1.9264900662251656,
      "grad_norm": 0.8437208508313229,
      "learning_rate": 0.000287895414493534,
      "loss": 2.0625,
      "step": 2909
    },
    {
      "epoch": 1.9271523178807946,
      "grad_norm": 0.7947660352935073,
      "learning_rate": 0.0002878691322069228,
      "loss": 2.125,
      "step": 2910
    },
    {
      "epoch": 1.927814569536424,
      "grad_norm": 0.8518070873263343,
      "learning_rate": 0.00028784282262064284,
      "loss": 2.4688,
      "step": 2911
    },
    {
      "epoch": 1.9284768211920529,
      "grad_norm": 0.8296905768915149,
      "learning_rate": 0.00028781648573990363,
      "loss": 2.4531,
      "step": 2912
    },
    {
      "epoch": 1.9291390728476823,
      "grad_norm": 0.8523138674206684,
      "learning_rate": 0.00028779012156992026,
      "loss": 2.625,
      "step": 2913
    },
    {
      "epoch": 1.9298013245033112,
      "grad_norm": 0.8118583335329524,
      "learning_rate": 0.0002877637301159131,
      "loss": 2.3125,
      "step": 2914
    },
    {
      "epoch": 1.9304635761589404,
      "grad_norm": 0.795400929278512,
      "learning_rate": 0.000287737311383108,
      "loss": 1.9922,
      "step": 2915
    },
    {
      "epoch": 1.9311258278145695,
      "grad_norm": 0.7856147758158786,
      "learning_rate": 0.0002877108653767361,
      "loss": 2.375,
      "step": 2916
    },
    {
      "epoch": 1.9317880794701987,
      "grad_norm": 0.7584332076625354,
      "learning_rate": 0.000287684392102034,
      "loss": 2.0312,
      "step": 2917
    },
    {
      "epoch": 1.9324503311258279,
      "grad_norm": 0.7853546435614522,
      "learning_rate": 0.0002876578915642439,
      "loss": 2.2812,
      "step": 2918
    },
    {
      "epoch": 1.9331125827814568,
      "grad_norm": 0.8100688802467343,
      "learning_rate": 0.000287631363768613,
      "loss": 2.4688,
      "step": 2919
    },
    {
      "epoch": 1.9337748344370862,
      "grad_norm": 0.8100790259363666,
      "learning_rate": 0.00028760480872039415,
      "loss": 2.5,
      "step": 2920
    },
    {
      "epoch": 1.9344370860927151,
      "grad_norm": 0.8053350334196528,
      "learning_rate": 0.0002875782264248456,
      "loss": 1.9922,
      "step": 2921
    },
    {
      "epoch": 1.9350993377483445,
      "grad_norm": 0.7718928961820292,
      "learning_rate": 0.00028755161688723095,
      "loss": 2.3438,
      "step": 2922
    },
    {
      "epoch": 1.9357615894039735,
      "grad_norm": 0.8266627231019702,
      "learning_rate": 0.0002875249801128192,
      "loss": 2.4219,
      "step": 2923
    },
    {
      "epoch": 1.9364238410596026,
      "grad_norm": 0.7908267174812844,
      "learning_rate": 0.00028749831610688466,
      "loss": 2.5156,
      "step": 2924
    },
    {
      "epoch": 1.9370860927152318,
      "grad_norm": 0.7784876997881196,
      "learning_rate": 0.0002874716248747072,
      "loss": 2.4219,
      "step": 2925
    },
    {
      "epoch": 1.937748344370861,
      "grad_norm": 0.9127165762173451,
      "learning_rate": 0.00028744490642157195,
      "loss": 2.5781,
      "step": 2926
    },
    {
      "epoch": 1.9384105960264901,
      "grad_norm": 0.81527355940955,
      "learning_rate": 0.0002874181607527695,
      "loss": 2.4844,
      "step": 2927
    },
    {
      "epoch": 1.939072847682119,
      "grad_norm": 0.8122161127053907,
      "learning_rate": 0.0002873913878735957,
      "loss": 2.25,
      "step": 2928
    },
    {
      "epoch": 1.9397350993377485,
      "grad_norm": 0.8595757228408031,
      "learning_rate": 0.00028736458778935204,
      "loss": 2.4062,
      "step": 2929
    },
    {
      "epoch": 1.9403973509933774,
      "grad_norm": 0.8285629984323323,
      "learning_rate": 0.0002873377605053451,
      "loss": 2.6875,
      "step": 2930
    },
    {
      "epoch": 1.9410596026490068,
      "grad_norm": 0.8238410643402478,
      "learning_rate": 0.0002873109060268871,
      "loss": 2.3438,
      "step": 2931
    },
    {
      "epoch": 1.9417218543046357,
      "grad_norm": 0.7108613583027408,
      "learning_rate": 0.00028728402435929546,
      "loss": 1.9688,
      "step": 2932
    },
    {
      "epoch": 1.942384105960265,
      "grad_norm": 1.0030386758670962,
      "learning_rate": 0.00028725711550789314,
      "loss": 2.5,
      "step": 2933
    },
    {
      "epoch": 1.943046357615894,
      "grad_norm": 1.0054361190941763,
      "learning_rate": 0.00028723017947800833,
      "loss": 2.7812,
      "step": 2934
    },
    {
      "epoch": 1.9437086092715232,
      "grad_norm": 0.8081275454921016,
      "learning_rate": 0.00028720321627497464,
      "loss": 2.3594,
      "step": 2935
    },
    {
      "epoch": 1.9443708609271524,
      "grad_norm": 0.7314984359958204,
      "learning_rate": 0.00028717622590413117,
      "loss": 2.2656,
      "step": 2936
    },
    {
      "epoch": 1.9450331125827813,
      "grad_norm": 0.7893275317431592,
      "learning_rate": 0.00028714920837082233,
      "loss": 2.4219,
      "step": 2937
    },
    {
      "epoch": 1.9456953642384107,
      "grad_norm": 0.9377042950172152,
      "learning_rate": 0.00028712216368039786,
      "loss": 2.5312,
      "step": 2938
    },
    {
      "epoch": 1.9463576158940397,
      "grad_norm": 0.829180220396996,
      "learning_rate": 0.0002870950918382129,
      "loss": 2.3594,
      "step": 2939
    },
    {
      "epoch": 1.9470198675496688,
      "grad_norm": 0.8748911234973903,
      "learning_rate": 0.000287067992849628,
      "loss": 2.25,
      "step": 2940
    },
    {
      "epoch": 1.947682119205298,
      "grad_norm": 0.8143908515718336,
      "learning_rate": 0.0002870408667200091,
      "loss": 1.9688,
      "step": 2941
    },
    {
      "epoch": 1.9483443708609272,
      "grad_norm": 0.7179007167088912,
      "learning_rate": 0.00028701371345472745,
      "loss": 2.1406,
      "step": 2942
    },
    {
      "epoch": 1.9490066225165563,
      "grad_norm": 0.9522570125357471,
      "learning_rate": 0.0002869865330591597,
      "loss": 2.7188,
      "step": 2943
    },
    {
      "epoch": 1.9496688741721855,
      "grad_norm": 0.7736543496263136,
      "learning_rate": 0.000286959325538688,
      "loss": 2.25,
      "step": 2944
    },
    {
      "epoch": 1.9503311258278146,
      "grad_norm": 0.9017343418215134,
      "learning_rate": 0.0002869320908986996,
      "loss": 2.4062,
      "step": 2945
    },
    {
      "epoch": 1.9509933774834436,
      "grad_norm": 0.9145655620579923,
      "learning_rate": 0.00028690482914458734,
      "loss": 2.4844,
      "step": 2946
    },
    {
      "epoch": 1.951655629139073,
      "grad_norm": 0.7629984312078882,
      "learning_rate": 0.00028687754028174935,
      "loss": 2.0312,
      "step": 2947
    },
    {
      "epoch": 1.952317880794702,
      "grad_norm": 0.8170916030967584,
      "learning_rate": 0.0002868502243155891,
      "loss": 2.375,
      "step": 2948
    },
    {
      "epoch": 1.952980132450331,
      "grad_norm": 0.7616419278151214,
      "learning_rate": 0.00028682288125151555,
      "loss": 1.9062,
      "step": 2949
    },
    {
      "epoch": 1.9536423841059603,
      "grad_norm": 0.8016584807303868,
      "learning_rate": 0.00028679551109494285,
      "loss": 2.4219,
      "step": 2950
    },
    {
      "epoch": 1.9543046357615894,
      "grad_norm": 0.7761216387124626,
      "learning_rate": 0.0002867681138512906,
      "loss": 2.3125,
      "step": 2951
    },
    {
      "epoch": 1.9549668874172186,
      "grad_norm": 0.7599671820765932,
      "learning_rate": 0.0002867406895259839,
      "loss": 2.3906,
      "step": 2952
    },
    {
      "epoch": 1.9556291390728475,
      "grad_norm": 0.8752925787182948,
      "learning_rate": 0.00028671323812445295,
      "loss": 2.3281,
      "step": 2953
    },
    {
      "epoch": 1.956291390728477,
      "grad_norm": 1.011514721219095,
      "learning_rate": 0.0002866857596521335,
      "loss": 2.3906,
      "step": 2954
    },
    {
      "epoch": 1.9569536423841059,
      "grad_norm": 0.8570594366666575,
      "learning_rate": 0.0002866582541144666,
      "loss": 2.2031,
      "step": 2955
    },
    {
      "epoch": 1.9576158940397352,
      "grad_norm": 0.8834455888493055,
      "learning_rate": 0.0002866307215168986,
      "loss": 2.5,
      "step": 2956
    },
    {
      "epoch": 1.9582781456953642,
      "grad_norm": 0.8042688311374435,
      "learning_rate": 0.0002866031618648814,
      "loss": 2.4844,
      "step": 2957
    },
    {
      "epoch": 1.9589403973509933,
      "grad_norm": 0.7930786158157322,
      "learning_rate": 0.00028657557516387195,
      "loss": 2.4062,
      "step": 2958
    },
    {
      "epoch": 1.9596026490066225,
      "grad_norm": 0.9466550603963453,
      "learning_rate": 0.00028654796141933293,
      "loss": 2.3438,
      "step": 2959
    },
    {
      "epoch": 1.9602649006622517,
      "grad_norm": 0.8094797870077459,
      "learning_rate": 0.00028652032063673197,
      "loss": 2.4531,
      "step": 2960
    },
    {
      "epoch": 1.9609271523178808,
      "grad_norm": 0.970401809793331,
      "learning_rate": 0.00028649265282154244,
      "loss": 2.4688,
      "step": 2961
    },
    {
      "epoch": 1.9615894039735098,
      "grad_norm": 0.6907469776593262,
      "learning_rate": 0.0002864649579792428,
      "loss": 2.0,
      "step": 2962
    },
    {
      "epoch": 1.9622516556291392,
      "grad_norm": 0.7883117431588597,
      "learning_rate": 0.0002864372361153169,
      "loss": 2.4688,
      "step": 2963
    },
    {
      "epoch": 1.9629139072847681,
      "grad_norm": 0.8262485500633999,
      "learning_rate": 0.00028640948723525415,
      "loss": 2.4688,
      "step": 2964
    },
    {
      "epoch": 1.9635761589403975,
      "grad_norm": 0.7804466498829276,
      "learning_rate": 0.000286381711344549,
      "loss": 2.2812,
      "step": 2965
    },
    {
      "epoch": 1.9642384105960264,
      "grad_norm": 0.8306677166668551,
      "learning_rate": 0.00028635390844870145,
      "loss": 2.4531,
      "step": 2966
    },
    {
      "epoch": 1.9649006622516556,
      "grad_norm": 0.7664082317708758,
      "learning_rate": 0.00028632607855321677,
      "loss": 2.3125,
      "step": 2967
    },
    {
      "epoch": 1.9655629139072848,
      "grad_norm": 0.7605676466756678,
      "learning_rate": 0.0002862982216636056,
      "loss": 2.3281,
      "step": 2968
    },
    {
      "epoch": 1.966225165562914,
      "grad_norm": 0.8373908271093768,
      "learning_rate": 0.00028627033778538396,
      "loss": 2.5781,
      "step": 2969
    },
    {
      "epoch": 1.966887417218543,
      "grad_norm": 0.8913532349662862,
      "learning_rate": 0.0002862424269240731,
      "loss": 2.4062,
      "step": 2970
    },
    {
      "epoch": 1.967549668874172,
      "grad_norm": 0.8062896553220953,
      "learning_rate": 0.00028621448908519977,
      "loss": 2.5156,
      "step": 2971
    },
    {
      "epoch": 1.9682119205298014,
      "grad_norm": 0.8537440183720169,
      "learning_rate": 0.000286186524274296,
      "loss": 2.2969,
      "step": 2972
    },
    {
      "epoch": 1.9688741721854304,
      "grad_norm": 0.6659432156567302,
      "learning_rate": 0.000286158532496899,
      "loss": 1.7422,
      "step": 2973
    },
    {
      "epoch": 1.9695364238410598,
      "grad_norm": 0.7637334146342779,
      "learning_rate": 0.00028613051375855165,
      "loss": 1.9844,
      "step": 2974
    },
    {
      "epoch": 1.9701986754966887,
      "grad_norm": 0.8661636695192214,
      "learning_rate": 0.00028610246806480186,
      "loss": 2.5469,
      "step": 2975
    },
    {
      "epoch": 1.9708609271523179,
      "grad_norm": 0.9010976718916557,
      "learning_rate": 0.000286074395421203,
      "loss": 2.5938,
      "step": 2976
    },
    {
      "epoch": 1.971523178807947,
      "grad_norm": 0.8404657917573464,
      "learning_rate": 0.0002860462958333138,
      "loss": 2.3906,
      "step": 2977
    },
    {
      "epoch": 1.9721854304635762,
      "grad_norm": 0.8256838659364901,
      "learning_rate": 0.00028601816930669835,
      "loss": 2.7031,
      "step": 2978
    },
    {
      "epoch": 1.9728476821192054,
      "grad_norm": 0.9506167543755507,
      "learning_rate": 0.00028599001584692595,
      "loss": 2.25,
      "step": 2979
    },
    {
      "epoch": 1.9735099337748343,
      "grad_norm": 0.7787009350264543,
      "learning_rate": 0.00028596183545957137,
      "loss": 1.875,
      "step": 2980
    },
    {
      "epoch": 1.9741721854304637,
      "grad_norm": 0.8541601514385739,
      "learning_rate": 0.0002859336281502146,
      "loss": 2.5938,
      "step": 2981
    },
    {
      "epoch": 1.9748344370860926,
      "grad_norm": 0.8227465942645088,
      "learning_rate": 0.000285905393924441,
      "loss": 2.4219,
      "step": 2982
    },
    {
      "epoch": 1.975496688741722,
      "grad_norm": 0.7358629404976218,
      "learning_rate": 0.0002858771327878413,
      "loss": 2.0312,
      "step": 2983
    },
    {
      "epoch": 1.976158940397351,
      "grad_norm": 0.8552340727553472,
      "learning_rate": 0.00028584884474601154,
      "loss": 2.5625,
      "step": 2984
    },
    {
      "epoch": 1.9768211920529801,
      "grad_norm": 0.7856535421710622,
      "learning_rate": 0.0002858205298045531,
      "loss": 2.3281,
      "step": 2985
    },
    {
      "epoch": 1.9774834437086093,
      "grad_norm": 0.7367140394663255,
      "learning_rate": 0.0002857921879690726,
      "loss": 2.4375,
      "step": 2986
    },
    {
      "epoch": 1.9781456953642385,
      "grad_norm": 0.8007103401880254,
      "learning_rate": 0.0002857638192451821,
      "loss": 2.2812,
      "step": 2987
    },
    {
      "epoch": 1.9788079470198676,
      "grad_norm": 0.8067678255289292,
      "learning_rate": 0.00028573542363849886,
      "loss": 2.4219,
      "step": 2988
    },
    {
      "epoch": 1.9794701986754966,
      "grad_norm": 0.7206682472996087,
      "learning_rate": 0.00028570700115464556,
      "loss": 2.0,
      "step": 2989
    },
    {
      "epoch": 1.980132450331126,
      "grad_norm": 0.79344168752213,
      "learning_rate": 0.0002856785517992503,
      "loss": 2.2969,
      "step": 2990
    },
    {
      "epoch": 1.980794701986755,
      "grad_norm": 0.8411552127274997,
      "learning_rate": 0.00028565007557794617,
      "loss": 2.4844,
      "step": 2991
    },
    {
      "epoch": 1.981456953642384,
      "grad_norm": 0.8069420007135865,
      "learning_rate": 0.000285621572496372,
      "loss": 2.4219,
      "step": 2992
    },
    {
      "epoch": 1.9821192052980132,
      "grad_norm": 0.7578209686554758,
      "learning_rate": 0.0002855930425601716,
      "loss": 2.3125,
      "step": 2993
    },
    {
      "epoch": 1.9827814569536424,
      "grad_norm": 0.7980714684551018,
      "learning_rate": 0.0002855644857749942,
      "loss": 2.4062,
      "step": 2994
    },
    {
      "epoch": 1.9834437086092715,
      "grad_norm": 0.7957242808934297,
      "learning_rate": 0.00028553590214649445,
      "loss": 2.25,
      "step": 2995
    },
    {
      "epoch": 1.9841059602649007,
      "grad_norm": 0.7906121709016272,
      "learning_rate": 0.0002855072916803322,
      "loss": 2.0625,
      "step": 2996
    },
    {
      "epoch": 1.9847682119205299,
      "grad_norm": 0.7418735925128951,
      "learning_rate": 0.00028547865438217267,
      "loss": 2.3906,
      "step": 2997
    },
    {
      "epoch": 1.9854304635761588,
      "grad_norm": 0.7724521520361076,
      "learning_rate": 0.00028544999025768635,
      "loss": 2.4531,
      "step": 2998
    },
    {
      "epoch": 1.9860927152317882,
      "grad_norm": 0.7349983842965367,
      "learning_rate": 0.0002854212993125491,
      "loss": 2.1562,
      "step": 2999
    },
    {
      "epoch": 1.9867549668874172,
      "grad_norm": 0.7906857144921268,
      "learning_rate": 0.000285392581552442,
      "loss": 2.3594,
      "step": 3000
    },
    {
      "epoch": 1.9874172185430463,
      "grad_norm": 0.7967094493937565,
      "learning_rate": 0.00028536383698305156,
      "loss": 2.2656,
      "step": 3001
    },
    {
      "epoch": 1.9880794701986755,
      "grad_norm": 0.7345940240802142,
      "learning_rate": 0.00028533506561006947,
      "loss": 2.4062,
      "step": 3002
    },
    {
      "epoch": 1.9887417218543046,
      "grad_norm": 0.7925910814584763,
      "learning_rate": 0.00028530626743919284,
      "loss": 2.0156,
      "step": 3003
    },
    {
      "epoch": 1.9894039735099338,
      "grad_norm": 0.7329176566343101,
      "learning_rate": 0.000285277442476124,
      "loss": 2.3594,
      "step": 3004
    },
    {
      "epoch": 1.9900662251655628,
      "grad_norm": 0.7688851320110776,
      "learning_rate": 0.0002852485907265706,
      "loss": 2.2969,
      "step": 3005
    },
    {
      "epoch": 1.9907284768211921,
      "grad_norm": 0.691042239287597,
      "learning_rate": 0.0002852197121962458,
      "loss": 2.1406,
      "step": 3006
    },
    {
      "epoch": 1.991390728476821,
      "grad_norm": 0.704431539848973,
      "learning_rate": 0.00028519080689086765,
      "loss": 1.875,
      "step": 3007
    },
    {
      "epoch": 1.9920529801324505,
      "grad_norm": 0.8634188034160398,
      "learning_rate": 0.0002851618748161598,
      "loss": 2.4688,
      "step": 3008
    },
    {
      "epoch": 1.9927152317880794,
      "grad_norm": 0.8347909373295213,
      "learning_rate": 0.0002851329159778512,
      "loss": 2.4531,
      "step": 3009
    },
    {
      "epoch": 1.9933774834437086,
      "grad_norm": 0.8376122542428929,
      "learning_rate": 0.000285103930381676,
      "loss": 2.3438,
      "step": 3010
    },
    {
      "epoch": 1.9940397350993377,
      "grad_norm": 0.8201923354550036,
      "learning_rate": 0.0002850749180333736,
      "loss": 2.25,
      "step": 3011
    },
    {
      "epoch": 1.994701986754967,
      "grad_norm": 0.7512952569807326,
      "learning_rate": 0.00028504587893868897,
      "loss": 2.375,
      "step": 3012
    },
    {
      "epoch": 1.995364238410596,
      "grad_norm": 0.7677195904631053,
      "learning_rate": 0.000285016813103372,
      "loss": 2.2188,
      "step": 3013
    },
    {
      "epoch": 1.996026490066225,
      "grad_norm": 0.8307319096118427,
      "learning_rate": 0.00028498772053317805,
      "loss": 2.375,
      "step": 3014
    },
    {
      "epoch": 1.9966887417218544,
      "grad_norm": 0.8725571263053491,
      "learning_rate": 0.0002849586012338679,
      "loss": 2.25,
      "step": 3015
    },
    {
      "epoch": 1.9973509933774833,
      "grad_norm": 0.8151400709447173,
      "learning_rate": 0.0002849294552112075,
      "loss": 2.2969,
      "step": 3016
    },
    {
      "epoch": 1.9980132450331127,
      "grad_norm": 0.6970740944504712,
      "learning_rate": 0.000284900282470968,
      "loss": 2.1094,
      "step": 3017
    },
    {
      "epoch": 1.9986754966887417,
      "grad_norm": 0.767726541052731,
      "learning_rate": 0.000284871083018926,
      "loss": 2.3906,
      "step": 3018
    },
    {
      "epoch": 1.9993377483443708,
      "grad_norm": 0.856240836324722,
      "learning_rate": 0.00028484185686086333,
      "loss": 2.3125,
      "step": 3019
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8080218053225915,
      "learning_rate": 0.00028481260400256705,
      "loss": 2.2344,
      "step": 3020
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.3295350074768066,
      "eval_runtime": 34.0055,
      "eval_samples_per_second": 9.94,
      "eval_steps_per_second": 9.94,
      "step": 3020
    },
    {
      "epoch": 2.000662251655629,
      "grad_norm": 0.7822085716206245,
      "learning_rate": 0.0002847833244498296,
      "loss": 2.2812,
      "step": 3021
    },
    {
      "epoch": 2.0013245033112583,
      "grad_norm": 0.7351924332954816,
      "learning_rate": 0.0002847540182084487,
      "loss": 2.2969,
      "step": 3022
    },
    {
      "epoch": 2.0019867549668873,
      "grad_norm": 0.8244890366369247,
      "learning_rate": 0.0002847246852842273,
      "loss": 2.0156,
      "step": 3023
    },
    {
      "epoch": 2.0026490066225167,
      "grad_norm": 0.7119041181724008,
      "learning_rate": 0.0002846953256829736,
      "loss": 1.4062,
      "step": 3024
    },
    {
      "epoch": 2.0033112582781456,
      "grad_norm": 0.7871132037714198,
      "learning_rate": 0.00028466593941050124,
      "loss": 1.9453,
      "step": 3025
    },
    {
      "epoch": 2.003973509933775,
      "grad_norm": 0.7605720798387549,
      "learning_rate": 0.00028463652647262895,
      "loss": 2.1406,
      "step": 3026
    },
    {
      "epoch": 2.004635761589404,
      "grad_norm": 0.8961684730068566,
      "learning_rate": 0.0002846070868751809,
      "loss": 2.2188,
      "step": 3027
    },
    {
      "epoch": 2.0052980132450333,
      "grad_norm": 0.837197825175647,
      "learning_rate": 0.0002845776206239864,
      "loss": 2.2812,
      "step": 3028
    },
    {
      "epoch": 2.0059602649006623,
      "grad_norm": 0.8984064290365772,
      "learning_rate": 0.0002845481277248801,
      "loss": 2.4219,
      "step": 3029
    },
    {
      "epoch": 2.006622516556291,
      "grad_norm": 0.8987708137622051,
      "learning_rate": 0.000284518608183702,
      "loss": 2.0781,
      "step": 3030
    },
    {
      "epoch": 2.0072847682119206,
      "grad_norm": 0.7767318586194267,
      "learning_rate": 0.0002844890620062973,
      "loss": 2.0781,
      "step": 3031
    },
    {
      "epoch": 2.0079470198675495,
      "grad_norm": 0.8110983207289155,
      "learning_rate": 0.0002844594891985164,
      "loss": 2.3125,
      "step": 3032
    },
    {
      "epoch": 2.008609271523179,
      "grad_norm": 0.7690697359258195,
      "learning_rate": 0.0002844298897662152,
      "loss": 1.9844,
      "step": 3033
    },
    {
      "epoch": 2.009271523178808,
      "grad_norm": 0.8046151668047622,
      "learning_rate": 0.0002844002637152545,
      "loss": 2.2188,
      "step": 3034
    },
    {
      "epoch": 2.0099337748344372,
      "grad_norm": 0.8182453332284277,
      "learning_rate": 0.00028437061105150084,
      "loss": 2.3125,
      "step": 3035
    },
    {
      "epoch": 2.010596026490066,
      "grad_norm": 0.8150776848494918,
      "learning_rate": 0.0002843409317808257,
      "loss": 2.3594,
      "step": 3036
    },
    {
      "epoch": 2.0112582781456956,
      "grad_norm": 0.696765242686899,
      "learning_rate": 0.0002843112259091058,
      "loss": 1.7969,
      "step": 3037
    },
    {
      "epoch": 2.0119205298013245,
      "grad_norm": 0.7662577119962337,
      "learning_rate": 0.0002842814934422234,
      "loss": 2.1562,
      "step": 3038
    },
    {
      "epoch": 2.0125827814569535,
      "grad_norm": 0.8412781589294703,
      "learning_rate": 0.0002842517343860658,
      "loss": 1.9688,
      "step": 3039
    },
    {
      "epoch": 2.013245033112583,
      "grad_norm": 0.8110549835615121,
      "learning_rate": 0.00028422194874652565,
      "loss": 2.3594,
      "step": 3040
    },
    {
      "epoch": 2.013907284768212,
      "grad_norm": 0.7956529026298629,
      "learning_rate": 0.00028419213652950087,
      "loss": 1.6484,
      "step": 3041
    },
    {
      "epoch": 2.014569536423841,
      "grad_norm": 0.8333843196351454,
      "learning_rate": 0.00028416229774089457,
      "loss": 1.8906,
      "step": 3042
    },
    {
      "epoch": 2.01523178807947,
      "grad_norm": 0.9274956110875618,
      "learning_rate": 0.00028413243238661516,
      "loss": 2.2344,
      "step": 3043
    },
    {
      "epoch": 2.0158940397350995,
      "grad_norm": 0.9788051570809375,
      "learning_rate": 0.0002841025404725764,
      "loss": 2.4531,
      "step": 3044
    },
    {
      "epoch": 2.0165562913907285,
      "grad_norm": 0.8414831485270553,
      "learning_rate": 0.0002840726220046972,
      "loss": 2.2344,
      "step": 3045
    },
    {
      "epoch": 2.017218543046358,
      "grad_norm": 0.8711942496434394,
      "learning_rate": 0.00028404267698890176,
      "loss": 2.1562,
      "step": 3046
    },
    {
      "epoch": 2.017880794701987,
      "grad_norm": 0.8454803230666261,
      "learning_rate": 0.00028401270543111954,
      "loss": 2.3906,
      "step": 3047
    },
    {
      "epoch": 2.0185430463576157,
      "grad_norm": 0.7521761194131156,
      "learning_rate": 0.00028398270733728516,
      "loss": 1.8984,
      "step": 3048
    },
    {
      "epoch": 2.019205298013245,
      "grad_norm": 0.8538737442365523,
      "learning_rate": 0.00028395268271333874,
      "loss": 2.3906,
      "step": 3049
    },
    {
      "epoch": 2.019867549668874,
      "grad_norm": 0.7523135474096135,
      "learning_rate": 0.0002839226315652254,
      "loss": 2.0938,
      "step": 3050
    },
    {
      "epoch": 2.0205298013245034,
      "grad_norm": 0.822975951350184,
      "learning_rate": 0.00028389255389889566,
      "loss": 2.125,
      "step": 3051
    },
    {
      "epoch": 2.0211920529801324,
      "grad_norm": 0.7688298353858256,
      "learning_rate": 0.00028386244972030517,
      "loss": 2.2188,
      "step": 3052
    },
    {
      "epoch": 2.0218543046357618,
      "grad_norm": 0.8568860086774109,
      "learning_rate": 0.000283832319035415,
      "loss": 1.9922,
      "step": 3053
    },
    {
      "epoch": 2.0225165562913907,
      "grad_norm": 0.8606099574985561,
      "learning_rate": 0.0002838021618501913,
      "loss": 2.0938,
      "step": 3054
    },
    {
      "epoch": 2.0231788079470197,
      "grad_norm": 0.9079625132824076,
      "learning_rate": 0.00028377197817060554,
      "loss": 2.4062,
      "step": 3055
    },
    {
      "epoch": 2.023841059602649,
      "grad_norm": 0.8956579168343832,
      "learning_rate": 0.0002837417680026345,
      "loss": 2.2344,
      "step": 3056
    },
    {
      "epoch": 2.024503311258278,
      "grad_norm": 0.796979935853134,
      "learning_rate": 0.00028371153135226,
      "loss": 1.9453,
      "step": 3057
    },
    {
      "epoch": 2.0251655629139074,
      "grad_norm": 0.7634717956919849,
      "learning_rate": 0.00028368126822546943,
      "loss": 2.1719,
      "step": 3058
    },
    {
      "epoch": 2.0258278145695363,
      "grad_norm": 1.2910630383165485,
      "learning_rate": 0.00028365097862825513,
      "loss": 2.2188,
      "step": 3059
    },
    {
      "epoch": 2.0264900662251657,
      "grad_norm": 0.9142860946154696,
      "learning_rate": 0.0002836206625666148,
      "loss": 2.4688,
      "step": 3060
    },
    {
      "epoch": 2.0271523178807946,
      "grad_norm": 0.7395723062096163,
      "learning_rate": 0.0002835903200465514,
      "loss": 1.8203,
      "step": 3061
    },
    {
      "epoch": 2.027814569536424,
      "grad_norm": 0.9038546366642554,
      "learning_rate": 0.000283559951074073,
      "loss": 2.5938,
      "step": 3062
    },
    {
      "epoch": 2.028476821192053,
      "grad_norm": 0.7863480424396265,
      "learning_rate": 0.00028352955565519316,
      "loss": 2.2188,
      "step": 3063
    },
    {
      "epoch": 2.029139072847682,
      "grad_norm": 0.8261949428134419,
      "learning_rate": 0.0002834991337959304,
      "loss": 2.2969,
      "step": 3064
    },
    {
      "epoch": 2.0298013245033113,
      "grad_norm": 0.7972801406164743,
      "learning_rate": 0.0002834686855023086,
      "loss": 1.9531,
      "step": 3065
    },
    {
      "epoch": 2.0304635761589402,
      "grad_norm": 0.7231815437325241,
      "learning_rate": 0.00028343821078035693,
      "loss": 1.75,
      "step": 3066
    },
    {
      "epoch": 2.0311258278145696,
      "grad_norm": 0.7521245507311504,
      "learning_rate": 0.0002834077096361097,
      "loss": 2.1406,
      "step": 3067
    },
    {
      "epoch": 2.0317880794701986,
      "grad_norm": 1.0264235434684348,
      "learning_rate": 0.00028337718207560646,
      "loss": 2.7969,
      "step": 3068
    },
    {
      "epoch": 2.032450331125828,
      "grad_norm": 0.716934773790205,
      "learning_rate": 0.00028334662810489207,
      "loss": 1.6797,
      "step": 3069
    },
    {
      "epoch": 2.033112582781457,
      "grad_norm": 0.8696441641148054,
      "learning_rate": 0.0002833160477300166,
      "loss": 2.375,
      "step": 3070
    },
    {
      "epoch": 2.0337748344370863,
      "grad_norm": 0.9122303223825191,
      "learning_rate": 0.0002832854409570352,
      "loss": 2.3125,
      "step": 3071
    },
    {
      "epoch": 2.0344370860927152,
      "grad_norm": 0.7640241438415241,
      "learning_rate": 0.00028325480779200845,
      "loss": 2.2344,
      "step": 3072
    },
    {
      "epoch": 2.035099337748344,
      "grad_norm": 0.9634486625711441,
      "learning_rate": 0.000283224148241002,
      "loss": 2.2344,
      "step": 3073
    },
    {
      "epoch": 2.0357615894039736,
      "grad_norm": 0.8418488970874002,
      "learning_rate": 0.00028319346231008686,
      "loss": 2.2656,
      "step": 3074
    },
    {
      "epoch": 2.0364238410596025,
      "grad_norm": 0.8826407865837091,
      "learning_rate": 0.0002831627500053392,
      "loss": 2.4062,
      "step": 3075
    },
    {
      "epoch": 2.037086092715232,
      "grad_norm": 0.7944930509678328,
      "learning_rate": 0.00028313201133284033,
      "loss": 1.7969,
      "step": 3076
    },
    {
      "epoch": 2.037748344370861,
      "grad_norm": 0.8108011502441784,
      "learning_rate": 0.00028310124629867694,
      "loss": 1.8984,
      "step": 3077
    },
    {
      "epoch": 2.03841059602649,
      "grad_norm": 0.7722033275737668,
      "learning_rate": 0.0002830704549089408,
      "loss": 1.8203,
      "step": 3078
    },
    {
      "epoch": 2.039072847682119,
      "grad_norm": 0.8060447789564797,
      "learning_rate": 0.000283039637169729,
      "loss": 2.0781,
      "step": 3079
    },
    {
      "epoch": 2.0397350993377485,
      "grad_norm": 0.8437134224555156,
      "learning_rate": 0.00028300879308714377,
      "loss": 1.9688,
      "step": 3080
    },
    {
      "epoch": 2.0403973509933775,
      "grad_norm": 0.9196681504395526,
      "learning_rate": 0.0002829779226672927,
      "loss": 2.2812,
      "step": 3081
    },
    {
      "epoch": 2.0410596026490064,
      "grad_norm": 3.101775531759244,
      "learning_rate": 0.0002829470259162883,
      "loss": 2.3125,
      "step": 3082
    },
    {
      "epoch": 2.041721854304636,
      "grad_norm": 0.8790655244893846,
      "learning_rate": 0.0002829161028402486,
      "loss": 2.4219,
      "step": 3083
    },
    {
      "epoch": 2.0423841059602648,
      "grad_norm": 0.823235163431939,
      "learning_rate": 0.00028288515344529675,
      "loss": 2.0781,
      "step": 3084
    },
    {
      "epoch": 2.043046357615894,
      "grad_norm": 0.7846403539578026,
      "learning_rate": 0.00028285417773756097,
      "loss": 2.0312,
      "step": 3085
    },
    {
      "epoch": 2.043708609271523,
      "grad_norm": 0.8710351265030196,
      "learning_rate": 0.0002828231757231749,
      "loss": 2.0,
      "step": 3086
    },
    {
      "epoch": 2.0443708609271525,
      "grad_norm": 0.9168808979990573,
      "learning_rate": 0.0002827921474082773,
      "loss": 2.0,
      "step": 3087
    },
    {
      "epoch": 2.0450331125827814,
      "grad_norm": 0.759742998506687,
      "learning_rate": 0.0002827610927990121,
      "loss": 1.6016,
      "step": 3088
    },
    {
      "epoch": 2.045695364238411,
      "grad_norm": 0.9232128403809053,
      "learning_rate": 0.00028273001190152845,
      "loss": 2.3125,
      "step": 3089
    },
    {
      "epoch": 2.0463576158940397,
      "grad_norm": 0.8102866278962713,
      "learning_rate": 0.0002826989047219807,
      "loss": 2.0,
      "step": 3090
    },
    {
      "epoch": 2.0470198675496687,
      "grad_norm": 0.8976244384168187,
      "learning_rate": 0.0002826677712665285,
      "loss": 2.2656,
      "step": 3091
    },
    {
      "epoch": 2.047682119205298,
      "grad_norm": 0.8184290883734581,
      "learning_rate": 0.0002826366115413366,
      "loss": 2.0469,
      "step": 3092
    },
    {
      "epoch": 2.048344370860927,
      "grad_norm": 0.7909456792494507,
      "learning_rate": 0.000282605425552575,
      "loss": 1.6797,
      "step": 3093
    },
    {
      "epoch": 2.0490066225165564,
      "grad_norm": 0.8799167734306641,
      "learning_rate": 0.0002825742133064189,
      "loss": 2.0625,
      "step": 3094
    },
    {
      "epoch": 2.0496688741721854,
      "grad_norm": 0.9324396640392418,
      "learning_rate": 0.0002825429748090487,
      "loss": 2.3125,
      "step": 3095
    },
    {
      "epoch": 2.0503311258278147,
      "grad_norm": 0.8530898999727972,
      "learning_rate": 0.0002825117100666498,
      "loss": 2.2188,
      "step": 3096
    },
    {
      "epoch": 2.0509933774834437,
      "grad_norm": 0.8245024540803908,
      "learning_rate": 0.0002824804190854132,
      "loss": 2.4531,
      "step": 3097
    },
    {
      "epoch": 2.0516556291390726,
      "grad_norm": 0.8573663069793667,
      "learning_rate": 0.0002824491018715347,
      "loss": 2.3438,
      "step": 3098
    },
    {
      "epoch": 2.052317880794702,
      "grad_norm": 0.9130798646665145,
      "learning_rate": 0.00028241775843121565,
      "loss": 2.0312,
      "step": 3099
    },
    {
      "epoch": 2.052980132450331,
      "grad_norm": 0.7987379246026312,
      "learning_rate": 0.00028238638877066223,
      "loss": 2.1406,
      "step": 3100
    },
    {
      "epoch": 2.0536423841059603,
      "grad_norm": 0.8467016153797152,
      "learning_rate": 0.0002823549928960861,
      "loss": 2.2344,
      "step": 3101
    },
    {
      "epoch": 2.0543046357615893,
      "grad_norm": 0.7651170264975742,
      "learning_rate": 0.000282323570813704,
      "loss": 2.1719,
      "step": 3102
    },
    {
      "epoch": 2.0549668874172187,
      "grad_norm": 0.7974496084290763,
      "learning_rate": 0.0002822921225297378,
      "loss": 2.2812,
      "step": 3103
    },
    {
      "epoch": 2.0556291390728476,
      "grad_norm": 0.8419218283380449,
      "learning_rate": 0.00028226064805041465,
      "loss": 2.2188,
      "step": 3104
    },
    {
      "epoch": 2.056291390728477,
      "grad_norm": 0.8183821891990928,
      "learning_rate": 0.0002822291473819669,
      "loss": 2.4688,
      "step": 3105
    },
    {
      "epoch": 2.056953642384106,
      "grad_norm": 0.7669416981098064,
      "learning_rate": 0.000282197620530632,
      "loss": 2.125,
      "step": 3106
    },
    {
      "epoch": 2.057615894039735,
      "grad_norm": 0.7933034425220477,
      "learning_rate": 0.00028216606750265265,
      "loss": 2.1719,
      "step": 3107
    },
    {
      "epoch": 2.0582781456953643,
      "grad_norm": 0.8410366188945075,
      "learning_rate": 0.0002821344883042767,
      "loss": 2.3125,
      "step": 3108
    },
    {
      "epoch": 2.058940397350993,
      "grad_norm": 0.7943328375176141,
      "learning_rate": 0.0002821028829417572,
      "loss": 2.3281,
      "step": 3109
    },
    {
      "epoch": 2.0596026490066226,
      "grad_norm": 0.8197908979710908,
      "learning_rate": 0.0002820712514213524,
      "loss": 2.2344,
      "step": 3110
    },
    {
      "epoch": 2.0602649006622515,
      "grad_norm": 0.7799788168987745,
      "learning_rate": 0.00028203959374932557,
      "loss": 1.9062,
      "step": 3111
    },
    {
      "epoch": 2.060927152317881,
      "grad_norm": 1.6397234326451111,
      "learning_rate": 0.00028200790993194546,
      "loss": 1.6406,
      "step": 3112
    },
    {
      "epoch": 2.06158940397351,
      "grad_norm": 0.9426110548715044,
      "learning_rate": 0.00028197619997548574,
      "loss": 2.2344,
      "step": 3113
    },
    {
      "epoch": 2.0622516556291393,
      "grad_norm": 0.8363704691023665,
      "learning_rate": 0.00028194446388622534,
      "loss": 2.0625,
      "step": 3114
    },
    {
      "epoch": 2.062913907284768,
      "grad_norm": 0.800036050381332,
      "learning_rate": 0.00028191270167044846,
      "loss": 2.1719,
      "step": 3115
    },
    {
      "epoch": 2.063576158940397,
      "grad_norm": 0.8868937666014706,
      "learning_rate": 0.0002818809133344443,
      "loss": 2.5625,
      "step": 3116
    },
    {
      "epoch": 2.0642384105960265,
      "grad_norm": 0.8638151015504766,
      "learning_rate": 0.00028184909888450733,
      "loss": 2.2656,
      "step": 3117
    },
    {
      "epoch": 2.0649006622516555,
      "grad_norm": 0.8893267453546309,
      "learning_rate": 0.0002818172583269372,
      "loss": 2.25,
      "step": 3118
    },
    {
      "epoch": 2.065562913907285,
      "grad_norm": 0.8067311127583038,
      "learning_rate": 0.0002817853916680386,
      "loss": 2.2812,
      "step": 3119
    },
    {
      "epoch": 2.066225165562914,
      "grad_norm": 0.7622862111987854,
      "learning_rate": 0.0002817534989141217,
      "loss": 2.1719,
      "step": 3120
    },
    {
      "epoch": 2.066887417218543,
      "grad_norm": 0.7926266710421536,
      "learning_rate": 0.00028172158007150143,
      "loss": 2.2656,
      "step": 3121
    },
    {
      "epoch": 2.067549668874172,
      "grad_norm": 0.755907987843484,
      "learning_rate": 0.0002816896351464982,
      "loss": 1.8828,
      "step": 3122
    },
    {
      "epoch": 2.0682119205298015,
      "grad_norm": 0.8265968534322612,
      "learning_rate": 0.0002816576641454374,
      "loss": 2.2188,
      "step": 3123
    },
    {
      "epoch": 2.0688741721854305,
      "grad_norm": 0.8199098589470273,
      "learning_rate": 0.0002816256670746498,
      "loss": 2.0781,
      "step": 3124
    },
    {
      "epoch": 2.0695364238410594,
      "grad_norm": 0.8530459411639963,
      "learning_rate": 0.00028159364394047095,
      "loss": 2.4375,
      "step": 3125
    },
    {
      "epoch": 2.070198675496689,
      "grad_norm": 0.7856947973609075,
      "learning_rate": 0.000281561594749242,
      "loss": 1.8984,
      "step": 3126
    },
    {
      "epoch": 2.0708609271523177,
      "grad_norm": 0.8166310147552497,
      "learning_rate": 0.00028152951950730895,
      "loss": 2.2031,
      "step": 3127
    },
    {
      "epoch": 2.071523178807947,
      "grad_norm": 0.8320901590465946,
      "learning_rate": 0.0002814974182210231,
      "loss": 2.0781,
      "step": 3128
    },
    {
      "epoch": 2.072185430463576,
      "grad_norm": 0.8519037425416551,
      "learning_rate": 0.0002814652908967409,
      "loss": 2.125,
      "step": 3129
    },
    {
      "epoch": 2.0728476821192054,
      "grad_norm": 0.7537013947035321,
      "learning_rate": 0.00028143313754082393,
      "loss": 1.8203,
      "step": 3130
    },
    {
      "epoch": 2.0735099337748344,
      "grad_norm": 0.8793310990728718,
      "learning_rate": 0.0002814009581596388,
      "loss": 2.1406,
      "step": 3131
    },
    {
      "epoch": 2.0741721854304638,
      "grad_norm": 0.811602230226561,
      "learning_rate": 0.0002813687527595576,
      "loss": 1.9844,
      "step": 3132
    },
    {
      "epoch": 2.0748344370860927,
      "grad_norm": 0.8777386683507972,
      "learning_rate": 0.0002813365213469572,
      "loss": 1.75,
      "step": 3133
    },
    {
      "epoch": 2.0754966887417217,
      "grad_norm": 0.875854239870614,
      "learning_rate": 0.00028130426392821984,
      "loss": 1.8359,
      "step": 3134
    },
    {
      "epoch": 2.076158940397351,
      "grad_norm": 0.8872111356958647,
      "learning_rate": 0.0002812719805097329,
      "loss": 1.9062,
      "step": 3135
    },
    {
      "epoch": 2.07682119205298,
      "grad_norm": 0.855470303928801,
      "learning_rate": 0.0002812396710978888,
      "loss": 1.9531,
      "step": 3136
    },
    {
      "epoch": 2.0774834437086094,
      "grad_norm": 1.0019710975758873,
      "learning_rate": 0.00028120733569908524,
      "loss": 2.375,
      "step": 3137
    },
    {
      "epoch": 2.0781456953642383,
      "grad_norm": 0.8133040534781214,
      "learning_rate": 0.0002811749743197249,
      "loss": 2.1094,
      "step": 3138
    },
    {
      "epoch": 2.0788079470198677,
      "grad_norm": 0.9238881120290313,
      "learning_rate": 0.0002811425869662159,
      "loss": 2.3594,
      "step": 3139
    },
    {
      "epoch": 2.0794701986754967,
      "grad_norm": 0.868338876743035,
      "learning_rate": 0.00028111017364497107,
      "loss": 2.0938,
      "step": 3140
    },
    {
      "epoch": 2.080132450331126,
      "grad_norm": 0.7834299580394366,
      "learning_rate": 0.0002810777343624088,
      "loss": 1.9531,
      "step": 3141
    },
    {
      "epoch": 2.080794701986755,
      "grad_norm": 0.8840221639877152,
      "learning_rate": 0.0002810452691249523,
      "loss": 2.25,
      "step": 3142
    },
    {
      "epoch": 2.081456953642384,
      "grad_norm": 0.825079397032191,
      "learning_rate": 0.0002810127779390302,
      "loss": 2.1719,
      "step": 3143
    },
    {
      "epoch": 2.0821192052980133,
      "grad_norm": 0.8970107011962454,
      "learning_rate": 0.00028098026081107603,
      "loss": 2.2188,
      "step": 3144
    },
    {
      "epoch": 2.0827814569536423,
      "grad_norm": 0.7661709812279206,
      "learning_rate": 0.00028094771774752854,
      "loss": 2.0625,
      "step": 3145
    },
    {
      "epoch": 2.0834437086092716,
      "grad_norm": 0.8724765531200942,
      "learning_rate": 0.0002809151487548317,
      "loss": 2.4375,
      "step": 3146
    },
    {
      "epoch": 2.0841059602649006,
      "grad_norm": 0.8027041321418783,
      "learning_rate": 0.00028088255383943453,
      "loss": 2.2656,
      "step": 3147
    },
    {
      "epoch": 2.08476821192053,
      "grad_norm": 0.7778881267098369,
      "learning_rate": 0.00028084993300779115,
      "loss": 2.2969,
      "step": 3148
    },
    {
      "epoch": 2.085430463576159,
      "grad_norm": 0.7473978371141099,
      "learning_rate": 0.0002808172862663609,
      "loss": 1.7734,
      "step": 3149
    },
    {
      "epoch": 2.0860927152317883,
      "grad_norm": 0.7427112478913946,
      "learning_rate": 0.00028078461362160824,
      "loss": 1.8516,
      "step": 3150
    },
    {
      "epoch": 2.0867549668874172,
      "grad_norm": 0.9033159801032535,
      "learning_rate": 0.0002807519150800026,
      "loss": 2.3125,
      "step": 3151
    },
    {
      "epoch": 2.087417218543046,
      "grad_norm": 0.8654264293321393,
      "learning_rate": 0.0002807191906480188,
      "loss": 2.0938,
      "step": 3152
    },
    {
      "epoch": 2.0880794701986756,
      "grad_norm": 0.7824684087055018,
      "learning_rate": 0.00028068644033213663,
      "loss": 2.2344,
      "step": 3153
    },
    {
      "epoch": 2.0887417218543045,
      "grad_norm": 0.9550076518351815,
      "learning_rate": 0.000280653664138841,
      "loss": 2.3594,
      "step": 3154
    },
    {
      "epoch": 2.089403973509934,
      "grad_norm": 0.9138409780467831,
      "learning_rate": 0.00028062086207462195,
      "loss": 2.1719,
      "step": 3155
    },
    {
      "epoch": 2.090066225165563,
      "grad_norm": 0.8675920601060703,
      "learning_rate": 0.0002805880341459747,
      "loss": 2.2969,
      "step": 3156
    },
    {
      "epoch": 2.0907284768211922,
      "grad_norm": 0.7901989534135593,
      "learning_rate": 0.0002805551803593995,
      "loss": 2.2031,
      "step": 3157
    },
    {
      "epoch": 2.091390728476821,
      "grad_norm": 0.8714151694948924,
      "learning_rate": 0.0002805223007214019,
      "loss": 2.2969,
      "step": 3158
    },
    {
      "epoch": 2.0920529801324506,
      "grad_norm": 0.7670837948045159,
      "learning_rate": 0.0002804893952384923,
      "loss": 1.8359,
      "step": 3159
    },
    {
      "epoch": 2.0927152317880795,
      "grad_norm": 0.8089192012055985,
      "learning_rate": 0.0002804564639171865,
      "loss": 2.2969,
      "step": 3160
    },
    {
      "epoch": 2.0933774834437084,
      "grad_norm": 0.6718848019256846,
      "learning_rate": 0.0002804235067640052,
      "loss": 1.6641,
      "step": 3161
    },
    {
      "epoch": 2.094039735099338,
      "grad_norm": 0.7069621690083887,
      "learning_rate": 0.00028039052378547424,
      "loss": 1.7891,
      "step": 3162
    },
    {
      "epoch": 2.0947019867549668,
      "grad_norm": 0.7467074023567285,
      "learning_rate": 0.00028035751498812475,
      "loss": 1.9062,
      "step": 3163
    },
    {
      "epoch": 2.095364238410596,
      "grad_norm": 0.7227779512173876,
      "learning_rate": 0.00028032448037849277,
      "loss": 1.6953,
      "step": 3164
    },
    {
      "epoch": 2.096026490066225,
      "grad_norm": 0.8376928072390296,
      "learning_rate": 0.0002802914199631195,
      "loss": 1.9922,
      "step": 3165
    },
    {
      "epoch": 2.0966887417218545,
      "grad_norm": 0.8034678708664186,
      "learning_rate": 0.0002802583337485514,
      "loss": 2.125,
      "step": 3166
    },
    {
      "epoch": 2.0973509933774834,
      "grad_norm": 0.8804627676317095,
      "learning_rate": 0.00028022522174133986,
      "loss": 2.4062,
      "step": 3167
    },
    {
      "epoch": 2.0980132450331124,
      "grad_norm": 0.9297674297291991,
      "learning_rate": 0.0002801920839480414,
      "loss": 2.1719,
      "step": 3168
    },
    {
      "epoch": 2.0986754966887418,
      "grad_norm": 0.8657994609469095,
      "learning_rate": 0.00028015892037521774,
      "loss": 2.3125,
      "step": 3169
    },
    {
      "epoch": 2.0993377483443707,
      "grad_norm": 0.7788842539952289,
      "learning_rate": 0.0002801257310294356,
      "loss": 2.0625,
      "step": 3170
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8869612062271457,
      "learning_rate": 0.0002800925159172669,
      "loss": 2.1719,
      "step": 3171
    },
    {
      "epoch": 2.100662251655629,
      "grad_norm": 0.8810528926701792,
      "learning_rate": 0.00028005927504528856,
      "loss": 2.5,
      "step": 3172
    },
    {
      "epoch": 2.1013245033112584,
      "grad_norm": 0.878707686483825,
      "learning_rate": 0.00028002600842008267,
      "loss": 2.2344,
      "step": 3173
    },
    {
      "epoch": 2.1019867549668874,
      "grad_norm": 0.8461113755429649,
      "learning_rate": 0.00027999271604823645,
      "loss": 2.0625,
      "step": 3174
    },
    {
      "epoch": 2.1026490066225167,
      "grad_norm": 0.7926432684279495,
      "learning_rate": 0.0002799593979363421,
      "loss": 2.375,
      "step": 3175
    },
    {
      "epoch": 2.1033112582781457,
      "grad_norm": 0.8229155785691155,
      "learning_rate": 0.00027992605409099704,
      "loss": 2.2812,
      "step": 3176
    },
    {
      "epoch": 2.1039735099337746,
      "grad_norm": 0.8720616583297459,
      "learning_rate": 0.00027989268451880374,
      "loss": 2.0781,
      "step": 3177
    },
    {
      "epoch": 2.104635761589404,
      "grad_norm": 0.9041400344498846,
      "learning_rate": 0.0002798592892263697,
      "loss": 2.5312,
      "step": 3178
    },
    {
      "epoch": 2.105298013245033,
      "grad_norm": 0.7435122171247798,
      "learning_rate": 0.0002798258682203076,
      "loss": 2.0312,
      "step": 3179
    },
    {
      "epoch": 2.1059602649006623,
      "grad_norm": 0.9496896881886252,
      "learning_rate": 0.0002797924215072352,
      "loss": 2.2969,
      "step": 3180
    },
    {
      "epoch": 2.1066225165562913,
      "grad_norm": 0.8611402563517535,
      "learning_rate": 0.0002797589490937754,
      "loss": 2.2656,
      "step": 3181
    },
    {
      "epoch": 2.1072847682119207,
      "grad_norm": 0.8187332982870288,
      "learning_rate": 0.0002797254509865559,
      "loss": 2.2344,
      "step": 3182
    },
    {
      "epoch": 2.1079470198675496,
      "grad_norm": 0.7861742796468008,
      "learning_rate": 0.00027969192719220993,
      "loss": 2.2188,
      "step": 3183
    },
    {
      "epoch": 2.108609271523179,
      "grad_norm": 0.8361914810712451,
      "learning_rate": 0.0002796583777173755,
      "loss": 1.9062,
      "step": 3184
    },
    {
      "epoch": 2.109271523178808,
      "grad_norm": 0.8873101087353407,
      "learning_rate": 0.0002796248025686958,
      "loss": 2.0312,
      "step": 3185
    },
    {
      "epoch": 2.109933774834437,
      "grad_norm": 0.7826078986265593,
      "learning_rate": 0.0002795912017528191,
      "loss": 1.9766,
      "step": 3186
    },
    {
      "epoch": 2.1105960264900663,
      "grad_norm": 0.7737710990102193,
      "learning_rate": 0.00027955757527639867,
      "loss": 1.9141,
      "step": 3187
    },
    {
      "epoch": 2.111258278145695,
      "grad_norm": 0.8069783206144504,
      "learning_rate": 0.0002795239231460931,
      "loss": 2.125,
      "step": 3188
    },
    {
      "epoch": 2.1119205298013246,
      "grad_norm": 0.8866085409493999,
      "learning_rate": 0.00027949024536856577,
      "loss": 2.4062,
      "step": 3189
    },
    {
      "epoch": 2.1125827814569536,
      "grad_norm": 0.8189520837471019,
      "learning_rate": 0.0002794565419504852,
      "loss": 2.125,
      "step": 3190
    },
    {
      "epoch": 2.113245033112583,
      "grad_norm": 0.8332274608180126,
      "learning_rate": 0.0002794228128985253,
      "loss": 2.2656,
      "step": 3191
    },
    {
      "epoch": 2.113907284768212,
      "grad_norm": 0.829259255187262,
      "learning_rate": 0.00027938905821936466,
      "loss": 2.25,
      "step": 3192
    },
    {
      "epoch": 2.1145695364238413,
      "grad_norm": 0.8144184450139884,
      "learning_rate": 0.000279355277919687,
      "loss": 1.9922,
      "step": 3193
    },
    {
      "epoch": 2.11523178807947,
      "grad_norm": 0.809308280103592,
      "learning_rate": 0.00027932147200618134,
      "loss": 1.8438,
      "step": 3194
    },
    {
      "epoch": 2.115894039735099,
      "grad_norm": 0.8177378772099362,
      "learning_rate": 0.00027928764048554164,
      "loss": 2.1562,
      "step": 3195
    },
    {
      "epoch": 2.1165562913907285,
      "grad_norm": 0.8935446760475563,
      "learning_rate": 0.0002792537833644668,
      "loss": 2.4531,
      "step": 3196
    },
    {
      "epoch": 2.1172185430463575,
      "grad_norm": 0.9399276644864418,
      "learning_rate": 0.00027921990064966106,
      "loss": 2.0625,
      "step": 3197
    },
    {
      "epoch": 2.117880794701987,
      "grad_norm": 0.7638756757537034,
      "learning_rate": 0.00027918599234783356,
      "loss": 1.6719,
      "step": 3198
    },
    {
      "epoch": 2.118543046357616,
      "grad_norm": 0.9448630519331106,
      "learning_rate": 0.0002791520584656985,
      "loss": 2.125,
      "step": 3199
    },
    {
      "epoch": 2.119205298013245,
      "grad_norm": 0.8166669794794548,
      "learning_rate": 0.00027911809900997515,
      "loss": 2.125,
      "step": 3200
    },
    {
      "epoch": 2.119867549668874,
      "grad_norm": 0.823302143531591,
      "learning_rate": 0.0002790841139873879,
      "loss": 1.8281,
      "step": 3201
    },
    {
      "epoch": 2.120529801324503,
      "grad_norm": 0.8108558188610947,
      "learning_rate": 0.00027905010340466616,
      "loss": 2.0781,
      "step": 3202
    },
    {
      "epoch": 2.1211920529801325,
      "grad_norm": 0.854018729909469,
      "learning_rate": 0.0002790160672685445,
      "loss": 2.4062,
      "step": 3203
    },
    {
      "epoch": 2.1218543046357614,
      "grad_norm": 0.8911762331053634,
      "learning_rate": 0.00027898200558576237,
      "loss": 1.8125,
      "step": 3204
    },
    {
      "epoch": 2.122516556291391,
      "grad_norm": 0.9254976710446435,
      "learning_rate": 0.0002789479183630644,
      "loss": 2.3438,
      "step": 3205
    },
    {
      "epoch": 2.1231788079470197,
      "grad_norm": 0.8610242415273918,
      "learning_rate": 0.00027891380560720024,
      "loss": 2.1875,
      "step": 3206
    },
    {
      "epoch": 2.123841059602649,
      "grad_norm": 0.8142693582448015,
      "learning_rate": 0.0002788796673249247,
      "loss": 2.1094,
      "step": 3207
    },
    {
      "epoch": 2.124503311258278,
      "grad_norm": 0.800838433199322,
      "learning_rate": 0.0002788455035229974,
      "loss": 2.3438,
      "step": 3208
    },
    {
      "epoch": 2.1251655629139075,
      "grad_norm": 0.7114633123115649,
      "learning_rate": 0.00027881131420818325,
      "loss": 1.9297,
      "step": 3209
    },
    {
      "epoch": 2.1258278145695364,
      "grad_norm": 0.7796314936291165,
      "learning_rate": 0.00027877709938725214,
      "loss": 1.8594,
      "step": 3210
    },
    {
      "epoch": 2.1264900662251653,
      "grad_norm": 0.9047998884298306,
      "learning_rate": 0.0002787428590669789,
      "loss": 2.5156,
      "step": 3211
    },
    {
      "epoch": 2.1271523178807947,
      "grad_norm": 0.8102270906604747,
      "learning_rate": 0.00027870859325414363,
      "loss": 2.1562,
      "step": 3212
    },
    {
      "epoch": 2.1278145695364237,
      "grad_norm": 0.8850990790499089,
      "learning_rate": 0.00027867430195553124,
      "loss": 2.1406,
      "step": 3213
    },
    {
      "epoch": 2.128476821192053,
      "grad_norm": 0.9041739049287638,
      "learning_rate": 0.000278639985177932,
      "loss": 2.2812,
      "step": 3214
    },
    {
      "epoch": 2.129139072847682,
      "grad_norm": 0.8249793512853625,
      "learning_rate": 0.00027860564292814076,
      "loss": 2.2188,
      "step": 3215
    },
    {
      "epoch": 2.1298013245033114,
      "grad_norm": 0.7939861418613539,
      "learning_rate": 0.0002785712752129578,
      "loss": 2.1562,
      "step": 3216
    },
    {
      "epoch": 2.1304635761589403,
      "grad_norm": 0.861220834938404,
      "learning_rate": 0.0002785368820391884,
      "loss": 2.3281,
      "step": 3217
    },
    {
      "epoch": 2.1311258278145697,
      "grad_norm": 0.7821094289211991,
      "learning_rate": 0.00027850246341364266,
      "loss": 2.0469,
      "step": 3218
    },
    {
      "epoch": 2.1317880794701987,
      "grad_norm": 0.7976263234562335,
      "learning_rate": 0.000278468019343136,
      "loss": 2.1875,
      "step": 3219
    },
    {
      "epoch": 2.1324503311258276,
      "grad_norm": 0.8077208724635396,
      "learning_rate": 0.00027843354983448867,
      "loss": 2.2344,
      "step": 3220
    },
    {
      "epoch": 2.133112582781457,
      "grad_norm": 0.7884602463280347,
      "learning_rate": 0.00027839905489452597,
      "loss": 2.0781,
      "step": 3221
    },
    {
      "epoch": 2.133774834437086,
      "grad_norm": 0.8085181043787424,
      "learning_rate": 0.00027836453453007837,
      "loss": 2.0156,
      "step": 3222
    },
    {
      "epoch": 2.1344370860927153,
      "grad_norm": 0.8606845761531365,
      "learning_rate": 0.0002783299887479813,
      "loss": 2.25,
      "step": 3223
    },
    {
      "epoch": 2.1350993377483443,
      "grad_norm": 0.9822583445688104,
      "learning_rate": 0.00027829541755507517,
      "loss": 2.3125,
      "step": 3224
    },
    {
      "epoch": 2.1357615894039736,
      "grad_norm": 0.8485399732284782,
      "learning_rate": 0.0002782608209582056,
      "loss": 2.2344,
      "step": 3225
    },
    {
      "epoch": 2.1364238410596026,
      "grad_norm": 0.9110699370383477,
      "learning_rate": 0.00027822619896422293,
      "loss": 2.5156,
      "step": 3226
    },
    {
      "epoch": 2.137086092715232,
      "grad_norm": 0.7998480976676432,
      "learning_rate": 0.0002781915515799828,
      "loss": 2.125,
      "step": 3227
    },
    {
      "epoch": 2.137748344370861,
      "grad_norm": 0.7787851916355009,
      "learning_rate": 0.0002781568788123458,
      "loss": 2.0156,
      "step": 3228
    },
    {
      "epoch": 2.13841059602649,
      "grad_norm": 0.8132257640511175,
      "learning_rate": 0.0002781221806681775,
      "loss": 2.0469,
      "step": 3229
    },
    {
      "epoch": 2.1390728476821192,
      "grad_norm": 0.8502061427062879,
      "learning_rate": 0.00027808745715434866,
      "loss": 2.1719,
      "step": 3230
    },
    {
      "epoch": 2.139735099337748,
      "grad_norm": 0.8815362692246591,
      "learning_rate": 0.00027805270827773473,
      "loss": 2.3594,
      "step": 3231
    },
    {
      "epoch": 2.1403973509933776,
      "grad_norm": 0.9117495523163015,
      "learning_rate": 0.00027801793404521653,
      "loss": 2.4219,
      "step": 3232
    },
    {
      "epoch": 2.1410596026490065,
      "grad_norm": 0.9410253026495625,
      "learning_rate": 0.0002779831344636797,
      "loss": 2.25,
      "step": 3233
    },
    {
      "epoch": 2.141721854304636,
      "grad_norm": 0.8857225887521325,
      "learning_rate": 0.00027794830954001496,
      "loss": 1.9766,
      "step": 3234
    },
    {
      "epoch": 2.142384105960265,
      "grad_norm": 0.8247598418075782,
      "learning_rate": 0.00027791345928111804,
      "loss": 2.3281,
      "step": 3235
    },
    {
      "epoch": 2.1430463576158942,
      "grad_norm": 0.798738937985844,
      "learning_rate": 0.00027787858369388973,
      "loss": 1.9453,
      "step": 3236
    },
    {
      "epoch": 2.143708609271523,
      "grad_norm": 0.8654240560178093,
      "learning_rate": 0.00027784368278523574,
      "loss": 2.2812,
      "step": 3237
    },
    {
      "epoch": 2.144370860927152,
      "grad_norm": 0.8855302169022395,
      "learning_rate": 0.0002778087565620669,
      "loss": 2.2188,
      "step": 3238
    },
    {
      "epoch": 2.1450331125827815,
      "grad_norm": 0.7676852723027048,
      "learning_rate": 0.00027777380503129893,
      "loss": 2.1562,
      "step": 3239
    },
    {
      "epoch": 2.1456953642384105,
      "grad_norm": 0.8381182494245285,
      "learning_rate": 0.0002777388281998528,
      "loss": 1.8828,
      "step": 3240
    },
    {
      "epoch": 2.14635761589404,
      "grad_norm": 0.7969592405784364,
      "learning_rate": 0.00027770382607465407,
      "loss": 1.8125,
      "step": 3241
    },
    {
      "epoch": 2.147019867549669,
      "grad_norm": 0.9181325691003049,
      "learning_rate": 0.00027766879866263376,
      "loss": 2.2344,
      "step": 3242
    },
    {
      "epoch": 2.147682119205298,
      "grad_norm": 0.8671542383801918,
      "learning_rate": 0.0002776337459707277,
      "loss": 2.1094,
      "step": 3243
    },
    {
      "epoch": 2.148344370860927,
      "grad_norm": 0.8123017923607242,
      "learning_rate": 0.0002775986680058766,
      "loss": 2.2188,
      "step": 3244
    },
    {
      "epoch": 2.1490066225165565,
      "grad_norm": 0.8090885548036824,
      "learning_rate": 0.0002775635647750264,
      "loss": 1.8906,
      "step": 3245
    },
    {
      "epoch": 2.1496688741721854,
      "grad_norm": 0.8521149077192472,
      "learning_rate": 0.00027752843628512785,
      "loss": 2.1719,
      "step": 3246
    },
    {
      "epoch": 2.1503311258278144,
      "grad_norm": 0.8645924082368771,
      "learning_rate": 0.0002774932825431369,
      "loss": 1.9375,
      "step": 3247
    },
    {
      "epoch": 2.1509933774834438,
      "grad_norm": 0.9222697095775695,
      "learning_rate": 0.00027745810355601425,
      "loss": 2.4688,
      "step": 3248
    },
    {
      "epoch": 2.1516556291390727,
      "grad_norm": 0.846568195720941,
      "learning_rate": 0.000277422899330726,
      "loss": 2.0312,
      "step": 3249
    },
    {
      "epoch": 2.152317880794702,
      "grad_norm": 0.9426840800579832,
      "learning_rate": 0.0002773876698742427,
      "loss": 2.5312,
      "step": 3250
    },
    {
      "epoch": 2.152980132450331,
      "grad_norm": 0.860878534310882,
      "learning_rate": 0.0002773524151935403,
      "loss": 2.375,
      "step": 3251
    },
    {
      "epoch": 2.1536423841059604,
      "grad_norm": 0.8100590988887295,
      "learning_rate": 0.0002773171352955997,
      "loss": 2.2031,
      "step": 3252
    },
    {
      "epoch": 2.1543046357615894,
      "grad_norm": 0.8948308204547479,
      "learning_rate": 0.0002772818301874067,
      "loss": 2.3906,
      "step": 3253
    },
    {
      "epoch": 2.1549668874172188,
      "grad_norm": 0.7919869371471651,
      "learning_rate": 0.0002772464998759521,
      "loss": 2.1875,
      "step": 3254
    },
    {
      "epoch": 2.1556291390728477,
      "grad_norm": 0.8043753337644759,
      "learning_rate": 0.0002772111443682316,
      "loss": 1.9844,
      "step": 3255
    },
    {
      "epoch": 2.1562913907284766,
      "grad_norm": 0.7974534895558583,
      "learning_rate": 0.0002771757636712461,
      "loss": 2.2031,
      "step": 3256
    },
    {
      "epoch": 2.156953642384106,
      "grad_norm": 0.7741637095045127,
      "learning_rate": 0.00027714035779200137,
      "loss": 1.7891,
      "step": 3257
    },
    {
      "epoch": 2.157615894039735,
      "grad_norm": 0.882262510540357,
      "learning_rate": 0.00027710492673750826,
      "loss": 2.2969,
      "step": 3258
    },
    {
      "epoch": 2.1582781456953644,
      "grad_norm": 0.8788913504587818,
      "learning_rate": 0.00027706947051478237,
      "loss": 2.2656,
      "step": 3259
    },
    {
      "epoch": 2.1589403973509933,
      "grad_norm": 0.8957276353185788,
      "learning_rate": 0.00027703398913084455,
      "loss": 2.3594,
      "step": 3260
    },
    {
      "epoch": 2.1596026490066227,
      "grad_norm": 0.826726458575409,
      "learning_rate": 0.00027699848259272043,
      "loss": 2.3281,
      "step": 3261
    },
    {
      "epoch": 2.1602649006622516,
      "grad_norm": 0.792872375528584,
      "learning_rate": 0.00027696295090744085,
      "loss": 2.0938,
      "step": 3262
    },
    {
      "epoch": 2.160927152317881,
      "grad_norm": 0.8331968500724979,
      "learning_rate": 0.00027692739408204136,
      "loss": 2.3594,
      "step": 3263
    },
    {
      "epoch": 2.16158940397351,
      "grad_norm": 0.7736567263164781,
      "learning_rate": 0.00027689181212356266,
      "loss": 1.8594,
      "step": 3264
    },
    {
      "epoch": 2.162251655629139,
      "grad_norm": 0.852998208040114,
      "learning_rate": 0.00027685620503905043,
      "loss": 2.4531,
      "step": 3265
    },
    {
      "epoch": 2.1629139072847683,
      "grad_norm": 0.8171302746452069,
      "learning_rate": 0.0002768205728355552,
      "loss": 2.2812,
      "step": 3266
    },
    {
      "epoch": 2.1635761589403972,
      "grad_norm": 0.7962997632248591,
      "learning_rate": 0.0002767849155201326,
      "loss": 2.1094,
      "step": 3267
    },
    {
      "epoch": 2.1642384105960266,
      "grad_norm": 0.8250086457383703,
      "learning_rate": 0.0002767492330998432,
      "loss": 2.3594,
      "step": 3268
    },
    {
      "epoch": 2.1649006622516556,
      "grad_norm": 0.7382925284379064,
      "learning_rate": 0.00027671352558175254,
      "loss": 2.0469,
      "step": 3269
    },
    {
      "epoch": 2.165562913907285,
      "grad_norm": 0.8640181501937213,
      "learning_rate": 0.00027667779297293105,
      "loss": 2.1719,
      "step": 3270
    },
    {
      "epoch": 2.166225165562914,
      "grad_norm": 0.8040154250911639,
      "learning_rate": 0.00027664203528045424,
      "loss": 2.0625,
      "step": 3271
    },
    {
      "epoch": 2.1668874172185433,
      "grad_norm": 0.8007769382247858,
      "learning_rate": 0.00027660625251140256,
      "loss": 2.2969,
      "step": 3272
    },
    {
      "epoch": 2.167549668874172,
      "grad_norm": 0.9299384799368382,
      "learning_rate": 0.0002765704446728614,
      "loss": 2.4531,
      "step": 3273
    },
    {
      "epoch": 2.168211920529801,
      "grad_norm": 0.8825890929324891,
      "learning_rate": 0.0002765346117719211,
      "loss": 2.1875,
      "step": 3274
    },
    {
      "epoch": 2.1688741721854305,
      "grad_norm": 0.852743835362979,
      "learning_rate": 0.00027649875381567695,
      "loss": 2.0156,
      "step": 3275
    },
    {
      "epoch": 2.1695364238410595,
      "grad_norm": 0.7881463871581443,
      "learning_rate": 0.00027646287081122933,
      "loss": 2.2344,
      "step": 3276
    },
    {
      "epoch": 2.170198675496689,
      "grad_norm": 0.8206073467339754,
      "learning_rate": 0.00027642696276568345,
      "loss": 2.125,
      "step": 3277
    },
    {
      "epoch": 2.170860927152318,
      "grad_norm": 0.8281263956557895,
      "learning_rate": 0.0002763910296861495,
      "loss": 2.1875,
      "step": 3278
    },
    {
      "epoch": 2.171523178807947,
      "grad_norm": 0.882125613475579,
      "learning_rate": 0.0002763550715797426,
      "loss": 2.3594,
      "step": 3279
    },
    {
      "epoch": 2.172185430463576,
      "grad_norm": 0.8278852714288011,
      "learning_rate": 0.00027631908845358297,
      "loss": 2.0469,
      "step": 3280
    },
    {
      "epoch": 2.172847682119205,
      "grad_norm": 0.785570592764887,
      "learning_rate": 0.0002762830803147956,
      "loss": 2.0312,
      "step": 3281
    },
    {
      "epoch": 2.1735099337748345,
      "grad_norm": 0.7919405724412538,
      "learning_rate": 0.0002762470471705105,
      "loss": 2.0938,
      "step": 3282
    },
    {
      "epoch": 2.1741721854304634,
      "grad_norm": 0.837810773588796,
      "learning_rate": 0.0002762109890278628,
      "loss": 2.3906,
      "step": 3283
    },
    {
      "epoch": 2.174834437086093,
      "grad_norm": 0.7872660246620158,
      "learning_rate": 0.00027617490589399227,
      "loss": 2.0938,
      "step": 3284
    },
    {
      "epoch": 2.1754966887417218,
      "grad_norm": 0.8323316699109079,
      "learning_rate": 0.00027613879777604373,
      "loss": 1.9922,
      "step": 3285
    },
    {
      "epoch": 2.176158940397351,
      "grad_norm": 0.8018992494949806,
      "learning_rate": 0.0002761026646811672,
      "loss": 2.0156,
      "step": 3286
    },
    {
      "epoch": 2.17682119205298,
      "grad_norm": 0.8302308794928105,
      "learning_rate": 0.0002760665066165173,
      "loss": 2.1875,
      "step": 3287
    },
    {
      "epoch": 2.1774834437086095,
      "grad_norm": 0.7558837631959101,
      "learning_rate": 0.00027603032358925383,
      "loss": 2.1406,
      "step": 3288
    },
    {
      "epoch": 2.1781456953642384,
      "grad_norm": 0.7632790741462315,
      "learning_rate": 0.0002759941156065413,
      "loss": 1.9688,
      "step": 3289
    },
    {
      "epoch": 2.1788079470198674,
      "grad_norm": 0.7733118085391162,
      "learning_rate": 0.0002759578826755495,
      "loss": 2.0469,
      "step": 3290
    },
    {
      "epoch": 2.1794701986754967,
      "grad_norm": 0.8137496040176969,
      "learning_rate": 0.00027592162480345283,
      "loss": 2.0781,
      "step": 3291
    },
    {
      "epoch": 2.1801324503311257,
      "grad_norm": 0.99050130317897,
      "learning_rate": 0.0002758853419974308,
      "loss": 2.4219,
      "step": 3292
    },
    {
      "epoch": 2.180794701986755,
      "grad_norm": 0.8032197797888396,
      "learning_rate": 0.0002758490342646678,
      "loss": 2.2188,
      "step": 3293
    },
    {
      "epoch": 2.181456953642384,
      "grad_norm": 0.8396491976663559,
      "learning_rate": 0.0002758127016123533,
      "loss": 2.3594,
      "step": 3294
    },
    {
      "epoch": 2.1821192052980134,
      "grad_norm": 0.8255061920189865,
      "learning_rate": 0.0002757763440476814,
      "loss": 2.4375,
      "step": 3295
    },
    {
      "epoch": 2.1827814569536423,
      "grad_norm": 0.8489285029923598,
      "learning_rate": 0.0002757399615778514,
      "loss": 1.8516,
      "step": 3296
    },
    {
      "epoch": 2.1834437086092717,
      "grad_norm": 0.8508286114644342,
      "learning_rate": 0.00027570355421006746,
      "loss": 2.1562,
      "step": 3297
    },
    {
      "epoch": 2.1841059602649007,
      "grad_norm": 0.7562689755418857,
      "learning_rate": 0.00027566712195153866,
      "loss": 1.9922,
      "step": 3298
    },
    {
      "epoch": 2.1847682119205296,
      "grad_norm": 0.8787242466692003,
      "learning_rate": 0.000275630664809479,
      "loss": 2.2969,
      "step": 3299
    },
    {
      "epoch": 2.185430463576159,
      "grad_norm": 0.9344491983192744,
      "learning_rate": 0.00027559418279110733,
      "loss": 2.1875,
      "step": 3300
    },
    {
      "epoch": 2.186092715231788,
      "grad_norm": 0.8252876463159317,
      "learning_rate": 0.0002755576759036476,
      "loss": 2.2188,
      "step": 3301
    },
    {
      "epoch": 2.1867549668874173,
      "grad_norm": 0.7956282359832736,
      "learning_rate": 0.00027552114415432864,
      "loss": 2.2031,
      "step": 3302
    },
    {
      "epoch": 2.1874172185430463,
      "grad_norm": 0.7222801729364425,
      "learning_rate": 0.00027548458755038404,
      "loss": 1.7344,
      "step": 3303
    },
    {
      "epoch": 2.1880794701986757,
      "grad_norm": 0.7940695992711349,
      "learning_rate": 0.00027544800609905244,
      "loss": 1.9766,
      "step": 3304
    },
    {
      "epoch": 2.1887417218543046,
      "grad_norm": 0.8861022600103349,
      "learning_rate": 0.0002754113998075775,
      "loss": 1.9766,
      "step": 3305
    },
    {
      "epoch": 2.1894039735099335,
      "grad_norm": 1.0031515963194808,
      "learning_rate": 0.0002753747686832075,
      "loss": 2.3281,
      "step": 3306
    },
    {
      "epoch": 2.190066225165563,
      "grad_norm": 0.7601196397115809,
      "learning_rate": 0.000275338112733196,
      "loss": 2.0625,
      "step": 3307
    },
    {
      "epoch": 2.190728476821192,
      "grad_norm": 0.8613390575439762,
      "learning_rate": 0.0002753014319648012,
      "loss": 2.2344,
      "step": 3308
    },
    {
      "epoch": 2.1913907284768213,
      "grad_norm": 0.7556465928778803,
      "learning_rate": 0.00027526472638528634,
      "loss": 1.8125,
      "step": 3309
    },
    {
      "epoch": 2.19205298013245,
      "grad_norm": 0.8319931456644523,
      "learning_rate": 0.0002752279960019195,
      "loss": 2.1094,
      "step": 3310
    },
    {
      "epoch": 2.1927152317880796,
      "grad_norm": 0.7421980487514149,
      "learning_rate": 0.0002751912408219738,
      "loss": 1.7422,
      "step": 3311
    },
    {
      "epoch": 2.1933774834437085,
      "grad_norm": 0.8275927958504509,
      "learning_rate": 0.0002751544608527272,
      "loss": 2.1875,
      "step": 3312
    },
    {
      "epoch": 2.194039735099338,
      "grad_norm": 0.8507836818395664,
      "learning_rate": 0.0002751176561014624,
      "loss": 2.2969,
      "step": 3313
    },
    {
      "epoch": 2.194701986754967,
      "grad_norm": 0.8285849358506636,
      "learning_rate": 0.00027508082657546734,
      "loss": 2.3281,
      "step": 3314
    },
    {
      "epoch": 2.195364238410596,
      "grad_norm": 0.766382384247634,
      "learning_rate": 0.00027504397228203455,
      "loss": 1.9453,
      "step": 3315
    },
    {
      "epoch": 2.196026490066225,
      "grad_norm": 0.8843905447887204,
      "learning_rate": 0.00027500709322846175,
      "loss": 2.5156,
      "step": 3316
    },
    {
      "epoch": 2.196688741721854,
      "grad_norm": 0.7845108208485224,
      "learning_rate": 0.00027497018942205126,
      "loss": 2.25,
      "step": 3317
    },
    {
      "epoch": 2.1973509933774835,
      "grad_norm": 0.8256990756936016,
      "learning_rate": 0.00027493326087011055,
      "loss": 2.2031,
      "step": 3318
    },
    {
      "epoch": 2.1980132450331125,
      "grad_norm": 0.7968136572426746,
      "learning_rate": 0.0002748963075799519,
      "loss": 2.2812,
      "step": 3319
    },
    {
      "epoch": 2.198675496688742,
      "grad_norm": 0.8327481165603114,
      "learning_rate": 0.0002748593295588924,
      "loss": 2.25,
      "step": 3320
    },
    {
      "epoch": 2.199337748344371,
      "grad_norm": 0.8198669467615958,
      "learning_rate": 0.00027482232681425424,
      "loss": 2.25,
      "step": 3321
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.7417890205642291,
      "learning_rate": 0.0002747852993533643,
      "loss": 1.9688,
      "step": 3322
    },
    {
      "epoch": 2.200662251655629,
      "grad_norm": 0.7768295405464892,
      "learning_rate": 0.00027474824718355445,
      "loss": 2.0938,
      "step": 3323
    },
    {
      "epoch": 2.201324503311258,
      "grad_norm": 0.9182396334096297,
      "learning_rate": 0.00027471117031216153,
      "loss": 2.3906,
      "step": 3324
    },
    {
      "epoch": 2.2019867549668874,
      "grad_norm": 0.9132219842886142,
      "learning_rate": 0.00027467406874652706,
      "loss": 2.1406,
      "step": 3325
    },
    {
      "epoch": 2.2026490066225164,
      "grad_norm": 0.8051668784575603,
      "learning_rate": 0.00027463694249399765,
      "loss": 1.8438,
      "step": 3326
    },
    {
      "epoch": 2.203311258278146,
      "grad_norm": 0.8779608231869319,
      "learning_rate": 0.0002745997915619247,
      "loss": 2.4062,
      "step": 3327
    },
    {
      "epoch": 2.2039735099337747,
      "grad_norm": 0.7657089739207261,
      "learning_rate": 0.0002745626159576646,
      "loss": 1.6797,
      "step": 3328
    },
    {
      "epoch": 2.204635761589404,
      "grad_norm": 0.8485164960940961,
      "learning_rate": 0.00027452541568857847,
      "loss": 2.0156,
      "step": 3329
    },
    {
      "epoch": 2.205298013245033,
      "grad_norm": 0.8355818547700689,
      "learning_rate": 0.0002744881907620323,
      "loss": 2.1875,
      "step": 3330
    },
    {
      "epoch": 2.2059602649006624,
      "grad_norm": 0.8558865706646701,
      "learning_rate": 0.0002744509411853972,
      "loss": 2.2656,
      "step": 3331
    },
    {
      "epoch": 2.2066225165562914,
      "grad_norm": 0.7968607240104667,
      "learning_rate": 0.00027441366696604895,
      "loss": 2.3438,
      "step": 3332
    },
    {
      "epoch": 2.2072847682119203,
      "grad_norm": 0.8115129909848648,
      "learning_rate": 0.0002743763681113683,
      "loss": 2.1719,
      "step": 3333
    },
    {
      "epoch": 2.2079470198675497,
      "grad_norm": 0.7745359407731947,
      "learning_rate": 0.0002743390446287409,
      "loss": 2.2812,
      "step": 3334
    },
    {
      "epoch": 2.2086092715231787,
      "grad_norm": 0.8870758405984903,
      "learning_rate": 0.00027430169652555703,
      "loss": 2.6406,
      "step": 3335
    },
    {
      "epoch": 2.209271523178808,
      "grad_norm": 0.8388509120974267,
      "learning_rate": 0.0002742643238092123,
      "loss": 2.375,
      "step": 3336
    },
    {
      "epoch": 2.209933774834437,
      "grad_norm": 0.7701954031581828,
      "learning_rate": 0.00027422692648710673,
      "loss": 2.1719,
      "step": 3337
    },
    {
      "epoch": 2.2105960264900664,
      "grad_norm": 0.7700654296594206,
      "learning_rate": 0.0002741895045666455,
      "loss": 2.0781,
      "step": 3338
    },
    {
      "epoch": 2.2112582781456953,
      "grad_norm": 0.7572760178076289,
      "learning_rate": 0.0002741520580552386,
      "loss": 1.9766,
      "step": 3339
    },
    {
      "epoch": 2.2119205298013247,
      "grad_norm": 0.7811646489989366,
      "learning_rate": 0.0002741145869603009,
      "loss": 1.9766,
      "step": 3340
    },
    {
      "epoch": 2.2125827814569536,
      "grad_norm": 0.7652073370408187,
      "learning_rate": 0.000274077091289252,
      "loss": 1.6328,
      "step": 3341
    },
    {
      "epoch": 2.2132450331125826,
      "grad_norm": 0.735328194049772,
      "learning_rate": 0.0002740395710495165,
      "loss": 1.8203,
      "step": 3342
    },
    {
      "epoch": 2.213907284768212,
      "grad_norm": 0.8286038328340698,
      "learning_rate": 0.00027400202624852394,
      "loss": 2.0625,
      "step": 3343
    },
    {
      "epoch": 2.214569536423841,
      "grad_norm": 0.8768491364725021,
      "learning_rate": 0.0002739644568937085,
      "loss": 2.2031,
      "step": 3344
    },
    {
      "epoch": 2.2152317880794703,
      "grad_norm": 0.8552570456101776,
      "learning_rate": 0.0002739268629925094,
      "loss": 2.2656,
      "step": 3345
    },
    {
      "epoch": 2.2158940397350992,
      "grad_norm": 0.9068003321966285,
      "learning_rate": 0.00027388924455237063,
      "loss": 2.125,
      "step": 3346
    },
    {
      "epoch": 2.2165562913907286,
      "grad_norm": 0.8675120392327351,
      "learning_rate": 0.00027385160158074115,
      "loss": 2.1094,
      "step": 3347
    },
    {
      "epoch": 2.2172185430463576,
      "grad_norm": 0.8327555600723997,
      "learning_rate": 0.00027381393408507456,
      "loss": 2.0781,
      "step": 3348
    },
    {
      "epoch": 2.217880794701987,
      "grad_norm": 0.8418742450561617,
      "learning_rate": 0.0002737762420728296,
      "loss": 2.2188,
      "step": 3349
    },
    {
      "epoch": 2.218543046357616,
      "grad_norm": 0.8699429060159691,
      "learning_rate": 0.0002737385255514697,
      "loss": 2.3906,
      "step": 3350
    },
    {
      "epoch": 2.219205298013245,
      "grad_norm": 0.8509907393944772,
      "learning_rate": 0.000273700784528463,
      "loss": 2.3438,
      "step": 3351
    },
    {
      "epoch": 2.2198675496688742,
      "grad_norm": 0.8841108347587866,
      "learning_rate": 0.0002736630190112829,
      "loss": 2.0938,
      "step": 3352
    },
    {
      "epoch": 2.220529801324503,
      "grad_norm": 0.8334422899623337,
      "learning_rate": 0.0002736252290074072,
      "loss": 1.9453,
      "step": 3353
    },
    {
      "epoch": 2.2211920529801326,
      "grad_norm": 0.7592264525080054,
      "learning_rate": 0.0002735874145243189,
      "loss": 1.8125,
      "step": 3354
    },
    {
      "epoch": 2.2218543046357615,
      "grad_norm": 0.7825530875598116,
      "learning_rate": 0.0002735495755695056,
      "loss": 2.2344,
      "step": 3355
    },
    {
      "epoch": 2.222516556291391,
      "grad_norm": 0.8848044485857259,
      "learning_rate": 0.0002735117121504599,
      "loss": 2.3438,
      "step": 3356
    },
    {
      "epoch": 2.22317880794702,
      "grad_norm": 0.8737085402343399,
      "learning_rate": 0.0002734738242746792,
      "loss": 2.2812,
      "step": 3357
    },
    {
      "epoch": 2.223841059602649,
      "grad_norm": 0.8442083587277228,
      "learning_rate": 0.0002734359119496657,
      "loss": 2.2344,
      "step": 3358
    },
    {
      "epoch": 2.224503311258278,
      "grad_norm": 0.7677761452029833,
      "learning_rate": 0.0002733979751829265,
      "loss": 2.0781,
      "step": 3359
    },
    {
      "epoch": 2.225165562913907,
      "grad_norm": 0.8103208074750657,
      "learning_rate": 0.0002733600139819735,
      "loss": 2.3594,
      "step": 3360
    },
    {
      "epoch": 2.2258278145695365,
      "grad_norm": 0.7631405913554564,
      "learning_rate": 0.00027332202835432344,
      "loss": 2.0938,
      "step": 3361
    },
    {
      "epoch": 2.2264900662251654,
      "grad_norm": 0.8657307515518499,
      "learning_rate": 0.0002732840183074979,
      "loss": 2.2969,
      "step": 3362
    },
    {
      "epoch": 2.227152317880795,
      "grad_norm": 0.8110410252870841,
      "learning_rate": 0.00027324598384902333,
      "loss": 2.1406,
      "step": 3363
    },
    {
      "epoch": 2.2278145695364238,
      "grad_norm": 0.8640540740701811,
      "learning_rate": 0.000273207924986431,
      "loss": 2.0312,
      "step": 3364
    },
    {
      "epoch": 2.228476821192053,
      "grad_norm": 0.7989624386226248,
      "learning_rate": 0.00027316984172725704,
      "loss": 2.2031,
      "step": 3365
    },
    {
      "epoch": 2.229139072847682,
      "grad_norm": 0.7494145080518465,
      "learning_rate": 0.0002731317340790423,
      "loss": 2.1406,
      "step": 3366
    },
    {
      "epoch": 2.2298013245033115,
      "grad_norm": 0.8311134685557905,
      "learning_rate": 0.00027309360204933256,
      "loss": 2.0938,
      "step": 3367
    },
    {
      "epoch": 2.2304635761589404,
      "grad_norm": 0.7508644414489907,
      "learning_rate": 0.0002730554456456783,
      "loss": 2.1406,
      "step": 3368
    },
    {
      "epoch": 2.2311258278145694,
      "grad_norm": 0.7550064483810309,
      "learning_rate": 0.00027301726487563517,
      "loss": 1.9453,
      "step": 3369
    },
    {
      "epoch": 2.2317880794701987,
      "grad_norm": 0.7435367177173581,
      "learning_rate": 0.00027297905974676316,
      "loss": 1.9297,
      "step": 3370
    },
    {
      "epoch": 2.2324503311258277,
      "grad_norm": 0.8487484380639131,
      "learning_rate": 0.0002729408302666274,
      "loss": 2.1875,
      "step": 3371
    },
    {
      "epoch": 2.233112582781457,
      "grad_norm": 0.7569325897381546,
      "learning_rate": 0.0002729025764427978,
      "loss": 2.0156,
      "step": 3372
    },
    {
      "epoch": 2.233774834437086,
      "grad_norm": 0.8441153267497679,
      "learning_rate": 0.000272864298282849,
      "loss": 2.2188,
      "step": 3373
    },
    {
      "epoch": 2.2344370860927154,
      "grad_norm": 0.8027582806848588,
      "learning_rate": 0.0002728259957943606,
      "loss": 2.1562,
      "step": 3374
    },
    {
      "epoch": 2.2350993377483444,
      "grad_norm": 0.8066221003452809,
      "learning_rate": 0.00027278766898491686,
      "loss": 2.0781,
      "step": 3375
    },
    {
      "epoch": 2.2357615894039737,
      "grad_norm": 0.8281426998927284,
      "learning_rate": 0.00027274931786210694,
      "loss": 2.2188,
      "step": 3376
    },
    {
      "epoch": 2.2364238410596027,
      "grad_norm": 0.8218323511052216,
      "learning_rate": 0.0002727109424335248,
      "loss": 2.2031,
      "step": 3377
    },
    {
      "epoch": 2.2370860927152316,
      "grad_norm": 0.7743687151133861,
      "learning_rate": 0.00027267254270676923,
      "loss": 2.1406,
      "step": 3378
    },
    {
      "epoch": 2.237748344370861,
      "grad_norm": 0.794579298472305,
      "learning_rate": 0.0002726341186894438,
      "loss": 1.8047,
      "step": 3379
    },
    {
      "epoch": 2.23841059602649,
      "grad_norm": 0.8208912133462046,
      "learning_rate": 0.00027259567038915694,
      "loss": 2.3281,
      "step": 3380
    },
    {
      "epoch": 2.2390728476821193,
      "grad_norm": 0.8330499760238012,
      "learning_rate": 0.0002725571978135218,
      "loss": 2.0469,
      "step": 3381
    },
    {
      "epoch": 2.2397350993377483,
      "grad_norm": 0.7804178423215485,
      "learning_rate": 0.00027251870097015645,
      "loss": 2.0781,
      "step": 3382
    },
    {
      "epoch": 2.2403973509933777,
      "grad_norm": 0.7975006795801435,
      "learning_rate": 0.0002724801798666837,
      "loss": 2.2656,
      "step": 3383
    },
    {
      "epoch": 2.2410596026490066,
      "grad_norm": 0.8617674106901189,
      "learning_rate": 0.00027244163451073107,
      "loss": 2.2969,
      "step": 3384
    },
    {
      "epoch": 2.241721854304636,
      "grad_norm": 0.7248057874259806,
      "learning_rate": 0.0002724030649099311,
      "loss": 1.7344,
      "step": 3385
    },
    {
      "epoch": 2.242384105960265,
      "grad_norm": 0.7608766067926409,
      "learning_rate": 0.000272364471071921,
      "loss": 2.0312,
      "step": 3386
    },
    {
      "epoch": 2.243046357615894,
      "grad_norm": 0.8349490815247091,
      "learning_rate": 0.00027232585300434274,
      "loss": 2.1406,
      "step": 3387
    },
    {
      "epoch": 2.2437086092715233,
      "grad_norm": 0.8948656507471446,
      "learning_rate": 0.0002722872107148431,
      "loss": 2.2969,
      "step": 3388
    },
    {
      "epoch": 2.244370860927152,
      "grad_norm": 0.8515136297390447,
      "learning_rate": 0.0002722485442110739,
      "loss": 2.1094,
      "step": 3389
    },
    {
      "epoch": 2.2450331125827816,
      "grad_norm": 0.7520520842508782,
      "learning_rate": 0.0002722098535006913,
      "loss": 1.9453,
      "step": 3390
    },
    {
      "epoch": 2.2456953642384105,
      "grad_norm": 0.7633417223611352,
      "learning_rate": 0.00027217113859135663,
      "loss": 2.0312,
      "step": 3391
    },
    {
      "epoch": 2.24635761589404,
      "grad_norm": 0.8743428750761952,
      "learning_rate": 0.0002721323994907359,
      "loss": 2.5312,
      "step": 3392
    },
    {
      "epoch": 2.247019867549669,
      "grad_norm": 0.8852405185858262,
      "learning_rate": 0.00027209363620649986,
      "loss": 2.2969,
      "step": 3393
    },
    {
      "epoch": 2.247682119205298,
      "grad_norm": 0.8798641678710732,
      "learning_rate": 0.0002720548487463241,
      "loss": 2.4688,
      "step": 3394
    },
    {
      "epoch": 2.248344370860927,
      "grad_norm": 0.7365239863128915,
      "learning_rate": 0.000272016037117889,
      "loss": 1.75,
      "step": 3395
    },
    {
      "epoch": 2.249006622516556,
      "grad_norm": 0.835785839902504,
      "learning_rate": 0.00027197720132887964,
      "loss": 1.8516,
      "step": 3396
    },
    {
      "epoch": 2.2496688741721855,
      "grad_norm": 0.7508971485986422,
      "learning_rate": 0.000271938341386986,
      "loss": 1.9297,
      "step": 3397
    },
    {
      "epoch": 2.2503311258278145,
      "grad_norm": 0.777671344037389,
      "learning_rate": 0.0002718994572999028,
      "loss": 2.2344,
      "step": 3398
    },
    {
      "epoch": 2.250993377483444,
      "grad_norm": 0.7987089376070218,
      "learning_rate": 0.0002718605490753295,
      "loss": 2.1094,
      "step": 3399
    },
    {
      "epoch": 2.251655629139073,
      "grad_norm": 0.7933753326309906,
      "learning_rate": 0.00027182161672097046,
      "loss": 2.1406,
      "step": 3400
    },
    {
      "epoch": 2.2523178807947017,
      "grad_norm": 0.7271138779708894,
      "learning_rate": 0.0002717826602445346,
      "loss": 2.0156,
      "step": 3401
    },
    {
      "epoch": 2.252980132450331,
      "grad_norm": 0.8040104440661278,
      "learning_rate": 0.0002717436796537358,
      "loss": 2.1875,
      "step": 3402
    },
    {
      "epoch": 2.25364238410596,
      "grad_norm": 0.9207808506519045,
      "learning_rate": 0.0002717046749562928,
      "loss": 2.4844,
      "step": 3403
    },
    {
      "epoch": 2.2543046357615895,
      "grad_norm": 0.7615877458930103,
      "learning_rate": 0.00027166564615992877,
      "loss": 2.0312,
      "step": 3404
    },
    {
      "epoch": 2.2549668874172184,
      "grad_norm": 0.8909187387123997,
      "learning_rate": 0.00027162659327237195,
      "loss": 2.1562,
      "step": 3405
    },
    {
      "epoch": 2.255629139072848,
      "grad_norm": 0.8784966515385724,
      "learning_rate": 0.0002715875163013553,
      "loss": 2.4844,
      "step": 3406
    },
    {
      "epoch": 2.2562913907284767,
      "grad_norm": 0.798228395516519,
      "learning_rate": 0.0002715484152546164,
      "loss": 2.1719,
      "step": 3407
    },
    {
      "epoch": 2.256953642384106,
      "grad_norm": 0.8096575769378106,
      "learning_rate": 0.0002715092901398978,
      "loss": 2.1562,
      "step": 3408
    },
    {
      "epoch": 2.257615894039735,
      "grad_norm": 0.8263312669295252,
      "learning_rate": 0.00027147014096494675,
      "loss": 2.25,
      "step": 3409
    },
    {
      "epoch": 2.258278145695364,
      "grad_norm": 0.8032506928755974,
      "learning_rate": 0.0002714309677375151,
      "loss": 1.7578,
      "step": 3410
    },
    {
      "epoch": 2.2589403973509934,
      "grad_norm": 0.8118776506411731,
      "learning_rate": 0.00027139177046535976,
      "loss": 2.0781,
      "step": 3411
    },
    {
      "epoch": 2.2596026490066223,
      "grad_norm": 0.7489189597481541,
      "learning_rate": 0.0002713525491562421,
      "loss": 2.25,
      "step": 3412
    },
    {
      "epoch": 2.2602649006622517,
      "grad_norm": 0.8356459198460788,
      "learning_rate": 0.00027131330381792847,
      "loss": 2.125,
      "step": 3413
    },
    {
      "epoch": 2.2609271523178807,
      "grad_norm": 0.8126627671433717,
      "learning_rate": 0.00027127403445818983,
      "loss": 2.2656,
      "step": 3414
    },
    {
      "epoch": 2.26158940397351,
      "grad_norm": 0.8761888272040123,
      "learning_rate": 0.00027123474108480206,
      "loss": 2.1562,
      "step": 3415
    },
    {
      "epoch": 2.262251655629139,
      "grad_norm": 0.9409405461740266,
      "learning_rate": 0.0002711954237055456,
      "loss": 2.6406,
      "step": 3416
    },
    {
      "epoch": 2.2629139072847684,
      "grad_norm": 0.8472126965510226,
      "learning_rate": 0.0002711560823282058,
      "loss": 2.2031,
      "step": 3417
    },
    {
      "epoch": 2.2635761589403973,
      "grad_norm": 0.8420486264164412,
      "learning_rate": 0.00027111671696057274,
      "loss": 2.3125,
      "step": 3418
    },
    {
      "epoch": 2.2642384105960263,
      "grad_norm": 0.7985861220922338,
      "learning_rate": 0.0002710773276104411,
      "loss": 2.2344,
      "step": 3419
    },
    {
      "epoch": 2.2649006622516556,
      "grad_norm": 0.810427869583552,
      "learning_rate": 0.0002710379142856105,
      "loss": 2.4531,
      "step": 3420
    },
    {
      "epoch": 2.2655629139072846,
      "grad_norm": 0.8618888251122615,
      "learning_rate": 0.0002709984769938852,
      "loss": 2.4688,
      "step": 3421
    },
    {
      "epoch": 2.266225165562914,
      "grad_norm": 0.8198747872523499,
      "learning_rate": 0.0002709590157430743,
      "loss": 2.0781,
      "step": 3422
    },
    {
      "epoch": 2.266887417218543,
      "grad_norm": 0.778355310806136,
      "learning_rate": 0.0002709195305409915,
      "loss": 2.1719,
      "step": 3423
    },
    {
      "epoch": 2.2675496688741723,
      "grad_norm": 0.6993554432064756,
      "learning_rate": 0.0002708800213954554,
      "loss": 1.7344,
      "step": 3424
    },
    {
      "epoch": 2.2682119205298013,
      "grad_norm": 0.8238243243100243,
      "learning_rate": 0.0002708404883142891,
      "loss": 2.1875,
      "step": 3425
    },
    {
      "epoch": 2.2688741721854306,
      "grad_norm": 0.8666749818245352,
      "learning_rate": 0.00027080093130532086,
      "loss": 2.1406,
      "step": 3426
    },
    {
      "epoch": 2.2695364238410596,
      "grad_norm": 0.8421336644096673,
      "learning_rate": 0.00027076135037638324,
      "loss": 2.4062,
      "step": 3427
    },
    {
      "epoch": 2.2701986754966885,
      "grad_norm": 0.8559991768060545,
      "learning_rate": 0.00027072174553531374,
      "loss": 2.3438,
      "step": 3428
    },
    {
      "epoch": 2.270860927152318,
      "grad_norm": 0.8303548829817005,
      "learning_rate": 0.0002706821167899546,
      "loss": 2.25,
      "step": 3429
    },
    {
      "epoch": 2.271523178807947,
      "grad_norm": 0.8382802285375041,
      "learning_rate": 0.0002706424641481528,
      "loss": 2.4062,
      "step": 3430
    },
    {
      "epoch": 2.2721854304635762,
      "grad_norm": 0.8803112071613581,
      "learning_rate": 0.0002706027876177599,
      "loss": 2.0156,
      "step": 3431
    },
    {
      "epoch": 2.272847682119205,
      "grad_norm": 0.7720090301112768,
      "learning_rate": 0.0002705630872066324,
      "loss": 2.0469,
      "step": 3432
    },
    {
      "epoch": 2.2735099337748346,
      "grad_norm": 0.7653652268116603,
      "learning_rate": 0.0002705233629226315,
      "loss": 2.1406,
      "step": 3433
    },
    {
      "epoch": 2.2741721854304635,
      "grad_norm": 0.8053940854900532,
      "learning_rate": 0.0002704836147736229,
      "loss": 2.3594,
      "step": 3434
    },
    {
      "epoch": 2.274834437086093,
      "grad_norm": 0.7385875211216661,
      "learning_rate": 0.00027044384276747737,
      "loss": 2.0469,
      "step": 3435
    },
    {
      "epoch": 2.275496688741722,
      "grad_norm": 0.8804774699265401,
      "learning_rate": 0.00027040404691207006,
      "loss": 2.2656,
      "step": 3436
    },
    {
      "epoch": 2.276158940397351,
      "grad_norm": 0.8636738229663521,
      "learning_rate": 0.0002703642272152811,
      "loss": 2.1406,
      "step": 3437
    },
    {
      "epoch": 2.27682119205298,
      "grad_norm": 0.8235057521386625,
      "learning_rate": 0.0002703243836849952,
      "loss": 2.25,
      "step": 3438
    },
    {
      "epoch": 2.277483443708609,
      "grad_norm": 0.8204884388212093,
      "learning_rate": 0.00027028451632910185,
      "loss": 2.4062,
      "step": 3439
    },
    {
      "epoch": 2.2781456953642385,
      "grad_norm": 0.7989171557033451,
      "learning_rate": 0.00027024462515549524,
      "loss": 2.2031,
      "step": 3440
    },
    {
      "epoch": 2.2788079470198674,
      "grad_norm": 0.7493318433626527,
      "learning_rate": 0.0002702047101720744,
      "loss": 1.7266,
      "step": 3441
    },
    {
      "epoch": 2.279470198675497,
      "grad_norm": 0.7815136002279841,
      "learning_rate": 0.0002701647713867427,
      "loss": 1.9922,
      "step": 3442
    },
    {
      "epoch": 2.2801324503311258,
      "grad_norm": 0.8637766627426595,
      "learning_rate": 0.0002701248088074087,
      "loss": 2.4844,
      "step": 3443
    },
    {
      "epoch": 2.280794701986755,
      "grad_norm": 0.8447658776345155,
      "learning_rate": 0.00027008482244198536,
      "loss": 2.3438,
      "step": 3444
    },
    {
      "epoch": 2.281456953642384,
      "grad_norm": 0.7316717558792969,
      "learning_rate": 0.0002700448122983904,
      "loss": 1.8906,
      "step": 3445
    },
    {
      "epoch": 2.282119205298013,
      "grad_norm": 0.7467793634492488,
      "learning_rate": 0.0002700047783845464,
      "loss": 2.1719,
      "step": 3446
    },
    {
      "epoch": 2.2827814569536424,
      "grad_norm": 0.8118183386161748,
      "learning_rate": 0.00026996472070838043,
      "loss": 2.2656,
      "step": 3447
    },
    {
      "epoch": 2.2834437086092714,
      "grad_norm": 0.8660195484110779,
      "learning_rate": 0.0002699246392778244,
      "loss": 2.3438,
      "step": 3448
    },
    {
      "epoch": 2.2841059602649008,
      "grad_norm": 0.82448801922249,
      "learning_rate": 0.000269884534100815,
      "loss": 1.9297,
      "step": 3449
    },
    {
      "epoch": 2.2847682119205297,
      "grad_norm": 0.8362884112837178,
      "learning_rate": 0.0002698444051852933,
      "loss": 2.3438,
      "step": 3450
    },
    {
      "epoch": 2.285430463576159,
      "grad_norm": 0.8277670515587894,
      "learning_rate": 0.00026980425253920543,
      "loss": 2.3906,
      "step": 3451
    },
    {
      "epoch": 2.286092715231788,
      "grad_norm": 0.8090071707804996,
      "learning_rate": 0.00026976407617050206,
      "loss": 2.1562,
      "step": 3452
    },
    {
      "epoch": 2.2867549668874174,
      "grad_norm": 0.7835278686907772,
      "learning_rate": 0.0002697238760871386,
      "loss": 2.2812,
      "step": 3453
    },
    {
      "epoch": 2.2874172185430464,
      "grad_norm": 0.8842711401990846,
      "learning_rate": 0.00026968365229707497,
      "loss": 2.4531,
      "step": 3454
    },
    {
      "epoch": 2.2880794701986753,
      "grad_norm": 0.9212893370067547,
      "learning_rate": 0.00026964340480827617,
      "loss": 2.125,
      "step": 3455
    },
    {
      "epoch": 2.2887417218543047,
      "grad_norm": 0.8342558015598699,
      "learning_rate": 0.0002696031336287115,
      "loss": 2.4531,
      "step": 3456
    },
    {
      "epoch": 2.2894039735099336,
      "grad_norm": 0.7207004619667601,
      "learning_rate": 0.0002695628387663552,
      "loss": 2.0938,
      "step": 3457
    },
    {
      "epoch": 2.290066225165563,
      "grad_norm": 0.7898243661231695,
      "learning_rate": 0.000269522520229186,
      "loss": 1.9297,
      "step": 3458
    },
    {
      "epoch": 2.290728476821192,
      "grad_norm": 0.882192972919901,
      "learning_rate": 0.0002694821780251876,
      "loss": 1.9531,
      "step": 3459
    },
    {
      "epoch": 2.2913907284768213,
      "grad_norm": 0.7844110008067673,
      "learning_rate": 0.0002694418121623481,
      "loss": 2.1562,
      "step": 3460
    },
    {
      "epoch": 2.2920529801324503,
      "grad_norm": 0.8541480116558099,
      "learning_rate": 0.00026940142264866043,
      "loss": 2.1875,
      "step": 3461
    },
    {
      "epoch": 2.2927152317880797,
      "grad_norm": 0.8364263343369465,
      "learning_rate": 0.00026936100949212223,
      "loss": 2.1719,
      "step": 3462
    },
    {
      "epoch": 2.2933774834437086,
      "grad_norm": 0.8427578151844962,
      "learning_rate": 0.00026932057270073566,
      "loss": 2.0469,
      "step": 3463
    },
    {
      "epoch": 2.2940397350993376,
      "grad_norm": 0.8043233239213405,
      "learning_rate": 0.0002692801122825078,
      "loss": 2.1719,
      "step": 3464
    },
    {
      "epoch": 2.294701986754967,
      "grad_norm": 0.8113074335858285,
      "learning_rate": 0.00026923962824545013,
      "loss": 2.2656,
      "step": 3465
    },
    {
      "epoch": 2.295364238410596,
      "grad_norm": 0.8139702581361432,
      "learning_rate": 0.00026919912059757904,
      "loss": 2.4375,
      "step": 3466
    },
    {
      "epoch": 2.2960264900662253,
      "grad_norm": 0.7799906404235268,
      "learning_rate": 0.00026915858934691554,
      "loss": 2.2031,
      "step": 3467
    },
    {
      "epoch": 2.296688741721854,
      "grad_norm": 0.8405551450197039,
      "learning_rate": 0.0002691180345014852,
      "loss": 2.3281,
      "step": 3468
    },
    {
      "epoch": 2.2973509933774836,
      "grad_norm": 0.9412805990773327,
      "learning_rate": 0.00026907745606931843,
      "loss": 2.2812,
      "step": 3469
    },
    {
      "epoch": 2.2980132450331126,
      "grad_norm": 0.8074816861066985,
      "learning_rate": 0.0002690368540584501,
      "loss": 1.7891,
      "step": 3470
    },
    {
      "epoch": 2.298675496688742,
      "grad_norm": 0.7915758279067221,
      "learning_rate": 0.00026899622847692,
      "loss": 2.2188,
      "step": 3471
    },
    {
      "epoch": 2.299337748344371,
      "grad_norm": 0.7805398828352382,
      "learning_rate": 0.00026895557933277243,
      "loss": 2.2656,
      "step": 3472
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8029758477770688,
      "learning_rate": 0.0002689149066340564,
      "loss": 2.1406,
      "step": 3473
    },
    {
      "epoch": 2.300662251655629,
      "grad_norm": 0.7411551043751734,
      "learning_rate": 0.0002688742103888255,
      "loss": 1.9688,
      "step": 3474
    },
    {
      "epoch": 2.301324503311258,
      "grad_norm": 0.8215263787747553,
      "learning_rate": 0.0002688334906051381,
      "loss": 2.2812,
      "step": 3475
    },
    {
      "epoch": 2.3019867549668875,
      "grad_norm": 0.8063639458068304,
      "learning_rate": 0.00026879274729105717,
      "loss": 2.3906,
      "step": 3476
    },
    {
      "epoch": 2.3026490066225165,
      "grad_norm": 0.7333760557926674,
      "learning_rate": 0.00026875198045465037,
      "loss": 1.7109,
      "step": 3477
    },
    {
      "epoch": 2.303311258278146,
      "grad_norm": 0.810451754095096,
      "learning_rate": 0.00026871119010399004,
      "loss": 2.1719,
      "step": 3478
    },
    {
      "epoch": 2.303973509933775,
      "grad_norm": 0.8072731258411935,
      "learning_rate": 0.000268670376247153,
      "loss": 2.3125,
      "step": 3479
    },
    {
      "epoch": 2.304635761589404,
      "grad_norm": 0.7285133021589374,
      "learning_rate": 0.000268629538892221,
      "loss": 1.6797,
      "step": 3480
    },
    {
      "epoch": 2.305298013245033,
      "grad_norm": 0.788643311835183,
      "learning_rate": 0.00026858867804728027,
      "loss": 2.1406,
      "step": 3481
    },
    {
      "epoch": 2.305960264900662,
      "grad_norm": 0.7993309623498945,
      "learning_rate": 0.0002685477937204217,
      "loss": 2.1094,
      "step": 3482
    },
    {
      "epoch": 2.3066225165562915,
      "grad_norm": 0.7779715105032944,
      "learning_rate": 0.00026850688591974084,
      "loss": 2.0469,
      "step": 3483
    },
    {
      "epoch": 2.3072847682119204,
      "grad_norm": 0.8859747129026393,
      "learning_rate": 0.00026846595465333795,
      "loss": 2.4531,
      "step": 3484
    },
    {
      "epoch": 2.30794701986755,
      "grad_norm": 0.9794557863363869,
      "learning_rate": 0.00026842499992931786,
      "loss": 2.2969,
      "step": 3485
    },
    {
      "epoch": 2.3086092715231787,
      "grad_norm": 0.711960240393943,
      "learning_rate": 0.0002683840217557901,
      "loss": 1.9609,
      "step": 3486
    },
    {
      "epoch": 2.309271523178808,
      "grad_norm": 0.8364098197557683,
      "learning_rate": 0.0002683430201408688,
      "loss": 2.0625,
      "step": 3487
    },
    {
      "epoch": 2.309933774834437,
      "grad_norm": 0.7489954030200862,
      "learning_rate": 0.00026830199509267274,
      "loss": 2.0469,
      "step": 3488
    },
    {
      "epoch": 2.3105960264900665,
      "grad_norm": 0.8383299262573207,
      "learning_rate": 0.0002682609466193254,
      "loss": 2.2188,
      "step": 3489
    },
    {
      "epoch": 2.3112582781456954,
      "grad_norm": 0.8322237063990285,
      "learning_rate": 0.0002682198747289548,
      "loss": 2.1562,
      "step": 3490
    },
    {
      "epoch": 2.3119205298013243,
      "grad_norm": 0.808984873461122,
      "learning_rate": 0.00026817877942969363,
      "loss": 2.2969,
      "step": 3491
    },
    {
      "epoch": 2.3125827814569537,
      "grad_norm": 0.8564167958122387,
      "learning_rate": 0.00026813766072967925,
      "loss": 2.2344,
      "step": 3492
    },
    {
      "epoch": 2.3132450331125827,
      "grad_norm": 0.8271129423968329,
      "learning_rate": 0.0002680965186370537,
      "loss": 1.6953,
      "step": 3493
    },
    {
      "epoch": 2.313907284768212,
      "grad_norm": 0.7825199759283893,
      "learning_rate": 0.00026805535315996347,
      "loss": 2.1094,
      "step": 3494
    },
    {
      "epoch": 2.314569536423841,
      "grad_norm": 0.7558963310563874,
      "learning_rate": 0.0002680141643065599,
      "loss": 1.7812,
      "step": 3495
    },
    {
      "epoch": 2.3152317880794704,
      "grad_norm": 0.8409712057374107,
      "learning_rate": 0.0002679729520849988,
      "loss": 2.0938,
      "step": 3496
    },
    {
      "epoch": 2.3158940397350993,
      "grad_norm": 1.0284948757191428,
      "learning_rate": 0.0002679317165034406,
      "loss": 2.3438,
      "step": 3497
    },
    {
      "epoch": 2.3165562913907287,
      "grad_norm": 0.823717652873751,
      "learning_rate": 0.00026789045757005055,
      "loss": 2.2812,
      "step": 3498
    },
    {
      "epoch": 2.3172185430463577,
      "grad_norm": 0.7742367679207242,
      "learning_rate": 0.00026784917529299834,
      "loss": 2.0469,
      "step": 3499
    },
    {
      "epoch": 2.3178807947019866,
      "grad_norm": 0.8300108887562018,
      "learning_rate": 0.0002678078696804583,
      "loss": 2.1406,
      "step": 3500
    },
    {
      "epoch": 2.318543046357616,
      "grad_norm": 0.8019883200688397,
      "learning_rate": 0.00026776654074060947,
      "loss": 2.0,
      "step": 3501
    },
    {
      "epoch": 2.319205298013245,
      "grad_norm": 0.7767040976168101,
      "learning_rate": 0.0002677251884816354,
      "loss": 2.1406,
      "step": 3502
    },
    {
      "epoch": 2.3198675496688743,
      "grad_norm": 0.7871385438991763,
      "learning_rate": 0.0002676838129117244,
      "loss": 2.125,
      "step": 3503
    },
    {
      "epoch": 2.3205298013245033,
      "grad_norm": 0.8101185525919736,
      "learning_rate": 0.0002676424140390691,
      "loss": 2.2031,
      "step": 3504
    },
    {
      "epoch": 2.321192052980132,
      "grad_norm": 0.8173896386150242,
      "learning_rate": 0.00026760099187186725,
      "loss": 2.2812,
      "step": 3505
    },
    {
      "epoch": 2.3218543046357616,
      "grad_norm": 0.7869629040036665,
      "learning_rate": 0.00026755954641832064,
      "loss": 1.8281,
      "step": 3506
    },
    {
      "epoch": 2.322516556291391,
      "grad_norm": 0.7880492843609324,
      "learning_rate": 0.00026751807768663616,
      "loss": 2.0469,
      "step": 3507
    },
    {
      "epoch": 2.32317880794702,
      "grad_norm": 0.8280483733652815,
      "learning_rate": 0.0002674765856850249,
      "loss": 2.2812,
      "step": 3508
    },
    {
      "epoch": 2.323841059602649,
      "grad_norm": 0.7558704103339472,
      "learning_rate": 0.0002674350704217029,
      "loss": 1.9922,
      "step": 3509
    },
    {
      "epoch": 2.3245033112582782,
      "grad_norm": 0.7772141431684703,
      "learning_rate": 0.00026739353190489057,
      "loss": 2.1562,
      "step": 3510
    },
    {
      "epoch": 2.325165562913907,
      "grad_norm": 0.754192091853306,
      "learning_rate": 0.0002673519701428131,
      "loss": 1.9609,
      "step": 3511
    },
    {
      "epoch": 2.3258278145695366,
      "grad_norm": 0.7710161717638109,
      "learning_rate": 0.0002673103851437001,
      "loss": 2.1562,
      "step": 3512
    },
    {
      "epoch": 2.3264900662251655,
      "grad_norm": 0.8134440167567158,
      "learning_rate": 0.000267268776915786,
      "loss": 1.9609,
      "step": 3513
    },
    {
      "epoch": 2.3271523178807945,
      "grad_norm": 0.9498468649480726,
      "learning_rate": 0.00026722714546730957,
      "loss": 2.3438,
      "step": 3514
    },
    {
      "epoch": 2.327814569536424,
      "grad_norm": 0.753497630990035,
      "learning_rate": 0.0002671854908065144,
      "loss": 1.8828,
      "step": 3515
    },
    {
      "epoch": 2.328476821192053,
      "grad_norm": 0.7475686125932557,
      "learning_rate": 0.00026714381294164853,
      "loss": 1.7578,
      "step": 3516
    },
    {
      "epoch": 2.329139072847682,
      "grad_norm": 0.7368471150432514,
      "learning_rate": 0.0002671021118809647,
      "loss": 1.7969,
      "step": 3517
    },
    {
      "epoch": 2.329801324503311,
      "grad_norm": 0.7627186588359872,
      "learning_rate": 0.0002670603876327202,
      "loss": 2.1094,
      "step": 3518
    },
    {
      "epoch": 2.3304635761589405,
      "grad_norm": 0.8241381966534143,
      "learning_rate": 0.00026701864020517685,
      "loss": 2.2656,
      "step": 3519
    },
    {
      "epoch": 2.3311258278145695,
      "grad_norm": 0.8102145676316767,
      "learning_rate": 0.00026697686960660123,
      "loss": 1.9922,
      "step": 3520
    },
    {
      "epoch": 2.331788079470199,
      "grad_norm": 0.8620441010616755,
      "learning_rate": 0.00026693507584526425,
      "loss": 2.3438,
      "step": 3521
    },
    {
      "epoch": 2.332450331125828,
      "grad_norm": 0.8842616765315734,
      "learning_rate": 0.0002668932589294417,
      "loss": 2.2969,
      "step": 3522
    },
    {
      "epoch": 2.3331125827814567,
      "grad_norm": 0.7528527379787083,
      "learning_rate": 0.0002668514188674137,
      "loss": 1.8359,
      "step": 3523
    },
    {
      "epoch": 2.333774834437086,
      "grad_norm": 0.8130074819141339,
      "learning_rate": 0.0002668095556674652,
      "loss": 2.0312,
      "step": 3524
    },
    {
      "epoch": 2.334437086092715,
      "grad_norm": 0.8015450699548855,
      "learning_rate": 0.0002667676693378854,
      "loss": 2.1875,
      "step": 3525
    },
    {
      "epoch": 2.3350993377483444,
      "grad_norm": 0.8550538713528306,
      "learning_rate": 0.00026672575988696844,
      "loss": 2.25,
      "step": 3526
    },
    {
      "epoch": 2.3357615894039734,
      "grad_norm": 0.7655186600584316,
      "learning_rate": 0.00026668382732301276,
      "loss": 2.0,
      "step": 3527
    },
    {
      "epoch": 2.3364238410596028,
      "grad_norm": 0.8703591867480458,
      "learning_rate": 0.0002666418716543215,
      "loss": 2.1719,
      "step": 3528
    },
    {
      "epoch": 2.3370860927152317,
      "grad_norm": 0.8716923787579354,
      "learning_rate": 0.0002665998928892025,
      "loss": 2.2031,
      "step": 3529
    },
    {
      "epoch": 2.337748344370861,
      "grad_norm": 0.7623059992171638,
      "learning_rate": 0.00026655789103596783,
      "loss": 1.8203,
      "step": 3530
    },
    {
      "epoch": 2.33841059602649,
      "grad_norm": 0.83047209439055,
      "learning_rate": 0.00026651586610293453,
      "loss": 2.25,
      "step": 3531
    },
    {
      "epoch": 2.339072847682119,
      "grad_norm": 0.7910681377650448,
      "learning_rate": 0.0002664738180984239,
      "loss": 2.0312,
      "step": 3532
    },
    {
      "epoch": 2.3397350993377484,
      "grad_norm": 0.8100789357020621,
      "learning_rate": 0.0002664317470307619,
      "loss": 2.0938,
      "step": 3533
    },
    {
      "epoch": 2.3403973509933773,
      "grad_norm": 0.9017538204288273,
      "learning_rate": 0.0002663896529082792,
      "loss": 2.1406,
      "step": 3534
    },
    {
      "epoch": 2.3410596026490067,
      "grad_norm": 0.7412165669742986,
      "learning_rate": 0.00026634753573931083,
      "loss": 2.0469,
      "step": 3535
    },
    {
      "epoch": 2.3417218543046356,
      "grad_norm": 0.8849553799040792,
      "learning_rate": 0.00026630539553219656,
      "loss": 2.2188,
      "step": 3536
    },
    {
      "epoch": 2.342384105960265,
      "grad_norm": 0.7832643093217307,
      "learning_rate": 0.00026626323229528054,
      "loss": 1.9609,
      "step": 3537
    },
    {
      "epoch": 2.343046357615894,
      "grad_norm": 0.8241418829860712,
      "learning_rate": 0.0002662210460369116,
      "loss": 2.2969,
      "step": 3538
    },
    {
      "epoch": 2.3437086092715234,
      "grad_norm": 0.8246287905762533,
      "learning_rate": 0.0002661788367654431,
      "loss": 2.1094,
      "step": 3539
    },
    {
      "epoch": 2.3443708609271523,
      "grad_norm": 0.8955625126478135,
      "learning_rate": 0.000266136604489233,
      "loss": 2.1719,
      "step": 3540
    },
    {
      "epoch": 2.3450331125827812,
      "grad_norm": 0.9896257990148383,
      "learning_rate": 0.00026609434921664375,
      "loss": 2.2812,
      "step": 3541
    },
    {
      "epoch": 2.3456953642384106,
      "grad_norm": 0.8209655093030814,
      "learning_rate": 0.0002660520709560424,
      "loss": 2.1875,
      "step": 3542
    },
    {
      "epoch": 2.3463576158940396,
      "grad_norm": 0.8153954341921122,
      "learning_rate": 0.0002660097697158005,
      "loss": 2.1719,
      "step": 3543
    },
    {
      "epoch": 2.347019867549669,
      "grad_norm": 0.7872034619718571,
      "learning_rate": 0.00026596744550429414,
      "loss": 2.1406,
      "step": 3544
    },
    {
      "epoch": 2.347682119205298,
      "grad_norm": 0.8290463355424399,
      "learning_rate": 0.00026592509832990403,
      "loss": 2.0781,
      "step": 3545
    },
    {
      "epoch": 2.3483443708609273,
      "grad_norm": 0.722934519215767,
      "learning_rate": 0.0002658827282010155,
      "loss": 2.0625,
      "step": 3546
    },
    {
      "epoch": 2.3490066225165562,
      "grad_norm": 0.8174556561968537,
      "learning_rate": 0.00026584033512601823,
      "loss": 2.0469,
      "step": 3547
    },
    {
      "epoch": 2.3496688741721856,
      "grad_norm": 0.7975791625499015,
      "learning_rate": 0.0002657979191133065,
      "loss": 1.8984,
      "step": 3548
    },
    {
      "epoch": 2.3503311258278146,
      "grad_norm": 0.9192925546141507,
      "learning_rate": 0.0002657554801712792,
      "loss": 2.3906,
      "step": 3549
    },
    {
      "epoch": 2.3509933774834435,
      "grad_norm": 0.7872758239517652,
      "learning_rate": 0.0002657130183083397,
      "loss": 2.2188,
      "step": 3550
    },
    {
      "epoch": 2.351655629139073,
      "grad_norm": 0.7702925193856948,
      "learning_rate": 0.000265670533532896,
      "loss": 2.2344,
      "step": 3551
    },
    {
      "epoch": 2.352317880794702,
      "grad_norm": 0.7936957568718466,
      "learning_rate": 0.00026562802585336053,
      "loss": 2.25,
      "step": 3552
    },
    {
      "epoch": 2.352980132450331,
      "grad_norm": 0.783762734306259,
      "learning_rate": 0.00026558549527815035,
      "loss": 2.1094,
      "step": 3553
    },
    {
      "epoch": 2.35364238410596,
      "grad_norm": 0.8405530902811719,
      "learning_rate": 0.000265542941815687,
      "loss": 1.7344,
      "step": 3554
    },
    {
      "epoch": 2.3543046357615895,
      "grad_norm": 0.7364878035352765,
      "learning_rate": 0.0002655003654743964,
      "loss": 2.0312,
      "step": 3555
    },
    {
      "epoch": 2.3549668874172185,
      "grad_norm": 0.8523626169481099,
      "learning_rate": 0.0002654577662627093,
      "loss": 2.2188,
      "step": 3556
    },
    {
      "epoch": 2.355629139072848,
      "grad_norm": 0.770420021299208,
      "learning_rate": 0.00026541514418906083,
      "loss": 2.0781,
      "step": 3557
    },
    {
      "epoch": 2.356291390728477,
      "grad_norm": 0.8365722566100738,
      "learning_rate": 0.0002653724992618906,
      "loss": 2.0781,
      "step": 3558
    },
    {
      "epoch": 2.3569536423841058,
      "grad_norm": 0.8492206160524566,
      "learning_rate": 0.00026532983148964284,
      "loss": 2.2031,
      "step": 3559
    },
    {
      "epoch": 2.357615894039735,
      "grad_norm": 0.8912623066850461,
      "learning_rate": 0.0002652871408807662,
      "loss": 2.0625,
      "step": 3560
    },
    {
      "epoch": 2.358278145695364,
      "grad_norm": 0.7814844799594335,
      "learning_rate": 0.000265244427443714,
      "loss": 1.9453,
      "step": 3561
    },
    {
      "epoch": 2.3589403973509935,
      "grad_norm": 0.8588486143474591,
      "learning_rate": 0.00026520169118694385,
      "loss": 2.2344,
      "step": 3562
    },
    {
      "epoch": 2.3596026490066224,
      "grad_norm": 0.7846538466182011,
      "learning_rate": 0.0002651589321189181,
      "loss": 1.9219,
      "step": 3563
    },
    {
      "epoch": 2.360264900662252,
      "grad_norm": 0.816868818092568,
      "learning_rate": 0.00026511615024810365,
      "loss": 1.9219,
      "step": 3564
    },
    {
      "epoch": 2.3609271523178808,
      "grad_norm": 0.9338972370468528,
      "learning_rate": 0.0002650733455829716,
      "loss": 2.2656,
      "step": 3565
    },
    {
      "epoch": 2.36158940397351,
      "grad_norm": 0.8599554968607911,
      "learning_rate": 0.00026503051813199784,
      "loss": 2.2812,
      "step": 3566
    },
    {
      "epoch": 2.362251655629139,
      "grad_norm": 0.8571277365505403,
      "learning_rate": 0.0002649876679036627,
      "loss": 2.25,
      "step": 3567
    },
    {
      "epoch": 2.362913907284768,
      "grad_norm": 0.787730185536069,
      "learning_rate": 0.0002649447949064511,
      "loss": 2.1719,
      "step": 3568
    },
    {
      "epoch": 2.3635761589403974,
      "grad_norm": 0.7541041324615795,
      "learning_rate": 0.0002649018991488523,
      "loss": 1.7422,
      "step": 3569
    },
    {
      "epoch": 2.3642384105960264,
      "grad_norm": 0.9222148801036135,
      "learning_rate": 0.0002648589806393601,
      "loss": 2.3125,
      "step": 3570
    },
    {
      "epoch": 2.3649006622516557,
      "grad_norm": 0.7649390027216305,
      "learning_rate": 0.0002648160393864729,
      "loss": 1.9922,
      "step": 3571
    },
    {
      "epoch": 2.3655629139072847,
      "grad_norm": 0.9568446131772311,
      "learning_rate": 0.00026477307539869363,
      "loss": 2.2031,
      "step": 3572
    },
    {
      "epoch": 2.366225165562914,
      "grad_norm": 0.7702773392732393,
      "learning_rate": 0.00026473008868452963,
      "loss": 2.125,
      "step": 3573
    },
    {
      "epoch": 2.366887417218543,
      "grad_norm": 0.7873764885522335,
      "learning_rate": 0.0002646870792524927,
      "loss": 2.0156,
      "step": 3574
    },
    {
      "epoch": 2.3675496688741724,
      "grad_norm": 0.9573384268443924,
      "learning_rate": 0.00026464404711109917,
      "loss": 2.4688,
      "step": 3575
    },
    {
      "epoch": 2.3682119205298013,
      "grad_norm": 0.7859632885560607,
      "learning_rate": 0.00026460099226887,
      "loss": 2.125,
      "step": 3576
    },
    {
      "epoch": 2.3688741721854303,
      "grad_norm": 0.7656407422301027,
      "learning_rate": 0.00026455791473433055,
      "loss": 2.1094,
      "step": 3577
    },
    {
      "epoch": 2.3695364238410597,
      "grad_norm": 0.8581622538286189,
      "learning_rate": 0.0002645148145160106,
      "loss": 2.3125,
      "step": 3578
    },
    {
      "epoch": 2.3701986754966886,
      "grad_norm": 0.8412551384807456,
      "learning_rate": 0.0002644716916224445,
      "loss": 2.1719,
      "step": 3579
    },
    {
      "epoch": 2.370860927152318,
      "grad_norm": 0.7527793173231068,
      "learning_rate": 0.0002644285460621711,
      "loss": 2.1562,
      "step": 3580
    },
    {
      "epoch": 2.371523178807947,
      "grad_norm": 0.7875784960554587,
      "learning_rate": 0.0002643853778437337,
      "loss": 2.2344,
      "step": 3581
    },
    {
      "epoch": 2.3721854304635763,
      "grad_norm": 0.7972729524228418,
      "learning_rate": 0.00026434218697568004,
      "loss": 2.3281,
      "step": 3582
    },
    {
      "epoch": 2.3728476821192053,
      "grad_norm": 0.7538283326101876,
      "learning_rate": 0.0002642989734665625,
      "loss": 2.1875,
      "step": 3583
    },
    {
      "epoch": 2.3735099337748347,
      "grad_norm": 0.7308391322920831,
      "learning_rate": 0.00026425573732493784,
      "loss": 2.0781,
      "step": 3584
    },
    {
      "epoch": 2.3741721854304636,
      "grad_norm": 0.7821157463649815,
      "learning_rate": 0.0002642124785593673,
      "loss": 2.0156,
      "step": 3585
    },
    {
      "epoch": 2.3748344370860925,
      "grad_norm": 0.7619751460233185,
      "learning_rate": 0.0002641691971784166,
      "loss": 1.9297,
      "step": 3586
    },
    {
      "epoch": 2.375496688741722,
      "grad_norm": 0.7937685605739325,
      "learning_rate": 0.0002641258931906559,
      "loss": 2.2344,
      "step": 3587
    },
    {
      "epoch": 2.376158940397351,
      "grad_norm": 0.8312520885940347,
      "learning_rate": 0.00026408256660466,
      "loss": 2.2969,
      "step": 3588
    },
    {
      "epoch": 2.3768211920529803,
      "grad_norm": 0.7833330596737437,
      "learning_rate": 0.00026403921742900794,
      "loss": 2.0,
      "step": 3589
    },
    {
      "epoch": 2.377483443708609,
      "grad_norm": 0.7929016774121538,
      "learning_rate": 0.00026399584567228344,
      "loss": 2.3281,
      "step": 3590
    },
    {
      "epoch": 2.3781456953642386,
      "grad_norm": 0.7753201205484034,
      "learning_rate": 0.0002639524513430746,
      "loss": 2.0312,
      "step": 3591
    },
    {
      "epoch": 2.3788079470198675,
      "grad_norm": 0.8070256858649795,
      "learning_rate": 0.00026390903444997396,
      "loss": 2.0938,
      "step": 3592
    },
    {
      "epoch": 2.379470198675497,
      "grad_norm": 0.8784035595165001,
      "learning_rate": 0.00026386559500157856,
      "loss": 2.2344,
      "step": 3593
    },
    {
      "epoch": 2.380132450331126,
      "grad_norm": 0.7934989968736885,
      "learning_rate": 0.0002638221330064899,
      "loss": 2.25,
      "step": 3594
    },
    {
      "epoch": 2.380794701986755,
      "grad_norm": 0.7999236763820265,
      "learning_rate": 0.00026377864847331405,
      "loss": 2.1719,
      "step": 3595
    },
    {
      "epoch": 2.381456953642384,
      "grad_norm": 0.810801944709112,
      "learning_rate": 0.0002637351414106613,
      "loss": 2.1406,
      "step": 3596
    },
    {
      "epoch": 2.382119205298013,
      "grad_norm": 0.8681317904333736,
      "learning_rate": 0.00026369161182714665,
      "loss": 2.2031,
      "step": 3597
    },
    {
      "epoch": 2.3827814569536425,
      "grad_norm": 0.8895130495198013,
      "learning_rate": 0.00026364805973138944,
      "loss": 2.2656,
      "step": 3598
    },
    {
      "epoch": 2.3834437086092715,
      "grad_norm": 0.7717012487635636,
      "learning_rate": 0.00026360448513201347,
      "loss": 2.1094,
      "step": 3599
    },
    {
      "epoch": 2.384105960264901,
      "grad_norm": 0.8782813227523034,
      "learning_rate": 0.000263560888037647,
      "loss": 2.5312,
      "step": 3600
    },
    {
      "epoch": 2.38476821192053,
      "grad_norm": 0.8088371036538807,
      "learning_rate": 0.0002635172684569227,
      "loss": 2.0781,
      "step": 3601
    },
    {
      "epoch": 2.385430463576159,
      "grad_norm": 0.7506189918319235,
      "learning_rate": 0.0002634736263984779,
      "loss": 2.0625,
      "step": 3602
    },
    {
      "epoch": 2.386092715231788,
      "grad_norm": 0.7448361186872774,
      "learning_rate": 0.0002634299618709541,
      "loss": 2.0156,
      "step": 3603
    },
    {
      "epoch": 2.386754966887417,
      "grad_norm": 0.7429560549759459,
      "learning_rate": 0.0002633862748829975,
      "loss": 1.9766,
      "step": 3604
    },
    {
      "epoch": 2.3874172185430464,
      "grad_norm": 0.7293237850457983,
      "learning_rate": 0.0002633425654432585,
      "loss": 2.0938,
      "step": 3605
    },
    {
      "epoch": 2.3880794701986754,
      "grad_norm": 0.8067632906532726,
      "learning_rate": 0.000263298833560392,
      "loss": 2.2969,
      "step": 3606
    },
    {
      "epoch": 2.388741721854305,
      "grad_norm": 0.8669901850886488,
      "learning_rate": 0.00026325507924305764,
      "loss": 2.1875,
      "step": 3607
    },
    {
      "epoch": 2.3894039735099337,
      "grad_norm": 0.7814395338743131,
      "learning_rate": 0.00026321130249991917,
      "loss": 2.2656,
      "step": 3608
    },
    {
      "epoch": 2.390066225165563,
      "grad_norm": 0.8058662060806352,
      "learning_rate": 0.0002631675033396448,
      "loss": 2.25,
      "step": 3609
    },
    {
      "epoch": 2.390728476821192,
      "grad_norm": 0.8135754639778354,
      "learning_rate": 0.0002631236817709074,
      "loss": 2.0938,
      "step": 3610
    },
    {
      "epoch": 2.3913907284768214,
      "grad_norm": 0.8231872854923911,
      "learning_rate": 0.00026307983780238405,
      "loss": 2.375,
      "step": 3611
    },
    {
      "epoch": 2.3920529801324504,
      "grad_norm": 0.7798938095782424,
      "learning_rate": 0.00026303597144275643,
      "loss": 2.0469,
      "step": 3612
    },
    {
      "epoch": 2.3927152317880793,
      "grad_norm": 0.769214196087789,
      "learning_rate": 0.00026299208270071057,
      "loss": 2.1406,
      "step": 3613
    },
    {
      "epoch": 2.3933774834437087,
      "grad_norm": 0.8931987315375328,
      "learning_rate": 0.0002629481715849369,
      "loss": 2.2031,
      "step": 3614
    },
    {
      "epoch": 2.3940397350993377,
      "grad_norm": 0.8324613832885025,
      "learning_rate": 0.00026290423810413033,
      "loss": 2.375,
      "step": 3615
    },
    {
      "epoch": 2.394701986754967,
      "grad_norm": 0.7790733031306819,
      "learning_rate": 0.00026286028226699023,
      "loss": 2.0156,
      "step": 3616
    },
    {
      "epoch": 2.395364238410596,
      "grad_norm": 0.8176880623287325,
      "learning_rate": 0.0002628163040822203,
      "loss": 2.375,
      "step": 3617
    },
    {
      "epoch": 2.396026490066225,
      "grad_norm": 0.842797782019801,
      "learning_rate": 0.00026277230355852883,
      "loss": 2.3125,
      "step": 3618
    },
    {
      "epoch": 2.3966887417218543,
      "grad_norm": 0.8096954669219841,
      "learning_rate": 0.0002627282807046283,
      "loss": 2.2969,
      "step": 3619
    },
    {
      "epoch": 2.3973509933774833,
      "grad_norm": 0.7591993295150838,
      "learning_rate": 0.0002626842355292359,
      "loss": 2.0781,
      "step": 3620
    },
    {
      "epoch": 2.3980132450331126,
      "grad_norm": 0.7455173718863316,
      "learning_rate": 0.0002626401680410729,
      "loss": 2.1875,
      "step": 3621
    },
    {
      "epoch": 2.3986754966887416,
      "grad_norm": 0.7332975584379399,
      "learning_rate": 0.00026259607824886526,
      "loss": 1.8047,
      "step": 3622
    },
    {
      "epoch": 2.399337748344371,
      "grad_norm": 0.8238404928995606,
      "learning_rate": 0.0002625519661613433,
      "loss": 2.2031,
      "step": 3623
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.7633460862918544,
      "learning_rate": 0.00026250783178724166,
      "loss": 2.2812,
      "step": 3624
    },
    {
      "epoch": 2.4006622516556293,
      "grad_norm": 0.7227952906113905,
      "learning_rate": 0.0002624636751352994,
      "loss": 2.0781,
      "step": 3625
    },
    {
      "epoch": 2.4013245033112582,
      "grad_norm": 0.8055733301251006,
      "learning_rate": 0.0002624194962142602,
      "loss": 2.125,
      "step": 3626
    },
    {
      "epoch": 2.401986754966887,
      "grad_norm": 0.7939315294182231,
      "learning_rate": 0.0002623752950328718,
      "loss": 2.3906,
      "step": 3627
    },
    {
      "epoch": 2.4026490066225166,
      "grad_norm": 0.757823821228571,
      "learning_rate": 0.0002623310715998867,
      "loss": 2.125,
      "step": 3628
    },
    {
      "epoch": 2.4033112582781455,
      "grad_norm": 0.8216954008180362,
      "learning_rate": 0.0002622868259240616,
      "loss": 2.0781,
      "step": 3629
    },
    {
      "epoch": 2.403973509933775,
      "grad_norm": 0.7742602116984382,
      "learning_rate": 0.00026224255801415765,
      "loss": 2.125,
      "step": 3630
    },
    {
      "epoch": 2.404635761589404,
      "grad_norm": 0.8687000798179293,
      "learning_rate": 0.0002621982678789404,
      "loss": 2.5938,
      "step": 3631
    },
    {
      "epoch": 2.4052980132450332,
      "grad_norm": 0.7554301587009465,
      "learning_rate": 0.00026215395552717973,
      "loss": 2.0781,
      "step": 3632
    },
    {
      "epoch": 2.405960264900662,
      "grad_norm": 0.7480187277505671,
      "learning_rate": 0.00026210962096765014,
      "loss": 2.25,
      "step": 3633
    },
    {
      "epoch": 2.4066225165562916,
      "grad_norm": 0.8014807323742578,
      "learning_rate": 0.00026206526420913023,
      "loss": 2.2969,
      "step": 3634
    },
    {
      "epoch": 2.4072847682119205,
      "grad_norm": 0.7786140637073341,
      "learning_rate": 0.0002620208852604033,
      "loss": 2.2344,
      "step": 3635
    },
    {
      "epoch": 2.4079470198675494,
      "grad_norm": 0.8985461561247333,
      "learning_rate": 0.00026197648413025674,
      "loss": 2.1094,
      "step": 3636
    },
    {
      "epoch": 2.408609271523179,
      "grad_norm": 0.7788293376187034,
      "learning_rate": 0.0002619320608274826,
      "loss": 2.0,
      "step": 3637
    },
    {
      "epoch": 2.4092715231788078,
      "grad_norm": 0.8471027246759101,
      "learning_rate": 0.0002618876153608771,
      "loss": 2.3594,
      "step": 3638
    },
    {
      "epoch": 2.409933774834437,
      "grad_norm": 0.7894665103156032,
      "learning_rate": 0.00026184314773924103,
      "loss": 2.0938,
      "step": 3639
    },
    {
      "epoch": 2.410596026490066,
      "grad_norm": 0.712404186045945,
      "learning_rate": 0.0002617986579713795,
      "loss": 1.7031,
      "step": 3640
    },
    {
      "epoch": 2.4112582781456955,
      "grad_norm": 0.7956530994207354,
      "learning_rate": 0.000261754146066102,
      "loss": 2.2656,
      "step": 3641
    },
    {
      "epoch": 2.4119205298013244,
      "grad_norm": 0.808948601121387,
      "learning_rate": 0.0002617096120322223,
      "loss": 2.1562,
      "step": 3642
    },
    {
      "epoch": 2.412582781456954,
      "grad_norm": 0.8156357415889068,
      "learning_rate": 0.0002616650558785587,
      "loss": 1.9219,
      "step": 3643
    },
    {
      "epoch": 2.4132450331125828,
      "grad_norm": 0.9026219565846519,
      "learning_rate": 0.00026162047761393387,
      "loss": 2.3906,
      "step": 3644
    },
    {
      "epoch": 2.4139072847682117,
      "grad_norm": 0.828435762212483,
      "learning_rate": 0.00026157587724717475,
      "loss": 2.0938,
      "step": 3645
    },
    {
      "epoch": 2.414569536423841,
      "grad_norm": 0.8252140024733946,
      "learning_rate": 0.00026153125478711275,
      "loss": 2.1875,
      "step": 3646
    },
    {
      "epoch": 2.41523178807947,
      "grad_norm": 0.8520896357360669,
      "learning_rate": 0.00026148661024258374,
      "loss": 2.4688,
      "step": 3647
    },
    {
      "epoch": 2.4158940397350994,
      "grad_norm": 0.8412381300223911,
      "learning_rate": 0.00026144194362242765,
      "loss": 2.2656,
      "step": 3648
    },
    {
      "epoch": 2.4165562913907284,
      "grad_norm": 0.7537007870515904,
      "learning_rate": 0.0002613972549354891,
      "loss": 2.0312,
      "step": 3649
    },
    {
      "epoch": 2.4172185430463577,
      "grad_norm": 0.727992445937327,
      "learning_rate": 0.00026135254419061695,
      "loss": 1.7969,
      "step": 3650
    },
    {
      "epoch": 2.4178807947019867,
      "grad_norm": 0.8413015170724273,
      "learning_rate": 0.00026130781139666444,
      "loss": 2.3906,
      "step": 3651
    },
    {
      "epoch": 2.418543046357616,
      "grad_norm": 0.7684723693713971,
      "learning_rate": 0.00026126305656248914,
      "loss": 2.0312,
      "step": 3652
    },
    {
      "epoch": 2.419205298013245,
      "grad_norm": 0.7508854474142914,
      "learning_rate": 0.0002612182796969531,
      "loss": 1.8125,
      "step": 3653
    },
    {
      "epoch": 2.419867549668874,
      "grad_norm": 0.732397409630897,
      "learning_rate": 0.00026117348080892264,
      "loss": 1.75,
      "step": 3654
    },
    {
      "epoch": 2.4205298013245033,
      "grad_norm": 0.8742543732150815,
      "learning_rate": 0.0002611286599072683,
      "loss": 2.4375,
      "step": 3655
    },
    {
      "epoch": 2.4211920529801323,
      "grad_norm": 0.7538414469191606,
      "learning_rate": 0.00026108381700086535,
      "loss": 2.1094,
      "step": 3656
    },
    {
      "epoch": 2.4218543046357617,
      "grad_norm": 0.7484323067327109,
      "learning_rate": 0.00026103895209859306,
      "loss": 2.0,
      "step": 3657
    },
    {
      "epoch": 2.4225165562913906,
      "grad_norm": 0.8465057680432724,
      "learning_rate": 0.00026099406520933523,
      "loss": 2.1875,
      "step": 3658
    },
    {
      "epoch": 2.42317880794702,
      "grad_norm": 1.0162209057975777,
      "learning_rate": 0.00026094915634198,
      "loss": 2.3594,
      "step": 3659
    },
    {
      "epoch": 2.423841059602649,
      "grad_norm": 0.7283380008014367,
      "learning_rate": 0.0002609042255054198,
      "loss": 1.9766,
      "step": 3660
    },
    {
      "epoch": 2.4245033112582783,
      "grad_norm": 0.7888572002465285,
      "learning_rate": 0.0002608592727085515,
      "loss": 2.1719,
      "step": 3661
    },
    {
      "epoch": 2.4251655629139073,
      "grad_norm": 0.8199963998615016,
      "learning_rate": 0.00026081429796027624,
      "loss": 2.0156,
      "step": 3662
    },
    {
      "epoch": 2.4258278145695362,
      "grad_norm": 0.8334344732561343,
      "learning_rate": 0.0002607693012694995,
      "loss": 2.1875,
      "step": 3663
    },
    {
      "epoch": 2.4264900662251656,
      "grad_norm": 0.8386123031054049,
      "learning_rate": 0.00026072428264513125,
      "loss": 2.1875,
      "step": 3664
    },
    {
      "epoch": 2.4271523178807946,
      "grad_norm": 0.8204409577548738,
      "learning_rate": 0.00026067924209608557,
      "loss": 2.2656,
      "step": 3665
    },
    {
      "epoch": 2.427814569536424,
      "grad_norm": 0.8790249933305871,
      "learning_rate": 0.0002606341796312811,
      "loss": 2.25,
      "step": 3666
    },
    {
      "epoch": 2.428476821192053,
      "grad_norm": 0.8651708312663978,
      "learning_rate": 0.0002605890952596406,
      "loss": 2.0625,
      "step": 3667
    },
    {
      "epoch": 2.4291390728476823,
      "grad_norm": 0.8130654986601692,
      "learning_rate": 0.00026054398899009144,
      "loss": 2.25,
      "step": 3668
    },
    {
      "epoch": 2.429801324503311,
      "grad_norm": 0.8081244852505798,
      "learning_rate": 0.00026049886083156517,
      "loss": 2.2812,
      "step": 3669
    },
    {
      "epoch": 2.4304635761589406,
      "grad_norm": 0.7721411828936044,
      "learning_rate": 0.00026045371079299755,
      "loss": 2.0469,
      "step": 3670
    },
    {
      "epoch": 2.4311258278145695,
      "grad_norm": 0.8499352959739105,
      "learning_rate": 0.00026040853888332886,
      "loss": 2.2969,
      "step": 3671
    },
    {
      "epoch": 2.4317880794701985,
      "grad_norm": 0.8580259171222804,
      "learning_rate": 0.0002603633451115037,
      "loss": 1.9453,
      "step": 3672
    },
    {
      "epoch": 2.432450331125828,
      "grad_norm": 0.7928081475423464,
      "learning_rate": 0.00026031812948647097,
      "loss": 2.1875,
      "step": 3673
    },
    {
      "epoch": 2.433112582781457,
      "grad_norm": 0.7808383073504149,
      "learning_rate": 0.00026027289201718384,
      "loss": 1.8906,
      "step": 3674
    },
    {
      "epoch": 2.433774834437086,
      "grad_norm": 0.8641230958121657,
      "learning_rate": 0.00026022763271259984,
      "loss": 2.375,
      "step": 3675
    },
    {
      "epoch": 2.434437086092715,
      "grad_norm": 0.842412673700197,
      "learning_rate": 0.00026018235158168084,
      "loss": 2.25,
      "step": 3676
    },
    {
      "epoch": 2.4350993377483445,
      "grad_norm": 0.9618234748444455,
      "learning_rate": 0.000260137048633393,
      "loss": 2.1719,
      "step": 3677
    },
    {
      "epoch": 2.4357615894039735,
      "grad_norm": 0.973700784335045,
      "learning_rate": 0.0002600917238767069,
      "loss": 2.2969,
      "step": 3678
    },
    {
      "epoch": 2.436423841059603,
      "grad_norm": 0.8418957199314805,
      "learning_rate": 0.00026004637732059734,
      "loss": 2.25,
      "step": 3679
    },
    {
      "epoch": 2.437086092715232,
      "grad_norm": 0.856697794327708,
      "learning_rate": 0.00026000100897404334,
      "loss": 2.2344,
      "step": 3680
    },
    {
      "epoch": 2.4377483443708607,
      "grad_norm": 0.9385182614973339,
      "learning_rate": 0.0002599556188460285,
      "loss": 2.2969,
      "step": 3681
    },
    {
      "epoch": 2.43841059602649,
      "grad_norm": 0.7859602931843207,
      "learning_rate": 0.0002599102069455405,
      "loss": 2.2188,
      "step": 3682
    },
    {
      "epoch": 2.439072847682119,
      "grad_norm": 0.9065440712055216,
      "learning_rate": 0.0002598647732815715,
      "loss": 2.4062,
      "step": 3683
    },
    {
      "epoch": 2.4397350993377485,
      "grad_norm": 0.8255287607682568,
      "learning_rate": 0.00025981931786311777,
      "loss": 2.25,
      "step": 3684
    },
    {
      "epoch": 2.4403973509933774,
      "grad_norm": 0.8603593157376187,
      "learning_rate": 0.00025977384069918007,
      "loss": 2.3281,
      "step": 3685
    },
    {
      "epoch": 2.441059602649007,
      "grad_norm": 0.7663879075345988,
      "learning_rate": 0.0002597283417987634,
      "loss": 2.0,
      "step": 3686
    },
    {
      "epoch": 2.4417218543046357,
      "grad_norm": 0.7949237063570226,
      "learning_rate": 0.0002596828211708771,
      "loss": 2.2812,
      "step": 3687
    },
    {
      "epoch": 2.442384105960265,
      "grad_norm": 0.7834429210955349,
      "learning_rate": 0.0002596372788245347,
      "loss": 2.4219,
      "step": 3688
    },
    {
      "epoch": 2.443046357615894,
      "grad_norm": 0.7931726668190414,
      "learning_rate": 0.00025959171476875416,
      "loss": 2.0156,
      "step": 3689
    },
    {
      "epoch": 2.443708609271523,
      "grad_norm": 0.8245492384761287,
      "learning_rate": 0.0002595461290125576,
      "loss": 2.5,
      "step": 3690
    },
    {
      "epoch": 2.4443708609271524,
      "grad_norm": 0.8745721561770207,
      "learning_rate": 0.00025950052156497173,
      "loss": 2.3438,
      "step": 3691
    },
    {
      "epoch": 2.4450331125827813,
      "grad_norm": 0.7579249099032168,
      "learning_rate": 0.0002594548924350271,
      "loss": 2.0156,
      "step": 3692
    },
    {
      "epoch": 2.4456953642384107,
      "grad_norm": 0.7940598235940702,
      "learning_rate": 0.0002594092416317589,
      "loss": 2.2812,
      "step": 3693
    },
    {
      "epoch": 2.4463576158940397,
      "grad_norm": 0.7181037377864162,
      "learning_rate": 0.00025936356916420657,
      "loss": 1.8906,
      "step": 3694
    },
    {
      "epoch": 2.447019867549669,
      "grad_norm": 0.7290275657274292,
      "learning_rate": 0.00025931787504141374,
      "loss": 2.2031,
      "step": 3695
    },
    {
      "epoch": 2.447682119205298,
      "grad_norm": 0.7641701887600025,
      "learning_rate": 0.0002592721592724283,
      "loss": 2.0469,
      "step": 3696
    },
    {
      "epoch": 2.4483443708609274,
      "grad_norm": 0.7735180077417082,
      "learning_rate": 0.0002592264218663026,
      "loss": 2.1875,
      "step": 3697
    },
    {
      "epoch": 2.4490066225165563,
      "grad_norm": 0.8266234811514622,
      "learning_rate": 0.00025918066283209316,
      "loss": 2.3906,
      "step": 3698
    },
    {
      "epoch": 2.4496688741721853,
      "grad_norm": 0.754399097723304,
      "learning_rate": 0.0002591348821788607,
      "loss": 2.0781,
      "step": 3699
    },
    {
      "epoch": 2.4503311258278146,
      "grad_norm": 0.7801089380252292,
      "learning_rate": 0.00025908907991567043,
      "loss": 2.125,
      "step": 3700
    },
    {
      "epoch": 2.4509933774834436,
      "grad_norm": 0.7558862254531314,
      "learning_rate": 0.0002590432560515917,
      "loss": 1.875,
      "step": 3701
    },
    {
      "epoch": 2.451655629139073,
      "grad_norm": 0.8260401207822209,
      "learning_rate": 0.000258997410595698,
      "loss": 2.0312,
      "step": 3702
    },
    {
      "epoch": 2.452317880794702,
      "grad_norm": 0.7345401689908106,
      "learning_rate": 0.0002589515435570675,
      "loss": 2.0781,
      "step": 3703
    },
    {
      "epoch": 2.4529801324503313,
      "grad_norm": 0.7947041494812258,
      "learning_rate": 0.0002589056549447823,
      "loss": 2.0938,
      "step": 3704
    },
    {
      "epoch": 2.4536423841059603,
      "grad_norm": 0.7550221168067907,
      "learning_rate": 0.0002588597447679288,
      "loss": 2.2031,
      "step": 3705
    },
    {
      "epoch": 2.4543046357615896,
      "grad_norm": 0.72700331959345,
      "learning_rate": 0.0002588138130355978,
      "loss": 1.8594,
      "step": 3706
    },
    {
      "epoch": 2.4549668874172186,
      "grad_norm": 0.7812857362563028,
      "learning_rate": 0.00025876785975688436,
      "loss": 2.125,
      "step": 3707
    },
    {
      "epoch": 2.4556291390728475,
      "grad_norm": 0.7338502259373492,
      "learning_rate": 0.00025872188494088766,
      "loss": 1.9766,
      "step": 3708
    },
    {
      "epoch": 2.456291390728477,
      "grad_norm": 0.8109446502930461,
      "learning_rate": 0.00025867588859671133,
      "loss": 1.7656,
      "step": 3709
    },
    {
      "epoch": 2.456953642384106,
      "grad_norm": 0.7663438984587653,
      "learning_rate": 0.00025862987073346305,
      "loss": 2.1406,
      "step": 3710
    },
    {
      "epoch": 2.4576158940397352,
      "grad_norm": 0.7769394682684838,
      "learning_rate": 0.00025858383136025505,
      "loss": 2.0156,
      "step": 3711
    },
    {
      "epoch": 2.458278145695364,
      "grad_norm": 0.8079150413107431,
      "learning_rate": 0.0002585377704862035,
      "loss": 2.0781,
      "step": 3712
    },
    {
      "epoch": 2.4589403973509936,
      "grad_norm": 0.767665684192676,
      "learning_rate": 0.0002584916881204291,
      "loss": 2.1406,
      "step": 3713
    },
    {
      "epoch": 2.4596026490066225,
      "grad_norm": 0.8659724611527772,
      "learning_rate": 0.00025844558427205657,
      "loss": 2.2656,
      "step": 3714
    },
    {
      "epoch": 2.460264900662252,
      "grad_norm": 0.8002472614255726,
      "learning_rate": 0.0002583994589502151,
      "loss": 1.8047,
      "step": 3715
    },
    {
      "epoch": 2.460927152317881,
      "grad_norm": 0.7759990484041954,
      "learning_rate": 0.00025835331216403807,
      "loss": 1.6719,
      "step": 3716
    },
    {
      "epoch": 2.46158940397351,
      "grad_norm": 0.8365995049284329,
      "learning_rate": 0.00025830714392266295,
      "loss": 2.2188,
      "step": 3717
    },
    {
      "epoch": 2.462251655629139,
      "grad_norm": 0.8401222617562081,
      "learning_rate": 0.0002582609542352316,
      "loss": 2.1094,
      "step": 3718
    },
    {
      "epoch": 2.462913907284768,
      "grad_norm": 0.8044452882262215,
      "learning_rate": 0.00025821474311089014,
      "loss": 2.25,
      "step": 3719
    },
    {
      "epoch": 2.4635761589403975,
      "grad_norm": 0.8219732114269241,
      "learning_rate": 0.00025816851055878896,
      "loss": 2.5469,
      "step": 3720
    },
    {
      "epoch": 2.4642384105960264,
      "grad_norm": 0.8174796258453926,
      "learning_rate": 0.00025812225658808255,
      "loss": 2.2188,
      "step": 3721
    },
    {
      "epoch": 2.4649006622516554,
      "grad_norm": 0.8108640060730444,
      "learning_rate": 0.00025807598120792976,
      "loss": 2.1094,
      "step": 3722
    },
    {
      "epoch": 2.4655629139072848,
      "grad_norm": 0.8683068022438082,
      "learning_rate": 0.00025802968442749364,
      "loss": 2.0938,
      "step": 3723
    },
    {
      "epoch": 2.466225165562914,
      "grad_norm": 0.750924543068475,
      "learning_rate": 0.0002579833662559414,
      "loss": 2.1094,
      "step": 3724
    },
    {
      "epoch": 2.466887417218543,
      "grad_norm": 0.7386933190562606,
      "learning_rate": 0.00025793702670244475,
      "loss": 2.1094,
      "step": 3725
    },
    {
      "epoch": 2.467549668874172,
      "grad_norm": 0.8536993131565537,
      "learning_rate": 0.00025789066577617934,
      "loss": 2.2188,
      "step": 3726
    },
    {
      "epoch": 2.4682119205298014,
      "grad_norm": 0.8332192958507407,
      "learning_rate": 0.00025784428348632515,
      "loss": 2.25,
      "step": 3727
    },
    {
      "epoch": 2.4688741721854304,
      "grad_norm": 0.7364051704302544,
      "learning_rate": 0.0002577978798420665,
      "loss": 1.8281,
      "step": 3728
    },
    {
      "epoch": 2.4695364238410598,
      "grad_norm": 0.7396566173883143,
      "learning_rate": 0.0002577514548525917,
      "loss": 2.2969,
      "step": 3729
    },
    {
      "epoch": 2.4701986754966887,
      "grad_norm": 0.740898850741441,
      "learning_rate": 0.00025770500852709353,
      "loss": 2.0156,
      "step": 3730
    },
    {
      "epoch": 2.4708609271523176,
      "grad_norm": 0.7729389738421756,
      "learning_rate": 0.00025765854087476893,
      "loss": 1.9375,
      "step": 3731
    },
    {
      "epoch": 2.471523178807947,
      "grad_norm": 0.8624638876107098,
      "learning_rate": 0.00025761205190481893,
      "loss": 2.2812,
      "step": 3732
    },
    {
      "epoch": 2.472185430463576,
      "grad_norm": 0.8023476474046847,
      "learning_rate": 0.0002575655416264489,
      "loss": 2.3125,
      "step": 3733
    },
    {
      "epoch": 2.4728476821192054,
      "grad_norm": 0.7255421905873592,
      "learning_rate": 0.0002575190100488684,
      "loss": 1.8281,
      "step": 3734
    },
    {
      "epoch": 2.4735099337748343,
      "grad_norm": 0.7931748315734833,
      "learning_rate": 0.0002574724571812913,
      "loss": 2.1562,
      "step": 3735
    },
    {
      "epoch": 2.4741721854304637,
      "grad_norm": 0.7662039930519703,
      "learning_rate": 0.00025742588303293555,
      "loss": 2.4062,
      "step": 3736
    },
    {
      "epoch": 2.4748344370860926,
      "grad_norm": 0.7825147613296451,
      "learning_rate": 0.00025737928761302337,
      "loss": 2.3438,
      "step": 3737
    },
    {
      "epoch": 2.475496688741722,
      "grad_norm": 0.7816595323637575,
      "learning_rate": 0.0002573326709307812,
      "loss": 2.2969,
      "step": 3738
    },
    {
      "epoch": 2.476158940397351,
      "grad_norm": 0.782474503950054,
      "learning_rate": 0.00025728603299543957,
      "loss": 2.2031,
      "step": 3739
    },
    {
      "epoch": 2.47682119205298,
      "grad_norm": 0.8265525780186007,
      "learning_rate": 0.00025723937381623343,
      "loss": 2.2344,
      "step": 3740
    },
    {
      "epoch": 2.4774834437086093,
      "grad_norm": 0.8073755991536216,
      "learning_rate": 0.00025719269340240185,
      "loss": 2.3438,
      "step": 3741
    },
    {
      "epoch": 2.4781456953642382,
      "grad_norm": 0.8453282986605583,
      "learning_rate": 0.0002571459917631881,
      "loss": 2.3125,
      "step": 3742
    },
    {
      "epoch": 2.4788079470198676,
      "grad_norm": 0.7701714877545602,
      "learning_rate": 0.0002570992689078395,
      "loss": 2.0156,
      "step": 3743
    },
    {
      "epoch": 2.4794701986754966,
      "grad_norm": 0.7535412329756577,
      "learning_rate": 0.00025705252484560786,
      "loss": 2.2188,
      "step": 3744
    },
    {
      "epoch": 2.480132450331126,
      "grad_norm": 0.7977866501667391,
      "learning_rate": 0.00025700575958574894,
      "loss": 2.1094,
      "step": 3745
    },
    {
      "epoch": 2.480794701986755,
      "grad_norm": 0.8567449504658754,
      "learning_rate": 0.0002569589731375229,
      "loss": 2.3594,
      "step": 3746
    },
    {
      "epoch": 2.4814569536423843,
      "grad_norm": 0.7443684422763787,
      "learning_rate": 0.0002569121655101939,
      "loss": 1.9609,
      "step": 3747
    },
    {
      "epoch": 2.482119205298013,
      "grad_norm": 0.754164299096999,
      "learning_rate": 0.0002568653367130304,
      "loss": 2.0625,
      "step": 3748
    },
    {
      "epoch": 2.482781456953642,
      "grad_norm": 0.7450370401636187,
      "learning_rate": 0.0002568184867553051,
      "loss": 1.7734,
      "step": 3749
    },
    {
      "epoch": 2.4834437086092715,
      "grad_norm": 0.7343611777486058,
      "learning_rate": 0.0002567716156462948,
      "loss": 1.7734,
      "step": 3750
    },
    {
      "epoch": 2.4841059602649005,
      "grad_norm": 0.8837410321717628,
      "learning_rate": 0.0002567247233952805,
      "loss": 2.0312,
      "step": 3751
    },
    {
      "epoch": 2.48476821192053,
      "grad_norm": 0.7554996689083994,
      "learning_rate": 0.00025667781001154733,
      "loss": 1.8984,
      "step": 3752
    },
    {
      "epoch": 2.485430463576159,
      "grad_norm": 0.8562522973241116,
      "learning_rate": 0.0002566308755043848,
      "loss": 2.2031,
      "step": 3753
    },
    {
      "epoch": 2.486092715231788,
      "grad_norm": 0.8354555574608559,
      "learning_rate": 0.00025658391988308643,
      "loss": 2.2188,
      "step": 3754
    },
    {
      "epoch": 2.486754966887417,
      "grad_norm": 0.8557149768317479,
      "learning_rate": 0.00025653694315695,
      "loss": 2.1094,
      "step": 3755
    },
    {
      "epoch": 2.4874172185430465,
      "grad_norm": 0.8166114702044995,
      "learning_rate": 0.00025648994533527744,
      "loss": 2.125,
      "step": 3756
    },
    {
      "epoch": 2.4880794701986755,
      "grad_norm": 0.7882667142255402,
      "learning_rate": 0.0002564429264273748,
      "loss": 2.2344,
      "step": 3757
    },
    {
      "epoch": 2.4887417218543044,
      "grad_norm": 0.7843107808145644,
      "learning_rate": 0.0002563958864425524,
      "loss": 2.1562,
      "step": 3758
    },
    {
      "epoch": 2.489403973509934,
      "grad_norm": 0.8913234831605202,
      "learning_rate": 0.00025634882539012473,
      "loss": 2.4375,
      "step": 3759
    },
    {
      "epoch": 2.4900662251655628,
      "grad_norm": 0.7760531183654524,
      "learning_rate": 0.00025630174327941036,
      "loss": 2.0312,
      "step": 3760
    },
    {
      "epoch": 2.490728476821192,
      "grad_norm": 0.8237800597839532,
      "learning_rate": 0.0002562546401197321,
      "loss": 2.2344,
      "step": 3761
    },
    {
      "epoch": 2.491390728476821,
      "grad_norm": 0.7538543525796487,
      "learning_rate": 0.00025620751592041695,
      "loss": 1.9688,
      "step": 3762
    },
    {
      "epoch": 2.4920529801324505,
      "grad_norm": 0.7671719916062989,
      "learning_rate": 0.000256160370690796,
      "loss": 2.125,
      "step": 3763
    },
    {
      "epoch": 2.4927152317880794,
      "grad_norm": 0.7264601185989904,
      "learning_rate": 0.00025611320444020454,
      "loss": 2.125,
      "step": 3764
    },
    {
      "epoch": 2.493377483443709,
      "grad_norm": 0.7540591425870986,
      "learning_rate": 0.00025606601717798207,
      "loss": 2.0469,
      "step": 3765
    },
    {
      "epoch": 2.4940397350993377,
      "grad_norm": 0.8000031191656632,
      "learning_rate": 0.00025601880891347223,
      "loss": 2.3281,
      "step": 3766
    },
    {
      "epoch": 2.4947019867549667,
      "grad_norm": 0.9173837159614713,
      "learning_rate": 0.00025597157965602275,
      "loss": 2.5156,
      "step": 3767
    },
    {
      "epoch": 2.495364238410596,
      "grad_norm": 0.8574803970528944,
      "learning_rate": 0.00025592432941498556,
      "loss": 2.4219,
      "step": 3768
    },
    {
      "epoch": 2.496026490066225,
      "grad_norm": 0.7116906170460164,
      "learning_rate": 0.00025587705819971675,
      "loss": 1.7266,
      "step": 3769
    },
    {
      "epoch": 2.4966887417218544,
      "grad_norm": 0.8645167719637118,
      "learning_rate": 0.0002558297660195766,
      "loss": 2.3594,
      "step": 3770
    },
    {
      "epoch": 2.4973509933774833,
      "grad_norm": 0.8022357297590467,
      "learning_rate": 0.00025578245288392944,
      "loss": 2.3281,
      "step": 3771
    },
    {
      "epoch": 2.4980132450331127,
      "grad_norm": 0.6870707544917136,
      "learning_rate": 0.0002557351188021439,
      "loss": 1.75,
      "step": 3772
    },
    {
      "epoch": 2.4986754966887417,
      "grad_norm": 0.7505084647297366,
      "learning_rate": 0.0002556877637835926,
      "loss": 2.0625,
      "step": 3773
    },
    {
      "epoch": 2.499337748344371,
      "grad_norm": 0.7825213787198477,
      "learning_rate": 0.0002556403878376524,
      "loss": 2.3594,
      "step": 3774
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8331305704984378,
      "learning_rate": 0.00025559299097370426,
      "loss": 2.2344,
      "step": 3775
    },
    {
      "epoch": 2.500662251655629,
      "grad_norm": 0.8127604660915183,
      "learning_rate": 0.00025554557320113335,
      "loss": 2.2188,
      "step": 3776
    },
    {
      "epoch": 2.5013245033112583,
      "grad_norm": 0.8757668774831419,
      "learning_rate": 0.00025549813452932887,
      "loss": 2.3281,
      "step": 3777
    },
    {
      "epoch": 2.5019867549668873,
      "grad_norm": 0.796343673496541,
      "learning_rate": 0.0002554506749676843,
      "loss": 2.0312,
      "step": 3778
    },
    {
      "epoch": 2.5026490066225167,
      "grad_norm": 0.697684377385825,
      "learning_rate": 0.00025540319452559715,
      "loss": 1.7422,
      "step": 3779
    },
    {
      "epoch": 2.5033112582781456,
      "grad_norm": 0.7508516375462173,
      "learning_rate": 0.000255355693212469,
      "loss": 2.0625,
      "step": 3780
    },
    {
      "epoch": 2.503973509933775,
      "grad_norm": 0.8995981351168479,
      "learning_rate": 0.0002553081710377058,
      "loss": 2.2656,
      "step": 3781
    },
    {
      "epoch": 2.504635761589404,
      "grad_norm": 0.9442778658895785,
      "learning_rate": 0.0002552606280107174,
      "loss": 2.4531,
      "step": 3782
    },
    {
      "epoch": 2.5052980132450333,
      "grad_norm": 0.848668594315267,
      "learning_rate": 0.0002552130641409179,
      "loss": 2.1875,
      "step": 3783
    },
    {
      "epoch": 2.5059602649006623,
      "grad_norm": 0.8128908052676902,
      "learning_rate": 0.00025516547943772543,
      "loss": 2.2188,
      "step": 3784
    },
    {
      "epoch": 2.506622516556291,
      "grad_norm": 0.817947238228866,
      "learning_rate": 0.00025511787391056237,
      "loss": 2.0781,
      "step": 3785
    },
    {
      "epoch": 2.5072847682119206,
      "grad_norm": 0.7906552605647277,
      "learning_rate": 0.0002550702475688551,
      "loss": 2.2969,
      "step": 3786
    },
    {
      "epoch": 2.5079470198675495,
      "grad_norm": 0.8487939423019191,
      "learning_rate": 0.0002550226004220343,
      "loss": 2.5,
      "step": 3787
    },
    {
      "epoch": 2.508609271523179,
      "grad_norm": 0.783383233932687,
      "learning_rate": 0.0002549749324795345,
      "loss": 2.0781,
      "step": 3788
    },
    {
      "epoch": 2.509271523178808,
      "grad_norm": 0.8424805306609977,
      "learning_rate": 0.00025492724375079457,
      "loss": 2.3125,
      "step": 3789
    },
    {
      "epoch": 2.5099337748344372,
      "grad_norm": 0.8537090061892594,
      "learning_rate": 0.00025487953424525747,
      "loss": 2.375,
      "step": 3790
    },
    {
      "epoch": 2.510596026490066,
      "grad_norm": 0.8162934038231799,
      "learning_rate": 0.0002548318039723702,
      "loss": 2.0625,
      "step": 3791
    },
    {
      "epoch": 2.5112582781456956,
      "grad_norm": 0.7981103742956785,
      "learning_rate": 0.0002547840529415838,
      "loss": 2.3281,
      "step": 3792
    },
    {
      "epoch": 2.5119205298013245,
      "grad_norm": 0.8942816627767144,
      "learning_rate": 0.0002547362811623537,
      "loss": 2.0625,
      "step": 3793
    },
    {
      "epoch": 2.5125827814569535,
      "grad_norm": 0.8452463509791648,
      "learning_rate": 0.00025468848864413913,
      "loss": 2.1875,
      "step": 3794
    },
    {
      "epoch": 2.513245033112583,
      "grad_norm": 0.8139976073956928,
      "learning_rate": 0.00025464067539640356,
      "loss": 1.8828,
      "step": 3795
    },
    {
      "epoch": 2.513907284768212,
      "grad_norm": 0.849179213799658,
      "learning_rate": 0.00025459284142861465,
      "loss": 2.0781,
      "step": 3796
    },
    {
      "epoch": 2.514569536423841,
      "grad_norm": 0.80543474258563,
      "learning_rate": 0.00025454498675024395,
      "loss": 2.3438,
      "step": 3797
    },
    {
      "epoch": 2.51523178807947,
      "grad_norm": 0.8449208722342026,
      "learning_rate": 0.00025449711137076733,
      "loss": 2.4531,
      "step": 3798
    },
    {
      "epoch": 2.515894039735099,
      "grad_norm": 0.7963480556022876,
      "learning_rate": 0.00025444921529966467,
      "loss": 1.7812,
      "step": 3799
    },
    {
      "epoch": 2.5165562913907285,
      "grad_norm": 0.7979864641611971,
      "learning_rate": 0.0002544012985464199,
      "loss": 1.7344,
      "step": 3800
    },
    {
      "epoch": 2.517218543046358,
      "grad_norm": 0.8261186208922794,
      "learning_rate": 0.0002543533611205211,
      "loss": 2.2188,
      "step": 3801
    },
    {
      "epoch": 2.517880794701987,
      "grad_norm": 0.7656984243107973,
      "learning_rate": 0.00025430540303146044,
      "loss": 2.0,
      "step": 3802
    },
    {
      "epoch": 2.5185430463576157,
      "grad_norm": 0.8187946922760879,
      "learning_rate": 0.0002542574242887341,
      "loss": 2.1406,
      "step": 3803
    },
    {
      "epoch": 2.519205298013245,
      "grad_norm": 0.7909393604313028,
      "learning_rate": 0.00025420942490184253,
      "loss": 2.2656,
      "step": 3804
    },
    {
      "epoch": 2.519867549668874,
      "grad_norm": 0.7923129991722448,
      "learning_rate": 0.0002541614048802901,
      "loss": 2.2188,
      "step": 3805
    },
    {
      "epoch": 2.5205298013245034,
      "grad_norm": 0.8009879228520259,
      "learning_rate": 0.00025411336423358533,
      "loss": 2.0938,
      "step": 3806
    },
    {
      "epoch": 2.5211920529801324,
      "grad_norm": 0.8765228361984988,
      "learning_rate": 0.0002540653029712408,
      "loss": 2.1562,
      "step": 3807
    },
    {
      "epoch": 2.5218543046357613,
      "grad_norm": 0.8177431591227649,
      "learning_rate": 0.00025401722110277323,
      "loss": 2.2031,
      "step": 3808
    },
    {
      "epoch": 2.5225165562913907,
      "grad_norm": 0.7852915290432646,
      "learning_rate": 0.0002539691186377034,
      "loss": 2.3125,
      "step": 3809
    },
    {
      "epoch": 2.52317880794702,
      "grad_norm": 0.7835910034076287,
      "learning_rate": 0.00025392099558555603,
      "loss": 1.9922,
      "step": 3810
    },
    {
      "epoch": 2.523841059602649,
      "grad_norm": 0.820804249496446,
      "learning_rate": 0.0002538728519558601,
      "loss": 2.1875,
      "step": 3811
    },
    {
      "epoch": 2.524503311258278,
      "grad_norm": 0.7604540774669745,
      "learning_rate": 0.0002538246877581487,
      "loss": 2.0625,
      "step": 3812
    },
    {
      "epoch": 2.5251655629139074,
      "grad_norm": 0.8402028053161055,
      "learning_rate": 0.0002537765030019588,
      "loss": 2.3281,
      "step": 3813
    },
    {
      "epoch": 2.5258278145695363,
      "grad_norm": 0.7518663424086643,
      "learning_rate": 0.0002537282976968316,
      "loss": 2.0938,
      "step": 3814
    },
    {
      "epoch": 2.5264900662251657,
      "grad_norm": 0.758418025852073,
      "learning_rate": 0.0002536800718523122,
      "loss": 2.125,
      "step": 3815
    },
    {
      "epoch": 2.5271523178807946,
      "grad_norm": 0.8030682646351185,
      "learning_rate": 0.00025363182547794987,
      "loss": 2.3281,
      "step": 3816
    },
    {
      "epoch": 2.5278145695364236,
      "grad_norm": 0.7977846492405214,
      "learning_rate": 0.00025358355858329804,
      "loss": 2.2969,
      "step": 3817
    },
    {
      "epoch": 2.528476821192053,
      "grad_norm": 0.697532931937724,
      "learning_rate": 0.00025353527117791405,
      "loss": 1.9531,
      "step": 3818
    },
    {
      "epoch": 2.5291390728476824,
      "grad_norm": 0.7431021991406611,
      "learning_rate": 0.0002534869632713594,
      "loss": 2.2031,
      "step": 3819
    },
    {
      "epoch": 2.5298013245033113,
      "grad_norm": 0.7206783428453616,
      "learning_rate": 0.00025343863487319957,
      "loss": 2.1719,
      "step": 3820
    },
    {
      "epoch": 2.5304635761589402,
      "grad_norm": 0.7655455629253477,
      "learning_rate": 0.0002533902859930041,
      "loss": 1.8594,
      "step": 3821
    },
    {
      "epoch": 2.5311258278145696,
      "grad_norm": 0.8014182720001765,
      "learning_rate": 0.00025334191664034677,
      "loss": 2.0625,
      "step": 3822
    },
    {
      "epoch": 2.5317880794701986,
      "grad_norm": 0.8222967477942553,
      "learning_rate": 0.0002532935268248051,
      "loss": 2.1875,
      "step": 3823
    },
    {
      "epoch": 2.532450331125828,
      "grad_norm": 0.8357026003372796,
      "learning_rate": 0.00025324511655596093,
      "loss": 2.125,
      "step": 3824
    },
    {
      "epoch": 2.533112582781457,
      "grad_norm": 0.7276083069883442,
      "learning_rate": 0.0002531966858434,
      "loss": 1.7656,
      "step": 3825
    },
    {
      "epoch": 2.533774834437086,
      "grad_norm": 0.7588331997235731,
      "learning_rate": 0.0002531482346967121,
      "loss": 1.8906,
      "step": 3826
    },
    {
      "epoch": 2.5344370860927152,
      "grad_norm": 0.7996064951243721,
      "learning_rate": 0.00025309976312549125,
      "loss": 2.125,
      "step": 3827
    },
    {
      "epoch": 2.5350993377483446,
      "grad_norm": 0.8378376099200675,
      "learning_rate": 0.00025305127113933534,
      "loss": 2.2969,
      "step": 3828
    },
    {
      "epoch": 2.5357615894039736,
      "grad_norm": 0.7013426592142689,
      "learning_rate": 0.0002530027587478462,
      "loss": 1.6953,
      "step": 3829
    },
    {
      "epoch": 2.5364238410596025,
      "grad_norm": 0.7935753869658606,
      "learning_rate": 0.00025295422596063,
      "loss": 2.2969,
      "step": 3830
    },
    {
      "epoch": 2.537086092715232,
      "grad_norm": 0.7367119358807327,
      "learning_rate": 0.00025290567278729673,
      "loss": 1.9375,
      "step": 3831
    },
    {
      "epoch": 2.537748344370861,
      "grad_norm": 0.82439960677723,
      "learning_rate": 0.0002528570992374604,
      "loss": 2.1875,
      "step": 3832
    },
    {
      "epoch": 2.53841059602649,
      "grad_norm": 0.8318401715741789,
      "learning_rate": 0.0002528085053207393,
      "loss": 2.4844,
      "step": 3833
    },
    {
      "epoch": 2.539072847682119,
      "grad_norm": 0.8423100442204545,
      "learning_rate": 0.00025275989104675545,
      "loss": 2.3438,
      "step": 3834
    },
    {
      "epoch": 2.539735099337748,
      "grad_norm": 0.7580464044835085,
      "learning_rate": 0.00025271125642513506,
      "loss": 2.0625,
      "step": 3835
    },
    {
      "epoch": 2.5403973509933775,
      "grad_norm": 0.8202448178531816,
      "learning_rate": 0.0002526626014655083,
      "loss": 2.3125,
      "step": 3836
    },
    {
      "epoch": 2.541059602649007,
      "grad_norm": 0.8581297181064826,
      "learning_rate": 0.0002526139261775096,
      "loss": 2.2656,
      "step": 3837
    },
    {
      "epoch": 2.541721854304636,
      "grad_norm": 0.7770333772543426,
      "learning_rate": 0.000252565230570777,
      "loss": 2.0938,
      "step": 3838
    },
    {
      "epoch": 2.5423841059602648,
      "grad_norm": 0.7600408267322143,
      "learning_rate": 0.0002525165146549528,
      "loss": 2.1719,
      "step": 3839
    },
    {
      "epoch": 2.543046357615894,
      "grad_norm": 0.8145785006120807,
      "learning_rate": 0.0002524677784396835,
      "loss": 2.3281,
      "step": 3840
    },
    {
      "epoch": 2.543708609271523,
      "grad_norm": 0.7932561086411951,
      "learning_rate": 0.00025241902193461925,
      "loss": 1.8594,
      "step": 3841
    },
    {
      "epoch": 2.5443708609271525,
      "grad_norm": 0.816912772459861,
      "learning_rate": 0.0002523702451494145,
      "loss": 2.0312,
      "step": 3842
    },
    {
      "epoch": 2.5450331125827814,
      "grad_norm": 0.7893623463166362,
      "learning_rate": 0.00025232144809372755,
      "loss": 1.8203,
      "step": 3843
    },
    {
      "epoch": 2.5456953642384104,
      "grad_norm": 0.8119976781239691,
      "learning_rate": 0.0002522726307772208,
      "loss": 2.0938,
      "step": 3844
    },
    {
      "epoch": 2.5463576158940397,
      "grad_norm": 0.737490234660676,
      "learning_rate": 0.0002522237932095607,
      "loss": 1.9531,
      "step": 3845
    },
    {
      "epoch": 2.547019867549669,
      "grad_norm": 0.8219803837715125,
      "learning_rate": 0.0002521749354004175,
      "loss": 2.2188,
      "step": 3846
    },
    {
      "epoch": 2.547682119205298,
      "grad_norm": 0.746898840525575,
      "learning_rate": 0.00025212605735946573,
      "loss": 2.0781,
      "step": 3847
    },
    {
      "epoch": 2.548344370860927,
      "grad_norm": 0.84047601011077,
      "learning_rate": 0.0002520771590963838,
      "loss": 2.3281,
      "step": 3848
    },
    {
      "epoch": 2.5490066225165564,
      "grad_norm": 0.7470349945710681,
      "learning_rate": 0.0002520282406208541,
      "loss": 1.9141,
      "step": 3849
    },
    {
      "epoch": 2.5496688741721854,
      "grad_norm": 0.886438730134503,
      "learning_rate": 0.00025197930194256296,
      "loss": 2.2344,
      "step": 3850
    },
    {
      "epoch": 2.5503311258278147,
      "grad_norm": 0.8008237416313609,
      "learning_rate": 0.00025193034307120093,
      "loss": 2.0156,
      "step": 3851
    },
    {
      "epoch": 2.5509933774834437,
      "grad_norm": 0.7750035070561504,
      "learning_rate": 0.00025188136401646234,
      "loss": 2.3281,
      "step": 3852
    },
    {
      "epoch": 2.5516556291390726,
      "grad_norm": 0.8144553860891628,
      "learning_rate": 0.00025183236478804565,
      "loss": 2.375,
      "step": 3853
    },
    {
      "epoch": 2.552317880794702,
      "grad_norm": 0.7571409837859496,
      "learning_rate": 0.0002517833453956533,
      "loss": 1.9219,
      "step": 3854
    },
    {
      "epoch": 2.5529801324503314,
      "grad_norm": 0.8286387279034257,
      "learning_rate": 0.0002517343058489917,
      "loss": 2.125,
      "step": 3855
    },
    {
      "epoch": 2.5536423841059603,
      "grad_norm": 0.7662489255544056,
      "learning_rate": 0.0002516852461577711,
      "loss": 2.125,
      "step": 3856
    },
    {
      "epoch": 2.5543046357615893,
      "grad_norm": 0.776737139507395,
      "learning_rate": 0.000251636166331706,
      "loss": 2.1094,
      "step": 3857
    },
    {
      "epoch": 2.5549668874172187,
      "grad_norm": 0.7450599419848773,
      "learning_rate": 0.0002515870663805148,
      "loss": 2.0469,
      "step": 3858
    },
    {
      "epoch": 2.5556291390728476,
      "grad_norm": 0.8460261715935387,
      "learning_rate": 0.0002515379463139197,
      "loss": 2.3281,
      "step": 3859
    },
    {
      "epoch": 2.556291390728477,
      "grad_norm": 0.7901148467077456,
      "learning_rate": 0.0002514888061416472,
      "loss": 2.0781,
      "step": 3860
    },
    {
      "epoch": 2.556953642384106,
      "grad_norm": 0.7571472625483164,
      "learning_rate": 0.0002514396458734276,
      "loss": 1.9453,
      "step": 3861
    },
    {
      "epoch": 2.557615894039735,
      "grad_norm": 0.7446337855542936,
      "learning_rate": 0.00025139046551899503,
      "loss": 1.7344,
      "step": 3862
    },
    {
      "epoch": 2.5582781456953643,
      "grad_norm": 0.8532667038400915,
      "learning_rate": 0.000251341265088088,
      "loss": 2.375,
      "step": 3863
    },
    {
      "epoch": 2.558940397350993,
      "grad_norm": 0.7420028567272337,
      "learning_rate": 0.00025129204459044854,
      "loss": 2.0781,
      "step": 3864
    },
    {
      "epoch": 2.5596026490066226,
      "grad_norm": 0.7596643386979913,
      "learning_rate": 0.0002512428040358231,
      "loss": 1.9062,
      "step": 3865
    },
    {
      "epoch": 2.5602649006622515,
      "grad_norm": 0.7574154303372109,
      "learning_rate": 0.00025119354343396165,
      "loss": 1.9766,
      "step": 3866
    },
    {
      "epoch": 2.560927152317881,
      "grad_norm": 0.8264921010946084,
      "learning_rate": 0.00025114426279461844,
      "loss": 2.2812,
      "step": 3867
    },
    {
      "epoch": 2.56158940397351,
      "grad_norm": 0.7804975081695901,
      "learning_rate": 0.00025109496212755167,
      "loss": 2.1875,
      "step": 3868
    },
    {
      "epoch": 2.5622516556291393,
      "grad_norm": 0.6984941253379536,
      "learning_rate": 0.00025104564144252337,
      "loss": 1.7578,
      "step": 3869
    },
    {
      "epoch": 2.562913907284768,
      "grad_norm": 0.795031467703789,
      "learning_rate": 0.0002509963007492996,
      "loss": 2.125,
      "step": 3870
    },
    {
      "epoch": 2.563576158940397,
      "grad_norm": 0.8906735255965582,
      "learning_rate": 0.0002509469400576504,
      "loss": 2.4375,
      "step": 3871
    },
    {
      "epoch": 2.5642384105960265,
      "grad_norm": 0.7837441486020402,
      "learning_rate": 0.0002508975593773498,
      "loss": 2.2344,
      "step": 3872
    },
    {
      "epoch": 2.5649006622516555,
      "grad_norm": 0.7754712567649199,
      "learning_rate": 0.00025084815871817567,
      "loss": 2.0312,
      "step": 3873
    },
    {
      "epoch": 2.565562913907285,
      "grad_norm": 0.8368026191313812,
      "learning_rate": 0.00025079873808990993,
      "loss": 2.4062,
      "step": 3874
    },
    {
      "epoch": 2.566225165562914,
      "grad_norm": 0.8056979895304843,
      "learning_rate": 0.00025074929750233834,
      "loss": 2.0469,
      "step": 3875
    },
    {
      "epoch": 2.566887417218543,
      "grad_norm": 0.7737076296081229,
      "learning_rate": 0.00025069983696525086,
      "loss": 2.0938,
      "step": 3876
    },
    {
      "epoch": 2.567549668874172,
      "grad_norm": 0.8821511327502463,
      "learning_rate": 0.0002506503564884412,
      "loss": 2.2188,
      "step": 3877
    },
    {
      "epoch": 2.5682119205298015,
      "grad_norm": 0.7850543575422075,
      "learning_rate": 0.000250600856081707,
      "loss": 2.2969,
      "step": 3878
    },
    {
      "epoch": 2.5688741721854305,
      "grad_norm": 0.7974938824474251,
      "learning_rate": 0.00025055133575485,
      "loss": 2.3438,
      "step": 3879
    },
    {
      "epoch": 2.5695364238410594,
      "grad_norm": 0.8057804249562326,
      "learning_rate": 0.00025050179551767564,
      "loss": 2.0781,
      "step": 3880
    },
    {
      "epoch": 2.570198675496689,
      "grad_norm": 0.8380163782506149,
      "learning_rate": 0.0002504522353799936,
      "loss": 2.3125,
      "step": 3881
    },
    {
      "epoch": 2.5708609271523177,
      "grad_norm": 0.793542440330441,
      "learning_rate": 0.0002504026553516173,
      "loss": 2.4219,
      "step": 3882
    },
    {
      "epoch": 2.571523178807947,
      "grad_norm": 0.7196688817113375,
      "learning_rate": 0.0002503530554423642,
      "loss": 2.125,
      "step": 3883
    },
    {
      "epoch": 2.572185430463576,
      "grad_norm": 0.790425489076223,
      "learning_rate": 0.0002503034356620556,
      "loss": 2.125,
      "step": 3884
    },
    {
      "epoch": 2.5728476821192054,
      "grad_norm": 0.8057085002503803,
      "learning_rate": 0.00025025379602051673,
      "loss": 2.1719,
      "step": 3885
    },
    {
      "epoch": 2.5735099337748344,
      "grad_norm": 0.8376618130457825,
      "learning_rate": 0.00025020413652757697,
      "loss": 2.2812,
      "step": 3886
    },
    {
      "epoch": 2.5741721854304638,
      "grad_norm": 0.8016221358236878,
      "learning_rate": 0.0002501544571930693,
      "loss": 2.0938,
      "step": 3887
    },
    {
      "epoch": 2.5748344370860927,
      "grad_norm": 0.7847848188366371,
      "learning_rate": 0.0002501047580268309,
      "loss": 1.8672,
      "step": 3888
    },
    {
      "epoch": 2.5754966887417217,
      "grad_norm": 0.8862670117657797,
      "learning_rate": 0.0002500550390387028,
      "loss": 2.1875,
      "step": 3889
    },
    {
      "epoch": 2.576158940397351,
      "grad_norm": 0.8455305159506907,
      "learning_rate": 0.0002500053002385298,
      "loss": 2.3438,
      "step": 3890
    },
    {
      "epoch": 2.57682119205298,
      "grad_norm": 0.7520496874918873,
      "learning_rate": 0.00024995554163616084,
      "loss": 2.0156,
      "step": 3891
    },
    {
      "epoch": 2.5774834437086094,
      "grad_norm": 0.7916388899951395,
      "learning_rate": 0.0002499057632414487,
      "loss": 2.1406,
      "step": 3892
    },
    {
      "epoch": 2.5781456953642383,
      "grad_norm": 0.7721191100194016,
      "learning_rate": 0.00024985596506425003,
      "loss": 2.1094,
      "step": 3893
    },
    {
      "epoch": 2.5788079470198677,
      "grad_norm": 0.8198154597827289,
      "learning_rate": 0.0002498061471144255,
      "loss": 2.4688,
      "step": 3894
    },
    {
      "epoch": 2.5794701986754967,
      "grad_norm": 0.8064199334058884,
      "learning_rate": 0.00024975630940183966,
      "loss": 2.2188,
      "step": 3895
    },
    {
      "epoch": 2.580132450331126,
      "grad_norm": 0.7305130426522344,
      "learning_rate": 0.0002497064519363608,
      "loss": 2.0781,
      "step": 3896
    },
    {
      "epoch": 2.580794701986755,
      "grad_norm": 0.7498111705044653,
      "learning_rate": 0.0002496565747278614,
      "loss": 1.8984,
      "step": 3897
    },
    {
      "epoch": 2.581456953642384,
      "grad_norm": 0.7590683379340506,
      "learning_rate": 0.00024960667778621765,
      "loss": 2.0312,
      "step": 3898
    },
    {
      "epoch": 2.5821192052980133,
      "grad_norm": 0.783864640611157,
      "learning_rate": 0.00024955676112130976,
      "loss": 2.0156,
      "step": 3899
    },
    {
      "epoch": 2.5827814569536423,
      "grad_norm": 0.7770826105235369,
      "learning_rate": 0.0002495068247430218,
      "loss": 2.0312,
      "step": 3900
    },
    {
      "epoch": 2.5834437086092716,
      "grad_norm": 0.800684155541999,
      "learning_rate": 0.0002494568686612417,
      "loss": 2.3281,
      "step": 3901
    },
    {
      "epoch": 2.5841059602649006,
      "grad_norm": 0.7634111481055363,
      "learning_rate": 0.00024940689288586136,
      "loss": 1.9531,
      "step": 3902
    },
    {
      "epoch": 2.58476821192053,
      "grad_norm": 0.7832052277772961,
      "learning_rate": 0.0002493568974267766,
      "loss": 2.1406,
      "step": 3903
    },
    {
      "epoch": 2.585430463576159,
      "grad_norm": 0.8647448520588311,
      "learning_rate": 0.000249306882293887,
      "loss": 2.4062,
      "step": 3904
    },
    {
      "epoch": 2.5860927152317883,
      "grad_norm": 0.7974701684893966,
      "learning_rate": 0.0002492568474970963,
      "loss": 2.1875,
      "step": 3905
    },
    {
      "epoch": 2.5867549668874172,
      "grad_norm": 0.8444881083020391,
      "learning_rate": 0.00024920679304631175,
      "loss": 2.3125,
      "step": 3906
    },
    {
      "epoch": 2.587417218543046,
      "grad_norm": 0.8457713890441648,
      "learning_rate": 0.00024915671895144487,
      "loss": 2.0469,
      "step": 3907
    },
    {
      "epoch": 2.5880794701986756,
      "grad_norm": 0.7906509776985224,
      "learning_rate": 0.0002491066252224108,
      "loss": 2.1719,
      "step": 3908
    },
    {
      "epoch": 2.5887417218543045,
      "grad_norm": 0.7299289061804439,
      "learning_rate": 0.0002490565118691288,
      "loss": 2.0469,
      "step": 3909
    },
    {
      "epoch": 2.589403973509934,
      "grad_norm": 0.7905565234884893,
      "learning_rate": 0.0002490063789015217,
      "loss": 2.125,
      "step": 3910
    },
    {
      "epoch": 2.590066225165563,
      "grad_norm": 0.7683221796802914,
      "learning_rate": 0.0002489562263295166,
      "loss": 2.0469,
      "step": 3911
    },
    {
      "epoch": 2.590728476821192,
      "grad_norm": 0.8803555791993387,
      "learning_rate": 0.0002489060541630441,
      "loss": 2.2031,
      "step": 3912
    },
    {
      "epoch": 2.591390728476821,
      "grad_norm": 0.7626386118997812,
      "learning_rate": 0.00024885586241203905,
      "loss": 1.75,
      "step": 3913
    },
    {
      "epoch": 2.5920529801324506,
      "grad_norm": 0.8097270998924269,
      "learning_rate": 0.0002488056510864398,
      "loss": 2.1875,
      "step": 3914
    },
    {
      "epoch": 2.5927152317880795,
      "grad_norm": 0.7915051937119802,
      "learning_rate": 0.0002487554201961889,
      "loss": 2.3438,
      "step": 3915
    },
    {
      "epoch": 2.5933774834437084,
      "grad_norm": 0.7859923727052803,
      "learning_rate": 0.0002487051697512326,
      "loss": 2.0781,
      "step": 3916
    },
    {
      "epoch": 2.594039735099338,
      "grad_norm": 0.8268539256371864,
      "learning_rate": 0.0002486548997615211,
      "loss": 2.0,
      "step": 3917
    },
    {
      "epoch": 2.5947019867549668,
      "grad_norm": 0.7464508124294178,
      "learning_rate": 0.0002486046102370084,
      "loss": 2.2344,
      "step": 3918
    },
    {
      "epoch": 2.595364238410596,
      "grad_norm": 0.8414754123614265,
      "learning_rate": 0.0002485543011876524,
      "loss": 2.1875,
      "step": 3919
    },
    {
      "epoch": 2.596026490066225,
      "grad_norm": 0.8364486476440537,
      "learning_rate": 0.0002485039726234149,
      "loss": 2.375,
      "step": 3920
    },
    {
      "epoch": 2.596688741721854,
      "grad_norm": 0.8499332486914752,
      "learning_rate": 0.00024845362455426143,
      "loss": 2.1562,
      "step": 3921
    },
    {
      "epoch": 2.5973509933774834,
      "grad_norm": 0.790382090333589,
      "learning_rate": 0.00024840325699016164,
      "loss": 2.2188,
      "step": 3922
    },
    {
      "epoch": 2.598013245033113,
      "grad_norm": 0.8460013384668975,
      "learning_rate": 0.0002483528699410888,
      "loss": 2.5,
      "step": 3923
    },
    {
      "epoch": 2.5986754966887418,
      "grad_norm": 0.8201312812497731,
      "learning_rate": 0.00024830246341702014,
      "loss": 2.4219,
      "step": 3924
    },
    {
      "epoch": 2.5993377483443707,
      "grad_norm": 0.8111659467613763,
      "learning_rate": 0.0002482520374279367,
      "loss": 2.2656,
      "step": 3925
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.740928949865739,
      "learning_rate": 0.00024820159198382343,
      "loss": 2.0625,
      "step": 3926
    },
    {
      "epoch": 2.600662251655629,
      "grad_norm": 0.8269110977117258,
      "learning_rate": 0.0002481511270946691,
      "loss": 2.2812,
      "step": 3927
    },
    {
      "epoch": 2.6013245033112584,
      "grad_norm": 0.753430514367967,
      "learning_rate": 0.0002481006427704663,
      "loss": 2.125,
      "step": 3928
    },
    {
      "epoch": 2.6019867549668874,
      "grad_norm": 0.7126898240367171,
      "learning_rate": 0.00024805013902121163,
      "loss": 1.7266,
      "step": 3929
    },
    {
      "epoch": 2.6026490066225163,
      "grad_norm": 0.8762948430242081,
      "learning_rate": 0.0002479996158569053,
      "loss": 2.2031,
      "step": 3930
    },
    {
      "epoch": 2.6033112582781457,
      "grad_norm": 0.7843506402537388,
      "learning_rate": 0.00024794907328755145,
      "loss": 2.0469,
      "step": 3931
    },
    {
      "epoch": 2.603973509933775,
      "grad_norm": 0.7381331086742765,
      "learning_rate": 0.00024789851132315816,
      "loss": 2.1094,
      "step": 3932
    },
    {
      "epoch": 2.604635761589404,
      "grad_norm": 0.7670185246648662,
      "learning_rate": 0.0002478479299737373,
      "loss": 2.125,
      "step": 3933
    },
    {
      "epoch": 2.605298013245033,
      "grad_norm": 0.8211015603506455,
      "learning_rate": 0.0002477973292493045,
      "loss": 2.2188,
      "step": 3934
    },
    {
      "epoch": 2.6059602649006623,
      "grad_norm": 0.8443723710307479,
      "learning_rate": 0.0002477467091598793,
      "loss": 2.3281,
      "step": 3935
    },
    {
      "epoch": 2.6066225165562913,
      "grad_norm": 0.7899296007156675,
      "learning_rate": 0.00024769606971548506,
      "loss": 2.3125,
      "step": 3936
    },
    {
      "epoch": 2.6072847682119207,
      "grad_norm": 0.9183019042447146,
      "learning_rate": 0.0002476454109261489,
      "loss": 2.2188,
      "step": 3937
    },
    {
      "epoch": 2.6079470198675496,
      "grad_norm": 0.7841027032669037,
      "learning_rate": 0.00024759473280190197,
      "loss": 2.1719,
      "step": 3938
    },
    {
      "epoch": 2.6086092715231786,
      "grad_norm": 0.7833066897542446,
      "learning_rate": 0.000247544035352779,
      "loss": 2.0625,
      "step": 3939
    },
    {
      "epoch": 2.609271523178808,
      "grad_norm": 0.7837985822577075,
      "learning_rate": 0.0002474933185888188,
      "loss": 1.7188,
      "step": 3940
    },
    {
      "epoch": 2.6099337748344373,
      "grad_norm": 0.7456934853250844,
      "learning_rate": 0.00024744258252006377,
      "loss": 1.9844,
      "step": 3941
    },
    {
      "epoch": 2.6105960264900663,
      "grad_norm": 0.7907739402003513,
      "learning_rate": 0.0002473918271565603,
      "loss": 2.0938,
      "step": 3942
    },
    {
      "epoch": 2.611258278145695,
      "grad_norm": 0.754651893790459,
      "learning_rate": 0.0002473410525083584,
      "loss": 2.1562,
      "step": 3943
    },
    {
      "epoch": 2.6119205298013246,
      "grad_norm": 0.7631920670205864,
      "learning_rate": 0.0002472902585855122,
      "loss": 2.1719,
      "step": 3944
    },
    {
      "epoch": 2.6125827814569536,
      "grad_norm": 0.8016581288052913,
      "learning_rate": 0.0002472394453980793,
      "loss": 1.9922,
      "step": 3945
    },
    {
      "epoch": 2.613245033112583,
      "grad_norm": 0.7526853919703954,
      "learning_rate": 0.00024718861295612153,
      "loss": 2.1875,
      "step": 3946
    },
    {
      "epoch": 2.613907284768212,
      "grad_norm": 0.7771445276399125,
      "learning_rate": 0.00024713776126970415,
      "loss": 2.2344,
      "step": 3947
    },
    {
      "epoch": 2.614569536423841,
      "grad_norm": 0.7729366534333044,
      "learning_rate": 0.0002470868903488963,
      "loss": 2.0312,
      "step": 3948
    },
    {
      "epoch": 2.61523178807947,
      "grad_norm": 0.888943939173716,
      "learning_rate": 0.00024703600020377115,
      "loss": 2.1875,
      "step": 3949
    },
    {
      "epoch": 2.6158940397350996,
      "grad_norm": 0.7831434195196432,
      "learning_rate": 0.0002469850908444055,
      "loss": 2.3281,
      "step": 3950
    },
    {
      "epoch": 2.6165562913907285,
      "grad_norm": 0.7718073110457113,
      "learning_rate": 0.0002469341622808799,
      "loss": 1.8359,
      "step": 3951
    },
    {
      "epoch": 2.6172185430463575,
      "grad_norm": 0.8134867977303436,
      "learning_rate": 0.0002468832145232789,
      "loss": 2.0469,
      "step": 3952
    },
    {
      "epoch": 2.617880794701987,
      "grad_norm": 0.8340184999293999,
      "learning_rate": 0.00024683224758169066,
      "loss": 1.8594,
      "step": 3953
    },
    {
      "epoch": 2.618543046357616,
      "grad_norm": 0.8092752229261603,
      "learning_rate": 0.00024678126146620725,
      "loss": 2.4062,
      "step": 3954
    },
    {
      "epoch": 2.619205298013245,
      "grad_norm": 0.7871138882713193,
      "learning_rate": 0.0002467302561869246,
      "loss": 2.0781,
      "step": 3955
    },
    {
      "epoch": 2.619867549668874,
      "grad_norm": 0.8734304829454115,
      "learning_rate": 0.00024667923175394213,
      "loss": 2.4531,
      "step": 3956
    },
    {
      "epoch": 2.620529801324503,
      "grad_norm": 0.7617866558802842,
      "learning_rate": 0.0002466281881773635,
      "loss": 2.0781,
      "step": 3957
    },
    {
      "epoch": 2.6211920529801325,
      "grad_norm": 0.8235196479221422,
      "learning_rate": 0.0002465771254672957,
      "loss": 2.125,
      "step": 3958
    },
    {
      "epoch": 2.621854304635762,
      "grad_norm": 0.8494079559319179,
      "learning_rate": 0.0002465260436338499,
      "loss": 2.1094,
      "step": 3959
    },
    {
      "epoch": 2.622516556291391,
      "grad_norm": 0.7950793798216379,
      "learning_rate": 0.0002464749426871408,
      "loss": 2.0156,
      "step": 3960
    },
    {
      "epoch": 2.6231788079470197,
      "grad_norm": 0.8087438620222626,
      "learning_rate": 0.000246423822637287,
      "loss": 2.2969,
      "step": 3961
    },
    {
      "epoch": 2.623841059602649,
      "grad_norm": 0.7970314748320818,
      "learning_rate": 0.00024637268349441083,
      "loss": 2.25,
      "step": 3962
    },
    {
      "epoch": 2.624503311258278,
      "grad_norm": 0.7772945032674179,
      "learning_rate": 0.0002463215252686385,
      "loss": 2.0781,
      "step": 3963
    },
    {
      "epoch": 2.6251655629139075,
      "grad_norm": 0.7574067714493823,
      "learning_rate": 0.0002462703479700998,
      "loss": 2.1719,
      "step": 3964
    },
    {
      "epoch": 2.6258278145695364,
      "grad_norm": 0.8778551151112557,
      "learning_rate": 0.0002462191516089286,
      "loss": 2.2812,
      "step": 3965
    },
    {
      "epoch": 2.6264900662251653,
      "grad_norm": 0.7474688429951181,
      "learning_rate": 0.0002461679361952622,
      "loss": 1.7812,
      "step": 3966
    },
    {
      "epoch": 2.6271523178807947,
      "grad_norm": 0.9117482978222983,
      "learning_rate": 0.00024611670173924196,
      "loss": 2.4375,
      "step": 3967
    },
    {
      "epoch": 2.627814569536424,
      "grad_norm": 0.8099232061960885,
      "learning_rate": 0.0002460654482510128,
      "loss": 2.2344,
      "step": 3968
    },
    {
      "epoch": 2.628476821192053,
      "grad_norm": 0.7650339671617369,
      "learning_rate": 0.00024601417574072353,
      "loss": 2.0312,
      "step": 3969
    },
    {
      "epoch": 2.629139072847682,
      "grad_norm": 0.7350188181502101,
      "learning_rate": 0.0002459628842185267,
      "loss": 2.1406,
      "step": 3970
    },
    {
      "epoch": 2.6298013245033114,
      "grad_norm": 0.7883223074748857,
      "learning_rate": 0.0002459115736945786,
      "loss": 2.0469,
      "step": 3971
    },
    {
      "epoch": 2.6304635761589403,
      "grad_norm": 0.9102055676408702,
      "learning_rate": 0.00024586024417903933,
      "loss": 2.3906,
      "step": 3972
    },
    {
      "epoch": 2.6311258278145697,
      "grad_norm": 0.8043078862466365,
      "learning_rate": 0.0002458088956820727,
      "loss": 2.2188,
      "step": 3973
    },
    {
      "epoch": 2.6317880794701987,
      "grad_norm": 0.8589414189521787,
      "learning_rate": 0.00024575752821384635,
      "loss": 2.1562,
      "step": 3974
    },
    {
      "epoch": 2.6324503311258276,
      "grad_norm": 0.7455054163511606,
      "learning_rate": 0.00024570614178453154,
      "loss": 1.9141,
      "step": 3975
    },
    {
      "epoch": 2.633112582781457,
      "grad_norm": 0.8856875005289148,
      "learning_rate": 0.00024565473640430347,
      "loss": 2.4688,
      "step": 3976
    },
    {
      "epoch": 2.633774834437086,
      "grad_norm": 0.7536702614283786,
      "learning_rate": 0.0002456033120833409,
      "loss": 2.0156,
      "step": 3977
    },
    {
      "epoch": 2.6344370860927153,
      "grad_norm": 0.7445244313387426,
      "learning_rate": 0.00024555186883182646,
      "loss": 2.0938,
      "step": 3978
    },
    {
      "epoch": 2.6350993377483443,
      "grad_norm": 0.8744795133062511,
      "learning_rate": 0.00024550040665994655,
      "loss": 2.2031,
      "step": 3979
    },
    {
      "epoch": 2.6357615894039736,
      "grad_norm": 0.7934506137321837,
      "learning_rate": 0.0002454489255778913,
      "loss": 2.3125,
      "step": 3980
    },
    {
      "epoch": 2.6364238410596026,
      "grad_norm": 0.8092477343568119,
      "learning_rate": 0.00024539742559585447,
      "loss": 2.0781,
      "step": 3981
    },
    {
      "epoch": 2.637086092715232,
      "grad_norm": 0.79372374669347,
      "learning_rate": 0.00024534590672403364,
      "loss": 2.0312,
      "step": 3982
    },
    {
      "epoch": 2.637748344370861,
      "grad_norm": 0.8424438035006498,
      "learning_rate": 0.00024529436897263024,
      "loss": 2.3438,
      "step": 3983
    },
    {
      "epoch": 2.63841059602649,
      "grad_norm": 0.7398599092464678,
      "learning_rate": 0.0002452428123518492,
      "loss": 1.9531,
      "step": 3984
    },
    {
      "epoch": 2.6390728476821192,
      "grad_norm": 0.8203314886054506,
      "learning_rate": 0.00024519123687189945,
      "loss": 2.0,
      "step": 3985
    },
    {
      "epoch": 2.639735099337748,
      "grad_norm": 2.3659414962017173,
      "learning_rate": 0.0002451396425429934,
      "loss": 2.1094,
      "step": 3986
    },
    {
      "epoch": 2.6403973509933776,
      "grad_norm": 0.7733714356852375,
      "learning_rate": 0.00024508802937534745,
      "loss": 2.0938,
      "step": 3987
    },
    {
      "epoch": 2.6410596026490065,
      "grad_norm": 0.8700694156492562,
      "learning_rate": 0.0002450363973791815,
      "loss": 2.1875,
      "step": 3988
    },
    {
      "epoch": 2.641721854304636,
      "grad_norm": 0.8500432520360289,
      "learning_rate": 0.0002449847465647193,
      "loss": 2.1719,
      "step": 3989
    },
    {
      "epoch": 2.642384105960265,
      "grad_norm": 0.9578931012029728,
      "learning_rate": 0.0002449330769421884,
      "loss": 2.3125,
      "step": 3990
    },
    {
      "epoch": 2.6430463576158942,
      "grad_norm": 0.8358342069810146,
      "learning_rate": 0.00024488138852181977,
      "loss": 2.3438,
      "step": 3991
    },
    {
      "epoch": 2.643708609271523,
      "grad_norm": 0.8090430422725822,
      "learning_rate": 0.00024482968131384845,
      "loss": 2.2344,
      "step": 3992
    },
    {
      "epoch": 2.644370860927152,
      "grad_norm": 0.8180062559859658,
      "learning_rate": 0.00024477795532851313,
      "loss": 2.2656,
      "step": 3993
    },
    {
      "epoch": 2.6450331125827815,
      "grad_norm": 0.8476654346437353,
      "learning_rate": 0.000244726210576056,
      "loss": 2.2188,
      "step": 3994
    },
    {
      "epoch": 2.6456953642384105,
      "grad_norm": 0.7187959212732824,
      "learning_rate": 0.0002446744470667232,
      "loss": 1.6953,
      "step": 3995
    },
    {
      "epoch": 2.64635761589404,
      "grad_norm": 0.8217597341706413,
      "learning_rate": 0.0002446226648107645,
      "loss": 1.7734,
      "step": 3996
    },
    {
      "epoch": 2.647019867549669,
      "grad_norm": 0.8090516019370594,
      "learning_rate": 0.0002445708638184333,
      "loss": 2.1406,
      "step": 3997
    },
    {
      "epoch": 2.647682119205298,
      "grad_norm": 0.774686018576076,
      "learning_rate": 0.00024451904409998685,
      "loss": 1.9219,
      "step": 3998
    },
    {
      "epoch": 2.648344370860927,
      "grad_norm": 0.7417660107897827,
      "learning_rate": 0.00024446720566568604,
      "loss": 2.0312,
      "step": 3999
    },
    {
      "epoch": 2.6490066225165565,
      "grad_norm": 0.8030055785026562,
      "learning_rate": 0.0002444153485257955,
      "loss": 2.3125,
      "step": 4000
    },
    {
      "epoch": 2.6496688741721854,
      "grad_norm": 0.7537488544454412,
      "learning_rate": 0.0002443634726905835,
      "loss": 2.0625,
      "step": 4001
    },
    {
      "epoch": 2.6503311258278144,
      "grad_norm": 0.7683872660779115,
      "learning_rate": 0.0002443115781703221,
      "loss": 2.0938,
      "step": 4002
    },
    {
      "epoch": 2.6509933774834438,
      "grad_norm": 0.8186140957195706,
      "learning_rate": 0.00024425966497528695,
      "loss": 2.3125,
      "step": 4003
    },
    {
      "epoch": 2.6516556291390727,
      "grad_norm": 0.8579295748754989,
      "learning_rate": 0.0002442077331157575,
      "loss": 2.3281,
      "step": 4004
    },
    {
      "epoch": 2.652317880794702,
      "grad_norm": 0.8304913858395438,
      "learning_rate": 0.0002441557826020168,
      "loss": 2.2344,
      "step": 4005
    },
    {
      "epoch": 2.652980132450331,
      "grad_norm": 0.7733390407939369,
      "learning_rate": 0.00024410381344435172,
      "loss": 1.8516,
      "step": 4006
    },
    {
      "epoch": 2.6536423841059604,
      "grad_norm": 0.748787941146656,
      "learning_rate": 0.00024405182565305267,
      "loss": 1.7969,
      "step": 4007
    },
    {
      "epoch": 2.6543046357615894,
      "grad_norm": 0.7143804740604952,
      "learning_rate": 0.00024399981923841393,
      "loss": 1.9453,
      "step": 4008
    },
    {
      "epoch": 2.6549668874172188,
      "grad_norm": 0.8282545433536421,
      "learning_rate": 0.00024394779421073326,
      "loss": 2.0781,
      "step": 4009
    },
    {
      "epoch": 2.6556291390728477,
      "grad_norm": 0.7340440023406857,
      "learning_rate": 0.00024389575058031225,
      "loss": 2.0781,
      "step": 4010
    },
    {
      "epoch": 2.6562913907284766,
      "grad_norm": 0.7749512111393109,
      "learning_rate": 0.00024384368835745615,
      "loss": 2.2656,
      "step": 4011
    },
    {
      "epoch": 2.656953642384106,
      "grad_norm": 0.8580400754296054,
      "learning_rate": 0.00024379160755247392,
      "loss": 2.5312,
      "step": 4012
    },
    {
      "epoch": 2.657615894039735,
      "grad_norm": 0.8333838902818076,
      "learning_rate": 0.0002437395081756781,
      "loss": 2.4062,
      "step": 4013
    },
    {
      "epoch": 2.6582781456953644,
      "grad_norm": 0.8579084646959299,
      "learning_rate": 0.00024368739023738494,
      "loss": 2.3438,
      "step": 4014
    },
    {
      "epoch": 2.6589403973509933,
      "grad_norm": 0.7854927141630277,
      "learning_rate": 0.00024363525374791444,
      "loss": 2.2656,
      "step": 4015
    },
    {
      "epoch": 2.6596026490066222,
      "grad_norm": 0.7361428925110682,
      "learning_rate": 0.0002435830987175902,
      "loss": 2.1094,
      "step": 4016
    },
    {
      "epoch": 2.6602649006622516,
      "grad_norm": 0.7222168902269966,
      "learning_rate": 0.00024353092515673951,
      "loss": 2.1094,
      "step": 4017
    },
    {
      "epoch": 2.660927152317881,
      "grad_norm": 0.7049282195647464,
      "learning_rate": 0.00024347873307569334,
      "loss": 2.2812,
      "step": 4018
    },
    {
      "epoch": 2.66158940397351,
      "grad_norm": 0.780329558805766,
      "learning_rate": 0.00024342652248478635,
      "loss": 2.1094,
      "step": 4019
    },
    {
      "epoch": 2.662251655629139,
      "grad_norm": 0.7108235463720955,
      "learning_rate": 0.0002433742933943568,
      "loss": 1.7812,
      "step": 4020
    },
    {
      "epoch": 2.6629139072847683,
      "grad_norm": 0.7431386827151667,
      "learning_rate": 0.00024332204581474664,
      "loss": 2.2969,
      "step": 4021
    },
    {
      "epoch": 2.6635761589403972,
      "grad_norm": 0.827379805174418,
      "learning_rate": 0.00024326977975630152,
      "loss": 2.1875,
      "step": 4022
    },
    {
      "epoch": 2.6642384105960266,
      "grad_norm": 0.7440660791752031,
      "learning_rate": 0.0002432174952293707,
      "loss": 1.9609,
      "step": 4023
    },
    {
      "epoch": 2.6649006622516556,
      "grad_norm": 0.756484251795818,
      "learning_rate": 0.0002431651922443071,
      "loss": 2.2031,
      "step": 4024
    },
    {
      "epoch": 2.6655629139072845,
      "grad_norm": 0.8497077727828823,
      "learning_rate": 0.0002431128708114674,
      "loss": 2.3125,
      "step": 4025
    },
    {
      "epoch": 2.666225165562914,
      "grad_norm": 0.8406128928214927,
      "learning_rate": 0.00024306053094121173,
      "loss": 2.3594,
      "step": 4026
    },
    {
      "epoch": 2.6668874172185433,
      "grad_norm": 0.7341871201159577,
      "learning_rate": 0.00024300817264390402,
      "loss": 2.0938,
      "step": 4027
    },
    {
      "epoch": 2.667549668874172,
      "grad_norm": 0.8009228829165581,
      "learning_rate": 0.00024295579592991182,
      "loss": 2.1562,
      "step": 4028
    },
    {
      "epoch": 2.668211920529801,
      "grad_norm": 0.7767802387545462,
      "learning_rate": 0.00024290340080960633,
      "loss": 2.4219,
      "step": 4029
    },
    {
      "epoch": 2.6688741721854305,
      "grad_norm": 0.7821838881125265,
      "learning_rate": 0.00024285098729336237,
      "loss": 2.2031,
      "step": 4030
    },
    {
      "epoch": 2.6695364238410595,
      "grad_norm": 0.7676080396302879,
      "learning_rate": 0.00024279855539155842,
      "loss": 2.1875,
      "step": 4031
    },
    {
      "epoch": 2.670198675496689,
      "grad_norm": 0.7184472072671353,
      "learning_rate": 0.00024274610511457664,
      "loss": 1.7891,
      "step": 4032
    },
    {
      "epoch": 2.670860927152318,
      "grad_norm": 0.7743859240278532,
      "learning_rate": 0.0002426936364728027,
      "loss": 2.2031,
      "step": 4033
    },
    {
      "epoch": 2.6715231788079468,
      "grad_norm": 0.7921646708238572,
      "learning_rate": 0.00024264114947662607,
      "loss": 2.0781,
      "step": 4034
    },
    {
      "epoch": 2.672185430463576,
      "grad_norm": 0.7675419751042976,
      "learning_rate": 0.00024258864413643976,
      "loss": 1.9297,
      "step": 4035
    },
    {
      "epoch": 2.6728476821192055,
      "grad_norm": 0.7522121126870421,
      "learning_rate": 0.00024253612046264042,
      "loss": 1.8438,
      "step": 4036
    },
    {
      "epoch": 2.6735099337748345,
      "grad_norm": 0.7766544368418868,
      "learning_rate": 0.00024248357846562828,
      "loss": 1.8047,
      "step": 4037
    },
    {
      "epoch": 2.6741721854304634,
      "grad_norm": 0.8508738094951402,
      "learning_rate": 0.00024243101815580735,
      "loss": 2.375,
      "step": 4038
    },
    {
      "epoch": 2.674834437086093,
      "grad_norm": 0.8154811171532356,
      "learning_rate": 0.00024237843954358514,
      "loss": 2.125,
      "step": 4039
    },
    {
      "epoch": 2.6754966887417218,
      "grad_norm": 0.8250995363598963,
      "learning_rate": 0.00024232584263937282,
      "loss": 2.0938,
      "step": 4040
    },
    {
      "epoch": 2.676158940397351,
      "grad_norm": 0.7586389656041453,
      "learning_rate": 0.00024227322745358517,
      "loss": 2.0781,
      "step": 4041
    },
    {
      "epoch": 2.67682119205298,
      "grad_norm": 0.7548252587657602,
      "learning_rate": 0.00024222059399664063,
      "loss": 2.0312,
      "step": 4042
    },
    {
      "epoch": 2.677483443708609,
      "grad_norm": 0.8453310966408427,
      "learning_rate": 0.00024216794227896113,
      "loss": 2.0781,
      "step": 4043
    },
    {
      "epoch": 2.6781456953642384,
      "grad_norm": 0.7950819885686008,
      "learning_rate": 0.00024211527231097243,
      "loss": 2.1719,
      "step": 4044
    },
    {
      "epoch": 2.678807947019868,
      "grad_norm": 0.9130463106753197,
      "learning_rate": 0.0002420625841031038,
      "loss": 2.3438,
      "step": 4045
    },
    {
      "epoch": 2.6794701986754967,
      "grad_norm": 0.8087856731625019,
      "learning_rate": 0.00024200987766578798,
      "loss": 2.2031,
      "step": 4046
    },
    {
      "epoch": 2.6801324503311257,
      "grad_norm": 0.7543452574716366,
      "learning_rate": 0.00024195715300946152,
      "loss": 2.0312,
      "step": 4047
    },
    {
      "epoch": 2.680794701986755,
      "grad_norm": 0.7674752929308315,
      "learning_rate": 0.00024190441014456459,
      "loss": 2.0469,
      "step": 4048
    },
    {
      "epoch": 2.681456953642384,
      "grad_norm": 0.8765161929005814,
      "learning_rate": 0.0002418516490815407,
      "loss": 2.3438,
      "step": 4049
    },
    {
      "epoch": 2.6821192052980134,
      "grad_norm": 0.7862718953691918,
      "learning_rate": 0.00024179886983083734,
      "loss": 2.3125,
      "step": 4050
    },
    {
      "epoch": 2.6827814569536423,
      "grad_norm": 0.8337623188986159,
      "learning_rate": 0.00024174607240290524,
      "loss": 2.2812,
      "step": 4051
    },
    {
      "epoch": 2.6834437086092713,
      "grad_norm": 0.855831125107986,
      "learning_rate": 0.000241693256808199,
      "loss": 2.25,
      "step": 4052
    },
    {
      "epoch": 2.6841059602649007,
      "grad_norm": 0.8105043828491679,
      "learning_rate": 0.0002416404230571767,
      "loss": 2.2031,
      "step": 4053
    },
    {
      "epoch": 2.68476821192053,
      "grad_norm": 0.747480014295251,
      "learning_rate": 0.0002415875711603,
      "loss": 1.9766,
      "step": 4054
    },
    {
      "epoch": 2.685430463576159,
      "grad_norm": 0.7977605909946265,
      "learning_rate": 0.0002415347011280342,
      "loss": 2.3281,
      "step": 4055
    },
    {
      "epoch": 2.686092715231788,
      "grad_norm": 0.8573859371801149,
      "learning_rate": 0.00024148181297084815,
      "loss": 2.4688,
      "step": 4056
    },
    {
      "epoch": 2.6867549668874173,
      "grad_norm": 0.7949748565568895,
      "learning_rate": 0.00024142890669921432,
      "loss": 2.0312,
      "step": 4057
    },
    {
      "epoch": 2.6874172185430463,
      "grad_norm": 0.7581592734645108,
      "learning_rate": 0.0002413759823236088,
      "loss": 2.0938,
      "step": 4058
    },
    {
      "epoch": 2.6880794701986757,
      "grad_norm": 0.7348714802604301,
      "learning_rate": 0.00024132303985451115,
      "loss": 1.9453,
      "step": 4059
    },
    {
      "epoch": 2.6887417218543046,
      "grad_norm": 0.8397994809086442,
      "learning_rate": 0.00024127007930240462,
      "loss": 1.8906,
      "step": 4060
    },
    {
      "epoch": 2.6894039735099335,
      "grad_norm": 0.7951997520906008,
      "learning_rate": 0.00024121710067777604,
      "loss": 2.2031,
      "step": 4061
    },
    {
      "epoch": 2.690066225165563,
      "grad_norm": 0.7796389670076017,
      "learning_rate": 0.00024116410399111575,
      "loss": 2.0156,
      "step": 4062
    },
    {
      "epoch": 2.6907284768211923,
      "grad_norm": 0.7922145430230525,
      "learning_rate": 0.00024111108925291774,
      "loss": 2.2656,
      "step": 4063
    },
    {
      "epoch": 2.6913907284768213,
      "grad_norm": 0.8331854757390216,
      "learning_rate": 0.00024105805647367946,
      "loss": 2.3906,
      "step": 4064
    },
    {
      "epoch": 2.69205298013245,
      "grad_norm": 0.7918863150750985,
      "learning_rate": 0.00024100500566390207,
      "loss": 2.2031,
      "step": 4065
    },
    {
      "epoch": 2.6927152317880796,
      "grad_norm": 0.752840711001402,
      "learning_rate": 0.0002409519368340902,
      "loss": 2.2969,
      "step": 4066
    },
    {
      "epoch": 2.6933774834437085,
      "grad_norm": 0.7106618693281918,
      "learning_rate": 0.0002408988499947521,
      "loss": 1.9141,
      "step": 4067
    },
    {
      "epoch": 2.694039735099338,
      "grad_norm": 0.7737054514426718,
      "learning_rate": 0.0002408457451563996,
      "loss": 2.0625,
      "step": 4068
    },
    {
      "epoch": 2.694701986754967,
      "grad_norm": 0.6913925395782106,
      "learning_rate": 0.00024079262232954804,
      "loss": 1.8516,
      "step": 4069
    },
    {
      "epoch": 2.695364238410596,
      "grad_norm": 0.7366946121495257,
      "learning_rate": 0.00024073948152471633,
      "loss": 2.1875,
      "step": 4070
    },
    {
      "epoch": 2.696026490066225,
      "grad_norm": 0.7768428160721613,
      "learning_rate": 0.00024068632275242707,
      "loss": 2.3594,
      "step": 4071
    },
    {
      "epoch": 2.6966887417218546,
      "grad_norm": 0.7396938121941722,
      "learning_rate": 0.00024063314602320614,
      "loss": 1.9922,
      "step": 4072
    },
    {
      "epoch": 2.6973509933774835,
      "grad_norm": 0.7397235321847527,
      "learning_rate": 0.00024057995134758325,
      "loss": 2.0781,
      "step": 4073
    },
    {
      "epoch": 2.6980132450331125,
      "grad_norm": 0.7385817568598578,
      "learning_rate": 0.00024052673873609155,
      "loss": 2.2031,
      "step": 4074
    },
    {
      "epoch": 2.698675496688742,
      "grad_norm": 0.7766812521645137,
      "learning_rate": 0.00024047350819926766,
      "loss": 2.0625,
      "step": 4075
    },
    {
      "epoch": 2.699337748344371,
      "grad_norm": 0.7229966825822027,
      "learning_rate": 0.00024042025974765194,
      "loss": 1.8125,
      "step": 4076
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.7041930470206346,
      "learning_rate": 0.00024036699339178815,
      "loss": 1.8594,
      "step": 4077
    },
    {
      "epoch": 2.700662251655629,
      "grad_norm": 0.747757219408075,
      "learning_rate": 0.00024031370914222365,
      "loss": 2.2031,
      "step": 4078
    },
    {
      "epoch": 2.701324503311258,
      "grad_norm": 0.7295858143041319,
      "learning_rate": 0.00024026040700950933,
      "loss": 1.9062,
      "step": 4079
    },
    {
      "epoch": 2.7019867549668874,
      "grad_norm": 0.8448136950281003,
      "learning_rate": 0.00024020708700419962,
      "loss": 2.0938,
      "step": 4080
    },
    {
      "epoch": 2.7026490066225164,
      "grad_norm": 0.8024408689093352,
      "learning_rate": 0.0002401537491368525,
      "loss": 2.0469,
      "step": 4081
    },
    {
      "epoch": 2.703311258278146,
      "grad_norm": 0.777671545450392,
      "learning_rate": 0.00024010039341802945,
      "loss": 2.1562,
      "step": 4082
    },
    {
      "epoch": 2.7039735099337747,
      "grad_norm": 0.7864116301787656,
      "learning_rate": 0.00024004701985829555,
      "loss": 2.375,
      "step": 4083
    },
    {
      "epoch": 2.704635761589404,
      "grad_norm": 0.820749465471322,
      "learning_rate": 0.00023999362846821934,
      "loss": 2.2031,
      "step": 4084
    },
    {
      "epoch": 2.705298013245033,
      "grad_norm": 0.6788171367973583,
      "learning_rate": 0.000239940219258373,
      "loss": 1.9141,
      "step": 4085
    },
    {
      "epoch": 2.7059602649006624,
      "grad_norm": 0.7762608211396803,
      "learning_rate": 0.00023988679223933205,
      "loss": 1.9141,
      "step": 4086
    },
    {
      "epoch": 2.7066225165562914,
      "grad_norm": 0.7785415522662342,
      "learning_rate": 0.00023983334742167577,
      "loss": 2.2188,
      "step": 4087
    },
    {
      "epoch": 2.7072847682119203,
      "grad_norm": 0.7513878047620702,
      "learning_rate": 0.00023977988481598675,
      "loss": 2.0781,
      "step": 4088
    },
    {
      "epoch": 2.7079470198675497,
      "grad_norm": 0.8292747168228135,
      "learning_rate": 0.0002397264044328513,
      "loss": 2.2188,
      "step": 4089
    },
    {
      "epoch": 2.7086092715231787,
      "grad_norm": 0.7144389068603944,
      "learning_rate": 0.00023967290628285908,
      "loss": 2.1094,
      "step": 4090
    },
    {
      "epoch": 2.709271523178808,
      "grad_norm": 0.7015301683643281,
      "learning_rate": 0.00023961939037660333,
      "loss": 1.7578,
      "step": 4091
    },
    {
      "epoch": 2.709933774834437,
      "grad_norm": 0.7604320477329195,
      "learning_rate": 0.0002395658567246808,
      "loss": 2.2812,
      "step": 4092
    },
    {
      "epoch": 2.7105960264900664,
      "grad_norm": 0.6960850765348368,
      "learning_rate": 0.00023951230533769186,
      "loss": 1.9844,
      "step": 4093
    },
    {
      "epoch": 2.7112582781456953,
      "grad_norm": 0.7404614022994347,
      "learning_rate": 0.00023945873622624022,
      "loss": 2.1094,
      "step": 4094
    },
    {
      "epoch": 2.7119205298013247,
      "grad_norm": 0.7436867083058829,
      "learning_rate": 0.00023940514940093318,
      "loss": 1.9141,
      "step": 4095
    },
    {
      "epoch": 2.7125827814569536,
      "grad_norm": 0.7693258187055638,
      "learning_rate": 0.0002393515448723816,
      "loss": 2.1875,
      "step": 4096
    },
    {
      "epoch": 2.7132450331125826,
      "grad_norm": 0.7990908124917558,
      "learning_rate": 0.00023929792265119971,
      "loss": 2.0781,
      "step": 4097
    },
    {
      "epoch": 2.713907284768212,
      "grad_norm": 0.8083064697611291,
      "learning_rate": 0.00023924428274800538,
      "loss": 2.0938,
      "step": 4098
    },
    {
      "epoch": 2.714569536423841,
      "grad_norm": 0.749556259702483,
      "learning_rate": 0.00023919062517341991,
      "loss": 2.2031,
      "step": 4099
    },
    {
      "epoch": 2.7152317880794703,
      "grad_norm": 0.8235749489946774,
      "learning_rate": 0.00023913694993806813,
      "loss": 2.2031,
      "step": 4100
    },
    {
      "epoch": 2.7158940397350992,
      "grad_norm": 0.8172201680805598,
      "learning_rate": 0.00023908325705257835,
      "loss": 2.1875,
      "step": 4101
    },
    {
      "epoch": 2.7165562913907286,
      "grad_norm": 0.7881146108637637,
      "learning_rate": 0.00023902954652758234,
      "loss": 2.1406,
      "step": 4102
    },
    {
      "epoch": 2.7172185430463576,
      "grad_norm": 0.7826072574917247,
      "learning_rate": 0.00023897581837371547,
      "loss": 2.1875,
      "step": 4103
    },
    {
      "epoch": 2.717880794701987,
      "grad_norm": 0.768843424187108,
      "learning_rate": 0.00023892207260161645,
      "loss": 2.0938,
      "step": 4104
    },
    {
      "epoch": 2.718543046357616,
      "grad_norm": 0.7755798806671983,
      "learning_rate": 0.00023886830922192757,
      "loss": 2.0469,
      "step": 4105
    },
    {
      "epoch": 2.719205298013245,
      "grad_norm": 0.7965834623867012,
      "learning_rate": 0.00023881452824529468,
      "loss": 2.1406,
      "step": 4106
    },
    {
      "epoch": 2.7198675496688742,
      "grad_norm": 0.7628454897438512,
      "learning_rate": 0.00023876072968236694,
      "loss": 2.0,
      "step": 4107
    },
    {
      "epoch": 2.720529801324503,
      "grad_norm": 0.8022246004686868,
      "learning_rate": 0.00023870691354379713,
      "loss": 2.375,
      "step": 4108
    },
    {
      "epoch": 2.7211920529801326,
      "grad_norm": 0.8422897363595795,
      "learning_rate": 0.00023865307984024145,
      "loss": 2.1875,
      "step": 4109
    },
    {
      "epoch": 2.7218543046357615,
      "grad_norm": 0.768094710039755,
      "learning_rate": 0.0002385992285823596,
      "loss": 2.0469,
      "step": 4110
    },
    {
      "epoch": 2.722516556291391,
      "grad_norm": 0.7653191461578709,
      "learning_rate": 0.00023854535978081465,
      "loss": 2.0,
      "step": 4111
    },
    {
      "epoch": 2.72317880794702,
      "grad_norm": 0.8050306173241116,
      "learning_rate": 0.00023849147344627341,
      "loss": 2.3594,
      "step": 4112
    },
    {
      "epoch": 2.723841059602649,
      "grad_norm": 0.7611514146372146,
      "learning_rate": 0.00023843756958940584,
      "loss": 1.9609,
      "step": 4113
    },
    {
      "epoch": 2.724503311258278,
      "grad_norm": 0.7287454424721603,
      "learning_rate": 0.00023838364822088566,
      "loss": 2.0,
      "step": 4114
    },
    {
      "epoch": 2.725165562913907,
      "grad_norm": 0.8622139876781421,
      "learning_rate": 0.00023832970935138984,
      "loss": 2.2344,
      "step": 4115
    },
    {
      "epoch": 2.7258278145695365,
      "grad_norm": 0.7317501265036545,
      "learning_rate": 0.00023827575299159885,
      "loss": 1.8281,
      "step": 4116
    },
    {
      "epoch": 2.7264900662251654,
      "grad_norm": 0.7808965905253575,
      "learning_rate": 0.00023822177915219675,
      "loss": 2.2344,
      "step": 4117
    },
    {
      "epoch": 2.727152317880795,
      "grad_norm": 0.8252352185844266,
      "learning_rate": 0.00023816778784387094,
      "loss": 2.3125,
      "step": 4118
    },
    {
      "epoch": 2.7278145695364238,
      "grad_norm": 0.8291781239868556,
      "learning_rate": 0.00023811377907731235,
      "loss": 2.2812,
      "step": 4119
    },
    {
      "epoch": 2.7284768211920527,
      "grad_norm": 0.8874748026170409,
      "learning_rate": 0.0002380597528632153,
      "loss": 2.4062,
      "step": 4120
    },
    {
      "epoch": 2.729139072847682,
      "grad_norm": 0.6895970878053972,
      "learning_rate": 0.0002380057092122776,
      "loss": 1.6328,
      "step": 4121
    },
    {
      "epoch": 2.7298013245033115,
      "grad_norm": 0.7610437522268005,
      "learning_rate": 0.00023795164813520054,
      "loss": 2.0938,
      "step": 4122
    },
    {
      "epoch": 2.7304635761589404,
      "grad_norm": 0.8267448102852251,
      "learning_rate": 0.00023789756964268875,
      "loss": 2.3125,
      "step": 4123
    },
    {
      "epoch": 2.7311258278145694,
      "grad_norm": 0.8067400875975469,
      "learning_rate": 0.00023784347374545056,
      "loss": 2.2031,
      "step": 4124
    },
    {
      "epoch": 2.7317880794701987,
      "grad_norm": 0.7904023176339622,
      "learning_rate": 0.00023778936045419738,
      "loss": 2.125,
      "step": 4125
    },
    {
      "epoch": 2.7324503311258277,
      "grad_norm": 0.800327730970933,
      "learning_rate": 0.00023773522977964446,
      "loss": 2.2969,
      "step": 4126
    },
    {
      "epoch": 2.733112582781457,
      "grad_norm": 0.7622801319974138,
      "learning_rate": 0.0002376810817325101,
      "loss": 2.0938,
      "step": 4127
    },
    {
      "epoch": 2.733774834437086,
      "grad_norm": 0.7460122558919458,
      "learning_rate": 0.00023762691632351636,
      "loss": 1.7969,
      "step": 4128
    },
    {
      "epoch": 2.734437086092715,
      "grad_norm": 0.7928885464972585,
      "learning_rate": 0.0002375727335633886,
      "loss": 2.0156,
      "step": 4129
    },
    {
      "epoch": 2.7350993377483444,
      "grad_norm": 0.8190237958611799,
      "learning_rate": 0.00023751853346285558,
      "loss": 2.1094,
      "step": 4130
    },
    {
      "epoch": 2.7357615894039737,
      "grad_norm": 0.7659991012713004,
      "learning_rate": 0.00023746431603264954,
      "loss": 2.0625,
      "step": 4131
    },
    {
      "epoch": 2.7364238410596027,
      "grad_norm": 0.8855012206324189,
      "learning_rate": 0.00023741008128350626,
      "loss": 2.1719,
      "step": 4132
    },
    {
      "epoch": 2.7370860927152316,
      "grad_norm": 0.8167691295429508,
      "learning_rate": 0.00023735582922616472,
      "loss": 2.1094,
      "step": 4133
    },
    {
      "epoch": 2.737748344370861,
      "grad_norm": 0.8057211479065577,
      "learning_rate": 0.00023730155987136743,
      "loss": 2.1406,
      "step": 4134
    },
    {
      "epoch": 2.73841059602649,
      "grad_norm": 0.7788521888088263,
      "learning_rate": 0.00023724727322986048,
      "loss": 2.0312,
      "step": 4135
    },
    {
      "epoch": 2.7390728476821193,
      "grad_norm": 0.8019006083722068,
      "learning_rate": 0.00023719296931239316,
      "loss": 2.25,
      "step": 4136
    },
    {
      "epoch": 2.7397350993377483,
      "grad_norm": 0.7612376619734238,
      "learning_rate": 0.00023713864812971826,
      "loss": 2.1094,
      "step": 4137
    },
    {
      "epoch": 2.7403973509933772,
      "grad_norm": 0.7900531342585587,
      "learning_rate": 0.00023708430969259206,
      "loss": 1.9766,
      "step": 4138
    },
    {
      "epoch": 2.7410596026490066,
      "grad_norm": 0.8022937409151758,
      "learning_rate": 0.00023702995401177413,
      "loss": 2.2812,
      "step": 4139
    },
    {
      "epoch": 2.741721854304636,
      "grad_norm": 0.7681654125560227,
      "learning_rate": 0.00023697558109802753,
      "loss": 1.9766,
      "step": 4140
    },
    {
      "epoch": 2.742384105960265,
      "grad_norm": 0.8203828712682121,
      "learning_rate": 0.00023692119096211872,
      "loss": 2.4062,
      "step": 4141
    },
    {
      "epoch": 2.743046357615894,
      "grad_norm": 0.7952976963611031,
      "learning_rate": 0.0002368667836148176,
      "loss": 1.9844,
      "step": 4142
    },
    {
      "epoch": 2.7437086092715233,
      "grad_norm": 0.7979788039837987,
      "learning_rate": 0.00023681235906689743,
      "loss": 2.0625,
      "step": 4143
    },
    {
      "epoch": 2.744370860927152,
      "grad_norm": 0.7770300980122986,
      "learning_rate": 0.00023675791732913492,
      "loss": 2.125,
      "step": 4144
    },
    {
      "epoch": 2.7450331125827816,
      "grad_norm": 0.825271057452493,
      "learning_rate": 0.0002367034584123101,
      "loss": 2.1875,
      "step": 4145
    },
    {
      "epoch": 2.7456953642384105,
      "grad_norm": 0.8546449864986183,
      "learning_rate": 0.00023664898232720645,
      "loss": 2.2344,
      "step": 4146
    },
    {
      "epoch": 2.7463576158940395,
      "grad_norm": 0.9085639667267498,
      "learning_rate": 0.00023659448908461096,
      "loss": 2.2969,
      "step": 4147
    },
    {
      "epoch": 2.747019867549669,
      "grad_norm": 0.8099066864178545,
      "learning_rate": 0.00023653997869531383,
      "loss": 2.1406,
      "step": 4148
    },
    {
      "epoch": 2.7476821192052983,
      "grad_norm": 0.7199504634925541,
      "learning_rate": 0.00023648545117010882,
      "loss": 2.0938,
      "step": 4149
    },
    {
      "epoch": 2.748344370860927,
      "grad_norm": 0.8373375348999462,
      "learning_rate": 0.0002364309065197929,
      "loss": 2.125,
      "step": 4150
    },
    {
      "epoch": 2.749006622516556,
      "grad_norm": 0.8153055606046666,
      "learning_rate": 0.0002363763447551666,
      "loss": 2.0156,
      "step": 4151
    },
    {
      "epoch": 2.7496688741721855,
      "grad_norm": 0.6951663824441948,
      "learning_rate": 0.0002363217658870338,
      "loss": 1.5703,
      "step": 4152
    },
    {
      "epoch": 2.7503311258278145,
      "grad_norm": 0.7829422799569432,
      "learning_rate": 0.00023626716992620166,
      "loss": 2.0469,
      "step": 4153
    },
    {
      "epoch": 2.750993377483444,
      "grad_norm": 0.8518814257582603,
      "learning_rate": 0.0002362125568834809,
      "loss": 2.1094,
      "step": 4154
    },
    {
      "epoch": 2.751655629139073,
      "grad_norm": 0.8401338197097309,
      "learning_rate": 0.00023615792676968546,
      "loss": 2.1094,
      "step": 4155
    },
    {
      "epoch": 2.7523178807947017,
      "grad_norm": 0.731568523519124,
      "learning_rate": 0.00023610327959563276,
      "loss": 1.7188,
      "step": 4156
    },
    {
      "epoch": 2.752980132450331,
      "grad_norm": 0.7103965737552937,
      "learning_rate": 0.00023604861537214353,
      "loss": 1.7734,
      "step": 4157
    },
    {
      "epoch": 2.7536423841059605,
      "grad_norm": 0.7834559987829549,
      "learning_rate": 0.00023599393411004196,
      "loss": 2.125,
      "step": 4158
    },
    {
      "epoch": 2.7543046357615895,
      "grad_norm": 0.7904140165785916,
      "learning_rate": 0.00023593923582015554,
      "loss": 2.125,
      "step": 4159
    },
    {
      "epoch": 2.7549668874172184,
      "grad_norm": 0.8115993772466429,
      "learning_rate": 0.00023588452051331516,
      "loss": 2.2812,
      "step": 4160
    },
    {
      "epoch": 2.755629139072848,
      "grad_norm": 0.8571107663303452,
      "learning_rate": 0.0002358297882003551,
      "loss": 2.3125,
      "step": 4161
    },
    {
      "epoch": 2.7562913907284767,
      "grad_norm": 0.7036030250550355,
      "learning_rate": 0.00023577503889211294,
      "loss": 1.6172,
      "step": 4162
    },
    {
      "epoch": 2.756953642384106,
      "grad_norm": 0.7058883361115429,
      "learning_rate": 0.00023572027259942976,
      "loss": 1.6328,
      "step": 4163
    },
    {
      "epoch": 2.757615894039735,
      "grad_norm": 0.7823379611647142,
      "learning_rate": 0.00023566548933314985,
      "loss": 2.0156,
      "step": 4164
    },
    {
      "epoch": 2.758278145695364,
      "grad_norm": 0.805409006269033,
      "learning_rate": 0.00023561068910412088,
      "loss": 2.0781,
      "step": 4165
    },
    {
      "epoch": 2.7589403973509934,
      "grad_norm": 0.8134100214667909,
      "learning_rate": 0.00023555587192319397,
      "loss": 2.3125,
      "step": 4166
    },
    {
      "epoch": 2.7596026490066228,
      "grad_norm": 0.8046440179290408,
      "learning_rate": 0.00023550103780122357,
      "loss": 2.0312,
      "step": 4167
    },
    {
      "epoch": 2.7602649006622517,
      "grad_norm": 0.7702042433966213,
      "learning_rate": 0.00023544618674906749,
      "loss": 2.1875,
      "step": 4168
    },
    {
      "epoch": 2.7609271523178807,
      "grad_norm": 0.7530132771734991,
      "learning_rate": 0.00023539131877758676,
      "loss": 2.0312,
      "step": 4169
    },
    {
      "epoch": 2.76158940397351,
      "grad_norm": 0.7976650708330737,
      "learning_rate": 0.00023533643389764595,
      "loss": 2.2812,
      "step": 4170
    },
    {
      "epoch": 2.762251655629139,
      "grad_norm": 0.84395738741019,
      "learning_rate": 0.0002352815321201129,
      "loss": 2.0625,
      "step": 4171
    },
    {
      "epoch": 2.7629139072847684,
      "grad_norm": 0.8406471270033072,
      "learning_rate": 0.00023522661345585876,
      "loss": 2.2812,
      "step": 4172
    },
    {
      "epoch": 2.7635761589403973,
      "grad_norm": 0.7794261055320769,
      "learning_rate": 0.00023517167791575806,
      "loss": 2.25,
      "step": 4173
    },
    {
      "epoch": 2.7642384105960263,
      "grad_norm": 0.8372643972903571,
      "learning_rate": 0.0002351167255106887,
      "loss": 2.1406,
      "step": 4174
    },
    {
      "epoch": 2.7649006622516556,
      "grad_norm": 0.8290953114904558,
      "learning_rate": 0.0002350617562515318,
      "loss": 2.1719,
      "step": 4175
    },
    {
      "epoch": 2.765562913907285,
      "grad_norm": 0.6910806801093337,
      "learning_rate": 0.000235006770149172,
      "loss": 1.6328,
      "step": 4176
    },
    {
      "epoch": 2.766225165562914,
      "grad_norm": 0.8245155037960146,
      "learning_rate": 0.00023495176721449715,
      "loss": 2.2969,
      "step": 4177
    },
    {
      "epoch": 2.766887417218543,
      "grad_norm": 0.7120429958507726,
      "learning_rate": 0.00023489674745839843,
      "loss": 1.7734,
      "step": 4178
    },
    {
      "epoch": 2.7675496688741723,
      "grad_norm": 0.7986479086853632,
      "learning_rate": 0.00023484171089177043,
      "loss": 2.25,
      "step": 4179
    },
    {
      "epoch": 2.7682119205298013,
      "grad_norm": 0.8412779935967443,
      "learning_rate": 0.000234786657525511,
      "loss": 2.3906,
      "step": 4180
    },
    {
      "epoch": 2.7688741721854306,
      "grad_norm": 0.820590249860559,
      "learning_rate": 0.00023473158737052127,
      "loss": 2.3438,
      "step": 4181
    },
    {
      "epoch": 2.7695364238410596,
      "grad_norm": 0.7755238200801939,
      "learning_rate": 0.00023467650043770586,
      "loss": 2.2969,
      "step": 4182
    },
    {
      "epoch": 2.7701986754966885,
      "grad_norm": 0.7949863308464199,
      "learning_rate": 0.0002346213967379726,
      "loss": 2.3125,
      "step": 4183
    },
    {
      "epoch": 2.770860927152318,
      "grad_norm": 0.7354806776657277,
      "learning_rate": 0.0002345662762822326,
      "loss": 2.2969,
      "step": 4184
    },
    {
      "epoch": 2.7715231788079473,
      "grad_norm": 0.7888059961844744,
      "learning_rate": 0.0002345111390814004,
      "loss": 2.3438,
      "step": 4185
    },
    {
      "epoch": 2.7721854304635762,
      "grad_norm": 0.7944500906467029,
      "learning_rate": 0.00023445598514639377,
      "loss": 2.0938,
      "step": 4186
    },
    {
      "epoch": 2.772847682119205,
      "grad_norm": 0.8131129034010449,
      "learning_rate": 0.00023440081448813382,
      "loss": 2.2969,
      "step": 4187
    },
    {
      "epoch": 2.7735099337748346,
      "grad_norm": 0.7092598273813782,
      "learning_rate": 0.00023434562711754494,
      "loss": 1.9688,
      "step": 4188
    },
    {
      "epoch": 2.7741721854304635,
      "grad_norm": 0.7843541153039585,
      "learning_rate": 0.00023429042304555495,
      "loss": 2.0938,
      "step": 4189
    },
    {
      "epoch": 2.774834437086093,
      "grad_norm": 0.7826794456698766,
      "learning_rate": 0.00023423520228309483,
      "loss": 1.7422,
      "step": 4190
    },
    {
      "epoch": 2.775496688741722,
      "grad_norm": 0.7961359691973965,
      "learning_rate": 0.0002341799648410989,
      "loss": 2.2031,
      "step": 4191
    },
    {
      "epoch": 2.776158940397351,
      "grad_norm": 0.748431233819658,
      "learning_rate": 0.00023412471073050485,
      "loss": 2.3125,
      "step": 4192
    },
    {
      "epoch": 2.77682119205298,
      "grad_norm": 0.7205158254930283,
      "learning_rate": 0.00023406943996225367,
      "loss": 1.9375,
      "step": 4193
    },
    {
      "epoch": 2.777483443708609,
      "grad_norm": 0.7511280263729264,
      "learning_rate": 0.00023401415254728952,
      "loss": 2.2188,
      "step": 4194
    },
    {
      "epoch": 2.7781456953642385,
      "grad_norm": 0.7612973725551422,
      "learning_rate": 0.00023395884849655992,
      "loss": 2.0625,
      "step": 4195
    },
    {
      "epoch": 2.7788079470198674,
      "grad_norm": 0.7559634877051492,
      "learning_rate": 0.00023390352782101586,
      "loss": 2.2031,
      "step": 4196
    },
    {
      "epoch": 2.779470198675497,
      "grad_norm": 0.7534210084281672,
      "learning_rate": 0.0002338481905316113,
      "loss": 2.1562,
      "step": 4197
    },
    {
      "epoch": 2.7801324503311258,
      "grad_norm": 0.7440041169058349,
      "learning_rate": 0.00023379283663930374,
      "loss": 1.9453,
      "step": 4198
    },
    {
      "epoch": 2.780794701986755,
      "grad_norm": 0.7814918334930664,
      "learning_rate": 0.00023373746615505388,
      "loss": 2.375,
      "step": 4199
    },
    {
      "epoch": 2.781456953642384,
      "grad_norm": 0.7219610379705318,
      "learning_rate": 0.00023368207908982573,
      "loss": 2.0781,
      "step": 4200
    },
    {
      "epoch": 2.782119205298013,
      "grad_norm": 0.751270972980044,
      "learning_rate": 0.00023362667545458646,
      "loss": 1.8594,
      "step": 4201
    },
    {
      "epoch": 2.7827814569536424,
      "grad_norm": 0.9004341400571496,
      "learning_rate": 0.0002335712552603068,
      "loss": 2.2812,
      "step": 4202
    },
    {
      "epoch": 2.7834437086092714,
      "grad_norm": 0.8119038622886741,
      "learning_rate": 0.00023351581851796042,
      "loss": 2.2812,
      "step": 4203
    },
    {
      "epoch": 2.7841059602649008,
      "grad_norm": 0.8298367742046188,
      "learning_rate": 0.00023346036523852446,
      "loss": 2.2344,
      "step": 4204
    },
    {
      "epoch": 2.7847682119205297,
      "grad_norm": 0.7239688367435931,
      "learning_rate": 0.00023340489543297944,
      "loss": 2.0156,
      "step": 4205
    },
    {
      "epoch": 2.785430463576159,
      "grad_norm": 0.790849640489346,
      "learning_rate": 0.00023334940911230885,
      "loss": 2.1094,
      "step": 4206
    },
    {
      "epoch": 2.786092715231788,
      "grad_norm": 0.7634660143756492,
      "learning_rate": 0.00023329390628749972,
      "loss": 1.8672,
      "step": 4207
    },
    {
      "epoch": 2.7867549668874174,
      "grad_norm": 0.7258488312771718,
      "learning_rate": 0.0002332383869695422,
      "loss": 1.8438,
      "step": 4208
    },
    {
      "epoch": 2.7874172185430464,
      "grad_norm": 0.7432328433894279,
      "learning_rate": 0.00023318285116942977,
      "loss": 1.7188,
      "step": 4209
    },
    {
      "epoch": 2.7880794701986753,
      "grad_norm": 0.6934020614599951,
      "learning_rate": 0.00023312729889815912,
      "loss": 1.8984,
      "step": 4210
    },
    {
      "epoch": 2.7887417218543047,
      "grad_norm": 0.8702525910495136,
      "learning_rate": 0.00023307173016673028,
      "loss": 2.1562,
      "step": 4211
    },
    {
      "epoch": 2.7894039735099336,
      "grad_norm": 0.8630892164784327,
      "learning_rate": 0.0002330161449861465,
      "loss": 2.0625,
      "step": 4212
    },
    {
      "epoch": 2.790066225165563,
      "grad_norm": 0.8658686835071944,
      "learning_rate": 0.00023296054336741423,
      "loss": 2.3906,
      "step": 4213
    },
    {
      "epoch": 2.790728476821192,
      "grad_norm": 0.8193324912709612,
      "learning_rate": 0.00023290492532154327,
      "loss": 2.4688,
      "step": 4214
    },
    {
      "epoch": 2.7913907284768213,
      "grad_norm": 0.8372812840759414,
      "learning_rate": 0.00023284929085954663,
      "loss": 2.3281,
      "step": 4215
    },
    {
      "epoch": 2.7920529801324503,
      "grad_norm": 0.8040277739781813,
      "learning_rate": 0.00023279363999244057,
      "loss": 2.375,
      "step": 4216
    },
    {
      "epoch": 2.7927152317880797,
      "grad_norm": 0.8025280096389449,
      "learning_rate": 0.0002327379727312446,
      "loss": 2.0469,
      "step": 4217
    },
    {
      "epoch": 2.7933774834437086,
      "grad_norm": 0.7568735397504575,
      "learning_rate": 0.00023268228908698144,
      "loss": 2.25,
      "step": 4218
    },
    {
      "epoch": 2.7940397350993376,
      "grad_norm": 0.8232854899606435,
      "learning_rate": 0.00023262658907067716,
      "loss": 2.2344,
      "step": 4219
    },
    {
      "epoch": 2.794701986754967,
      "grad_norm": 0.7832514849770001,
      "learning_rate": 0.0002325708726933609,
      "loss": 2.4219,
      "step": 4220
    },
    {
      "epoch": 2.795364238410596,
      "grad_norm": 0.7821770340460427,
      "learning_rate": 0.00023251513996606525,
      "loss": 1.9844,
      "step": 4221
    },
    {
      "epoch": 2.7960264900662253,
      "grad_norm": 0.7492770406860467,
      "learning_rate": 0.00023245939089982593,
      "loss": 2.0469,
      "step": 4222
    },
    {
      "epoch": 2.796688741721854,
      "grad_norm": 0.7023535552829512,
      "learning_rate": 0.00023240362550568176,
      "loss": 1.8516,
      "step": 4223
    },
    {
      "epoch": 2.7973509933774836,
      "grad_norm": 0.7435860054478869,
      "learning_rate": 0.00023234784379467502,
      "loss": 2.0469,
      "step": 4224
    },
    {
      "epoch": 2.7980132450331126,
      "grad_norm": 0.7292393886325331,
      "learning_rate": 0.00023229204577785117,
      "loss": 1.7812,
      "step": 4225
    },
    {
      "epoch": 2.798675496688742,
      "grad_norm": 0.7764089202843507,
      "learning_rate": 0.0002322362314662588,
      "loss": 2.2188,
      "step": 4226
    },
    {
      "epoch": 2.799337748344371,
      "grad_norm": 0.804420319958684,
      "learning_rate": 0.00023218040087094973,
      "loss": 2.1406,
      "step": 4227
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8154269358915601,
      "learning_rate": 0.00023212455400297918,
      "loss": 2.3281,
      "step": 4228
    },
    {
      "epoch": 2.800662251655629,
      "grad_norm": 0.7926866844395691,
      "learning_rate": 0.00023206869087340537,
      "loss": 2.0312,
      "step": 4229
    },
    {
      "epoch": 2.801324503311258,
      "grad_norm": 0.7730473252715683,
      "learning_rate": 0.0002320128114932899,
      "loss": 1.9219,
      "step": 4230
    },
    {
      "epoch": 2.8019867549668875,
      "grad_norm": 0.8574153571066321,
      "learning_rate": 0.00023195691587369753,
      "loss": 2.3281,
      "step": 4231
    },
    {
      "epoch": 2.8026490066225165,
      "grad_norm": 0.7965322680777257,
      "learning_rate": 0.00023190100402569621,
      "loss": 2.125,
      "step": 4232
    },
    {
      "epoch": 2.8033112582781454,
      "grad_norm": 0.7403512501193728,
      "learning_rate": 0.00023184507596035706,
      "loss": 2.125,
      "step": 4233
    },
    {
      "epoch": 2.803973509933775,
      "grad_norm": 0.7262448849710594,
      "learning_rate": 0.0002317891316887546,
      "loss": 2.0469,
      "step": 4234
    },
    {
      "epoch": 2.804635761589404,
      "grad_norm": 0.796876405321901,
      "learning_rate": 0.0002317331712219664,
      "loss": 2.2344,
      "step": 4235
    },
    {
      "epoch": 2.805298013245033,
      "grad_norm": 0.8038449890552815,
      "learning_rate": 0.00023167719457107327,
      "loss": 2.0,
      "step": 4236
    },
    {
      "epoch": 2.805960264900662,
      "grad_norm": 0.8296033059326662,
      "learning_rate": 0.00023162120174715918,
      "loss": 2.1094,
      "step": 4237
    },
    {
      "epoch": 2.8066225165562915,
      "grad_norm": 0.8307762074808884,
      "learning_rate": 0.00023156519276131137,
      "loss": 2.2031,
      "step": 4238
    },
    {
      "epoch": 2.8072847682119204,
      "grad_norm": 0.8177472708275646,
      "learning_rate": 0.00023150916762462032,
      "loss": 2.2344,
      "step": 4239
    },
    {
      "epoch": 2.80794701986755,
      "grad_norm": 0.7533977738077938,
      "learning_rate": 0.00023145312634817957,
      "loss": 1.7734,
      "step": 4240
    },
    {
      "epoch": 2.8086092715231787,
      "grad_norm": 0.7999699882788476,
      "learning_rate": 0.000231397068943086,
      "loss": 2.375,
      "step": 4241
    },
    {
      "epoch": 2.8092715231788077,
      "grad_norm": 0.8064547779666964,
      "learning_rate": 0.0002313409954204396,
      "loss": 2.5469,
      "step": 4242
    },
    {
      "epoch": 2.809933774834437,
      "grad_norm": 0.7755315750614865,
      "learning_rate": 0.00023128490579134352,
      "loss": 2.0469,
      "step": 4243
    },
    {
      "epoch": 2.8105960264900665,
      "grad_norm": 0.7084488353709306,
      "learning_rate": 0.00023122880006690424,
      "loss": 1.7344,
      "step": 4244
    },
    {
      "epoch": 2.8112582781456954,
      "grad_norm": 0.7947730780597543,
      "learning_rate": 0.0002311726782582313,
      "loss": 2.1406,
      "step": 4245
    },
    {
      "epoch": 2.8119205298013243,
      "grad_norm": 0.8110006656971359,
      "learning_rate": 0.0002311165403764374,
      "loss": 2.4062,
      "step": 4246
    },
    {
      "epoch": 2.8125827814569537,
      "grad_norm": 0.809472967421911,
      "learning_rate": 0.00023106038643263855,
      "loss": 2.3594,
      "step": 4247
    },
    {
      "epoch": 2.8132450331125827,
      "grad_norm": 0.7726614726711921,
      "learning_rate": 0.00023100421643795384,
      "loss": 2.1719,
      "step": 4248
    },
    {
      "epoch": 2.813907284768212,
      "grad_norm": 0.7781945326326388,
      "learning_rate": 0.0002309480304035056,
      "loss": 1.6953,
      "step": 4249
    },
    {
      "epoch": 2.814569536423841,
      "grad_norm": 0.7463990337871235,
      "learning_rate": 0.00023089182834041934,
      "loss": 1.7734,
      "step": 4250
    },
    {
      "epoch": 2.81523178807947,
      "grad_norm": 0.8109852964760289,
      "learning_rate": 0.00023083561025982367,
      "loss": 2.0938,
      "step": 4251
    },
    {
      "epoch": 2.8158940397350993,
      "grad_norm": 0.7577545593568166,
      "learning_rate": 0.0002307793761728504,
      "loss": 1.9766,
      "step": 4252
    },
    {
      "epoch": 2.8165562913907287,
      "grad_norm": 0.8677207931541057,
      "learning_rate": 0.00023072312609063455,
      "loss": 2.2344,
      "step": 4253
    },
    {
      "epoch": 2.8172185430463577,
      "grad_norm": 0.7908274967945879,
      "learning_rate": 0.00023066686002431422,
      "loss": 2.2031,
      "step": 4254
    },
    {
      "epoch": 2.8178807947019866,
      "grad_norm": 0.7375562409068035,
      "learning_rate": 0.00023061057798503087,
      "loss": 1.7656,
      "step": 4255
    },
    {
      "epoch": 2.818543046357616,
      "grad_norm": 0.8453015759524933,
      "learning_rate": 0.0002305542799839289,
      "loss": 2.1719,
      "step": 4256
    },
    {
      "epoch": 2.819205298013245,
      "grad_norm": 0.7843556437360832,
      "learning_rate": 0.00023049796603215593,
      "loss": 1.9922,
      "step": 4257
    },
    {
      "epoch": 2.8198675496688743,
      "grad_norm": 0.901672951311254,
      "learning_rate": 0.00023044163614086276,
      "loss": 2.3906,
      "step": 4258
    },
    {
      "epoch": 2.8205298013245033,
      "grad_norm": 0.8081337320036998,
      "learning_rate": 0.0002303852903212035,
      "loss": 2.3438,
      "step": 4259
    },
    {
      "epoch": 2.821192052980132,
      "grad_norm": 0.749074456780704,
      "learning_rate": 0.00023032892858433512,
      "loss": 2.0781,
      "step": 4260
    },
    {
      "epoch": 2.8218543046357616,
      "grad_norm": 0.7442823720580068,
      "learning_rate": 0.00023027255094141796,
      "loss": 2.0938,
      "step": 4261
    },
    {
      "epoch": 2.822516556291391,
      "grad_norm": 0.7446990067807711,
      "learning_rate": 0.00023021615740361542,
      "loss": 2.0312,
      "step": 4262
    },
    {
      "epoch": 2.82317880794702,
      "grad_norm": 0.748039169482607,
      "learning_rate": 0.00023015974798209406,
      "loss": 2.0625,
      "step": 4263
    },
    {
      "epoch": 2.823841059602649,
      "grad_norm": 0.772051645178878,
      "learning_rate": 0.0002301033226880236,
      "loss": 2.1875,
      "step": 4264
    },
    {
      "epoch": 2.8245033112582782,
      "grad_norm": 0.8062786589231989,
      "learning_rate": 0.00023004688153257686,
      "loss": 2.3594,
      "step": 4265
    },
    {
      "epoch": 2.825165562913907,
      "grad_norm": 0.7238089584695271,
      "learning_rate": 0.0002299904245269299,
      "loss": 2.0938,
      "step": 4266
    },
    {
      "epoch": 2.8258278145695366,
      "grad_norm": 0.7901511011357996,
      "learning_rate": 0.00022993395168226181,
      "loss": 2.2656,
      "step": 4267
    },
    {
      "epoch": 2.8264900662251655,
      "grad_norm": 0.7569493336175531,
      "learning_rate": 0.00022987746300975488,
      "loss": 2.1875,
      "step": 4268
    },
    {
      "epoch": 2.8271523178807945,
      "grad_norm": 0.7019069743014565,
      "learning_rate": 0.00022982095852059454,
      "loss": 1.9141,
      "step": 4269
    },
    {
      "epoch": 2.827814569536424,
      "grad_norm": 0.754147260268682,
      "learning_rate": 0.00022976443822596927,
      "loss": 2.1094,
      "step": 4270
    },
    {
      "epoch": 2.8284768211920532,
      "grad_norm": 0.7727647126383128,
      "learning_rate": 0.00022970790213707075,
      "loss": 2.1562,
      "step": 4271
    },
    {
      "epoch": 2.829139072847682,
      "grad_norm": 0.7663979618079061,
      "learning_rate": 0.0002296513502650938,
      "loss": 2.2188,
      "step": 4272
    },
    {
      "epoch": 2.829801324503311,
      "grad_norm": 0.7963869587440184,
      "learning_rate": 0.00022959478262123627,
      "loss": 2.4062,
      "step": 4273
    },
    {
      "epoch": 2.8304635761589405,
      "grad_norm": 0.7863170127716259,
      "learning_rate": 0.00022953819921669934,
      "loss": 2.1562,
      "step": 4274
    },
    {
      "epoch": 2.8311258278145695,
      "grad_norm": 0.7436850672127658,
      "learning_rate": 0.00022948160006268703,
      "loss": 2.0,
      "step": 4275
    },
    {
      "epoch": 2.831788079470199,
      "grad_norm": 0.8183350955159097,
      "learning_rate": 0.0002294249851704067,
      "loss": 2.3594,
      "step": 4276
    },
    {
      "epoch": 2.832450331125828,
      "grad_norm": 0.7268090199507281,
      "learning_rate": 0.00022936835455106876,
      "loss": 2.0469,
      "step": 4277
    },
    {
      "epoch": 2.8331125827814567,
      "grad_norm": 0.7554174818230153,
      "learning_rate": 0.00022931170821588665,
      "loss": 2.2031,
      "step": 4278
    },
    {
      "epoch": 2.833774834437086,
      "grad_norm": 0.8212927948745228,
      "learning_rate": 0.00022925504617607703,
      "loss": 2.4219,
      "step": 4279
    },
    {
      "epoch": 2.8344370860927155,
      "grad_norm": 0.8338430484338789,
      "learning_rate": 0.0002291983684428597,
      "loss": 2.1875,
      "step": 4280
    },
    {
      "epoch": 2.8350993377483444,
      "grad_norm": 0.8443267893319859,
      "learning_rate": 0.00022914167502745736,
      "loss": 2.0781,
      "step": 4281
    },
    {
      "epoch": 2.8357615894039734,
      "grad_norm": 0.7461390764597494,
      "learning_rate": 0.00022908496594109608,
      "loss": 2.0312,
      "step": 4282
    },
    {
      "epoch": 2.8364238410596028,
      "grad_norm": 0.847820641004548,
      "learning_rate": 0.00022902824119500485,
      "loss": 2.1875,
      "step": 4283
    },
    {
      "epoch": 2.8370860927152317,
      "grad_norm": 0.7877234222825876,
      "learning_rate": 0.00022897150080041585,
      "loss": 2.2188,
      "step": 4284
    },
    {
      "epoch": 2.837748344370861,
      "grad_norm": 0.7795510068562231,
      "learning_rate": 0.00022891474476856427,
      "loss": 2.1406,
      "step": 4285
    },
    {
      "epoch": 2.83841059602649,
      "grad_norm": 0.6357193554690472,
      "learning_rate": 0.00022885797311068855,
      "loss": 1.6406,
      "step": 4286
    },
    {
      "epoch": 2.839072847682119,
      "grad_norm": 2.7534272466246597,
      "learning_rate": 0.00022880118583803004,
      "loss": 2.0781,
      "step": 4287
    },
    {
      "epoch": 2.8397350993377484,
      "grad_norm": 0.7569681892451441,
      "learning_rate": 0.0002287443829618333,
      "loss": 1.6562,
      "step": 4288
    },
    {
      "epoch": 2.8403973509933778,
      "grad_norm": 0.8361652217296667,
      "learning_rate": 0.00022868756449334602,
      "loss": 2.0938,
      "step": 4289
    },
    {
      "epoch": 2.8410596026490067,
      "grad_norm": 0.6898048692278205,
      "learning_rate": 0.0002286307304438188,
      "loss": 1.7031,
      "step": 4290
    },
    {
      "epoch": 2.8417218543046356,
      "grad_norm": 0.7978115771443376,
      "learning_rate": 0.00022857388082450546,
      "loss": 2.2656,
      "step": 4291
    },
    {
      "epoch": 2.842384105960265,
      "grad_norm": 0.815597262414709,
      "learning_rate": 0.00022851701564666297,
      "loss": 2.1719,
      "step": 4292
    },
    {
      "epoch": 2.843046357615894,
      "grad_norm": 0.7722317800372288,
      "learning_rate": 0.00022846013492155116,
      "loss": 2.1406,
      "step": 4293
    },
    {
      "epoch": 2.8437086092715234,
      "grad_norm": 0.8040546933072937,
      "learning_rate": 0.00022840323866043315,
      "loss": 2.4375,
      "step": 4294
    },
    {
      "epoch": 2.8443708609271523,
      "grad_norm": 0.8414941405427296,
      "learning_rate": 0.000228346326874575,
      "loss": 2.2344,
      "step": 4295
    },
    {
      "epoch": 2.8450331125827812,
      "grad_norm": 0.7571066148674479,
      "learning_rate": 0.0002282893995752459,
      "loss": 2.1562,
      "step": 4296
    },
    {
      "epoch": 2.8456953642384106,
      "grad_norm": 0.8581498371455727,
      "learning_rate": 0.00022823245677371817,
      "loss": 2.4062,
      "step": 4297
    },
    {
      "epoch": 2.8463576158940396,
      "grad_norm": 0.7389811800639423,
      "learning_rate": 0.00022817549848126704,
      "loss": 2.1406,
      "step": 4298
    },
    {
      "epoch": 2.847019867549669,
      "grad_norm": 0.7816227614483723,
      "learning_rate": 0.000228118524709171,
      "loss": 2.2812,
      "step": 4299
    },
    {
      "epoch": 2.847682119205298,
      "grad_norm": 0.779748630393211,
      "learning_rate": 0.00022806153546871141,
      "loss": 2.2656,
      "step": 4300
    },
    {
      "epoch": 2.8483443708609273,
      "grad_norm": 0.8140458559140917,
      "learning_rate": 0.0002280045307711729,
      "loss": 2.2812,
      "step": 4301
    },
    {
      "epoch": 2.8490066225165562,
      "grad_norm": 0.7533871208436292,
      "learning_rate": 0.00022794751062784298,
      "loss": 2.2812,
      "step": 4302
    },
    {
      "epoch": 2.8496688741721856,
      "grad_norm": 0.8561658084528757,
      "learning_rate": 0.0002278904750500124,
      "loss": 2.2969,
      "step": 4303
    },
    {
      "epoch": 2.8503311258278146,
      "grad_norm": 0.9174690832625928,
      "learning_rate": 0.00022783342404897466,
      "loss": 2.4062,
      "step": 4304
    },
    {
      "epoch": 2.8509933774834435,
      "grad_norm": 0.7892232529139972,
      "learning_rate": 0.00022777635763602668,
      "loss": 2.125,
      "step": 4305
    },
    {
      "epoch": 2.851655629139073,
      "grad_norm": 0.7830108639125648,
      "learning_rate": 0.0002277192758224682,
      "loss": 2.1562,
      "step": 4306
    },
    {
      "epoch": 2.852317880794702,
      "grad_norm": 0.8714669452056019,
      "learning_rate": 0.0002276621786196021,
      "loss": 2.1562,
      "step": 4307
    },
    {
      "epoch": 2.852980132450331,
      "grad_norm": 0.8183496932179902,
      "learning_rate": 0.00022760506603873424,
      "loss": 2.4688,
      "step": 4308
    },
    {
      "epoch": 2.85364238410596,
      "grad_norm": 0.8637671451819452,
      "learning_rate": 0.00022754793809117362,
      "loss": 2.375,
      "step": 4309
    },
    {
      "epoch": 2.8543046357615895,
      "grad_norm": 0.7191367872969532,
      "learning_rate": 0.0002274907947882322,
      "loss": 2.0312,
      "step": 4310
    },
    {
      "epoch": 2.8549668874172185,
      "grad_norm": 0.9191924080651089,
      "learning_rate": 0.00022743363614122496,
      "loss": 2.2969,
      "step": 4311
    },
    {
      "epoch": 2.855629139072848,
      "grad_norm": 0.8355350713643267,
      "learning_rate": 0.0002273764621614701,
      "loss": 2.2656,
      "step": 4312
    },
    {
      "epoch": 2.856291390728477,
      "grad_norm": 0.8078895178755016,
      "learning_rate": 0.0002273192728602886,
      "loss": 2.4062,
      "step": 4313
    },
    {
      "epoch": 2.8569536423841058,
      "grad_norm": 0.7620700589753664,
      "learning_rate": 0.00022726206824900459,
      "loss": 2.1562,
      "step": 4314
    },
    {
      "epoch": 2.857615894039735,
      "grad_norm": 0.7995824727746912,
      "learning_rate": 0.00022720484833894534,
      "loss": 2.0156,
      "step": 4315
    },
    {
      "epoch": 2.858278145695364,
      "grad_norm": 0.7845277744047192,
      "learning_rate": 0.00022714761314144094,
      "loss": 2.1719,
      "step": 4316
    },
    {
      "epoch": 2.8589403973509935,
      "grad_norm": 0.7322900159291628,
      "learning_rate": 0.00022709036266782472,
      "loss": 2.125,
      "step": 4317
    },
    {
      "epoch": 2.8596026490066224,
      "grad_norm": 0.7694931190787067,
      "learning_rate": 0.0002270330969294329,
      "loss": 2.0625,
      "step": 4318
    },
    {
      "epoch": 2.860264900662252,
      "grad_norm": 0.8118855161875603,
      "learning_rate": 0.00022697581593760467,
      "loss": 2.0938,
      "step": 4319
    },
    {
      "epoch": 2.8609271523178808,
      "grad_norm": 0.8222520650845837,
      "learning_rate": 0.00022691851970368243,
      "loss": 2.1875,
      "step": 4320
    },
    {
      "epoch": 2.86158940397351,
      "grad_norm": 0.7920715466540987,
      "learning_rate": 0.00022686120823901148,
      "loss": 2.0156,
      "step": 4321
    },
    {
      "epoch": 2.862251655629139,
      "grad_norm": 0.7375070520555159,
      "learning_rate": 0.00022680388155494005,
      "loss": 2.0938,
      "step": 4322
    },
    {
      "epoch": 2.862913907284768,
      "grad_norm": 0.7413996166535582,
      "learning_rate": 0.00022674653966281956,
      "loss": 1.8672,
      "step": 4323
    },
    {
      "epoch": 2.8635761589403974,
      "grad_norm": 0.8045071123117419,
      "learning_rate": 0.00022668918257400436,
      "loss": 2.0938,
      "step": 4324
    },
    {
      "epoch": 2.8642384105960264,
      "grad_norm": 0.7838641917039068,
      "learning_rate": 0.00022663181029985175,
      "loss": 2.1094,
      "step": 4325
    },
    {
      "epoch": 2.8649006622516557,
      "grad_norm": 0.7712171496477863,
      "learning_rate": 0.0002265744228517222,
      "loss": 2.0469,
      "step": 4326
    },
    {
      "epoch": 2.8655629139072847,
      "grad_norm": 0.8891649412909731,
      "learning_rate": 0.00022651702024097903,
      "loss": 2.3594,
      "step": 4327
    },
    {
      "epoch": 2.866225165562914,
      "grad_norm": 0.80274163503277,
      "learning_rate": 0.00022645960247898858,
      "loss": 2.2031,
      "step": 4328
    },
    {
      "epoch": 2.866887417218543,
      "grad_norm": 0.765788969812281,
      "learning_rate": 0.00022640216957712025,
      "loss": 1.8594,
      "step": 4329
    },
    {
      "epoch": 2.8675496688741724,
      "grad_norm": 0.7909543757650036,
      "learning_rate": 0.00022634472154674646,
      "loss": 2.0625,
      "step": 4330
    },
    {
      "epoch": 2.8682119205298013,
      "grad_norm": 0.7820752640135349,
      "learning_rate": 0.0002262872583992425,
      "loss": 2.2031,
      "step": 4331
    },
    {
      "epoch": 2.8688741721854303,
      "grad_norm": 0.7748591209060456,
      "learning_rate": 0.00022622978014598684,
      "loss": 2.0312,
      "step": 4332
    },
    {
      "epoch": 2.8695364238410597,
      "grad_norm": 0.7784011873118347,
      "learning_rate": 0.00022617228679836068,
      "loss": 1.9766,
      "step": 4333
    },
    {
      "epoch": 2.8701986754966886,
      "grad_norm": 0.7875525626763957,
      "learning_rate": 0.0002261147783677485,
      "loss": 2.2656,
      "step": 4334
    },
    {
      "epoch": 2.870860927152318,
      "grad_norm": 0.6849342536124694,
      "learning_rate": 0.00022605725486553758,
      "loss": 1.8594,
      "step": 4335
    },
    {
      "epoch": 2.871523178807947,
      "grad_norm": 0.7565642297294322,
      "learning_rate": 0.00022599971630311822,
      "loss": 2.0312,
      "step": 4336
    },
    {
      "epoch": 2.872185430463576,
      "grad_norm": 0.6705695012665237,
      "learning_rate": 0.00022594216269188365,
      "loss": 1.7422,
      "step": 4337
    },
    {
      "epoch": 2.8728476821192053,
      "grad_norm": 0.8470494697029474,
      "learning_rate": 0.0002258845940432303,
      "loss": 2.4062,
      "step": 4338
    },
    {
      "epoch": 2.8735099337748347,
      "grad_norm": 0.764785123387304,
      "learning_rate": 0.00022582701036855734,
      "loss": 2.1406,
      "step": 4339
    },
    {
      "epoch": 2.8741721854304636,
      "grad_norm": 0.7943550482459245,
      "learning_rate": 0.00022576941167926697,
      "loss": 2.1719,
      "step": 4340
    },
    {
      "epoch": 2.8748344370860925,
      "grad_norm": 0.7530760275827229,
      "learning_rate": 0.00022571179798676444,
      "loss": 1.9922,
      "step": 4341
    },
    {
      "epoch": 2.875496688741722,
      "grad_norm": 0.730586421804568,
      "learning_rate": 0.00022565416930245792,
      "loss": 1.7109,
      "step": 4342
    },
    {
      "epoch": 2.876158940397351,
      "grad_norm": 0.8256585877735351,
      "learning_rate": 0.00022559652563775847,
      "loss": 2.125,
      "step": 4343
    },
    {
      "epoch": 2.8768211920529803,
      "grad_norm": 0.7843769049176272,
      "learning_rate": 0.0002255388670040803,
      "loss": 2.0469,
      "step": 4344
    },
    {
      "epoch": 2.877483443708609,
      "grad_norm": 0.7322742060131598,
      "learning_rate": 0.0002254811934128404,
      "loss": 2.1406,
      "step": 4345
    },
    {
      "epoch": 2.878145695364238,
      "grad_norm": 0.8760053975187727,
      "learning_rate": 0.00022542350487545892,
      "loss": 2.1875,
      "step": 4346
    },
    {
      "epoch": 2.8788079470198675,
      "grad_norm": 0.7857875978118242,
      "learning_rate": 0.0002253658014033587,
      "loss": 2.1875,
      "step": 4347
    },
    {
      "epoch": 2.879470198675497,
      "grad_norm": 0.7911363324108679,
      "learning_rate": 0.0002253080830079658,
      "loss": 2.1562,
      "step": 4348
    },
    {
      "epoch": 2.880132450331126,
      "grad_norm": 0.7964718063339266,
      "learning_rate": 0.00022525034970070906,
      "loss": 1.7031,
      "step": 4349
    },
    {
      "epoch": 2.880794701986755,
      "grad_norm": 0.736117924383515,
      "learning_rate": 0.00022519260149302038,
      "loss": 2.1094,
      "step": 4350
    },
    {
      "epoch": 2.881456953642384,
      "grad_norm": 0.8001213062960864,
      "learning_rate": 0.00022513483839633455,
      "loss": 1.9062,
      "step": 4351
    },
    {
      "epoch": 2.882119205298013,
      "grad_norm": 0.823698029731299,
      "learning_rate": 0.00022507706042208936,
      "loss": 2.2969,
      "step": 4352
    },
    {
      "epoch": 2.8827814569536425,
      "grad_norm": 0.765387513114179,
      "learning_rate": 0.00022501926758172538,
      "loss": 2.2188,
      "step": 4353
    },
    {
      "epoch": 2.8834437086092715,
      "grad_norm": 0.8016743427599137,
      "learning_rate": 0.00022496145988668643,
      "loss": 2.2031,
      "step": 4354
    },
    {
      "epoch": 2.8841059602649004,
      "grad_norm": 0.7342281739638434,
      "learning_rate": 0.00022490363734841898,
      "loss": 2.125,
      "step": 4355
    },
    {
      "epoch": 2.88476821192053,
      "grad_norm": 0.8330067617330862,
      "learning_rate": 0.0002248457999783726,
      "loss": 2.25,
      "step": 4356
    },
    {
      "epoch": 2.885430463576159,
      "grad_norm": 0.7497081607502295,
      "learning_rate": 0.00022478794778799978,
      "loss": 2.0938,
      "step": 4357
    },
    {
      "epoch": 2.886092715231788,
      "grad_norm": 0.8236186269902726,
      "learning_rate": 0.00022473008078875584,
      "loss": 2.25,
      "step": 4358
    },
    {
      "epoch": 2.886754966887417,
      "grad_norm": 0.8053080230822051,
      "learning_rate": 0.00022467219899209917,
      "loss": 2.0781,
      "step": 4359
    },
    {
      "epoch": 2.8874172185430464,
      "grad_norm": 0.7164664587738089,
      "learning_rate": 0.00022461430240949096,
      "loss": 1.8516,
      "step": 4360
    },
    {
      "epoch": 2.8880794701986754,
      "grad_norm": 0.8142001223518115,
      "learning_rate": 0.00022455639105239554,
      "loss": 2.25,
      "step": 4361
    },
    {
      "epoch": 2.888741721854305,
      "grad_norm": 0.7040437330515209,
      "learning_rate": 0.00022449846493227985,
      "loss": 1.8828,
      "step": 4362
    },
    {
      "epoch": 2.8894039735099337,
      "grad_norm": 0.7220505009445266,
      "learning_rate": 0.00022444052406061404,
      "loss": 1.8828,
      "step": 4363
    },
    {
      "epoch": 2.8900662251655627,
      "grad_norm": 0.7856552309358815,
      "learning_rate": 0.00022438256844887103,
      "loss": 2.2031,
      "step": 4364
    },
    {
      "epoch": 2.890728476821192,
      "grad_norm": 0.8502959740448913,
      "learning_rate": 0.0002243245981085267,
      "loss": 2.3281,
      "step": 4365
    },
    {
      "epoch": 2.8913907284768214,
      "grad_norm": 0.8097844518806273,
      "learning_rate": 0.00022426661305105982,
      "loss": 1.7422,
      "step": 4366
    },
    {
      "epoch": 2.8920529801324504,
      "grad_norm": 0.8382135619585039,
      "learning_rate": 0.0002242086132879522,
      "loss": 2.2812,
      "step": 4367
    },
    {
      "epoch": 2.8927152317880793,
      "grad_norm": 0.8040259502722577,
      "learning_rate": 0.00022415059883068827,
      "loss": 2.1562,
      "step": 4368
    },
    {
      "epoch": 2.8933774834437087,
      "grad_norm": 0.7487370394253613,
      "learning_rate": 0.00022409256969075573,
      "loss": 1.8984,
      "step": 4369
    },
    {
      "epoch": 2.8940397350993377,
      "grad_norm": 0.8121435968410278,
      "learning_rate": 0.00022403452587964491,
      "loss": 2.2188,
      "step": 4370
    },
    {
      "epoch": 2.894701986754967,
      "grad_norm": 0.8112797344685503,
      "learning_rate": 0.00022397646740884923,
      "loss": 2.3438,
      "step": 4371
    },
    {
      "epoch": 2.895364238410596,
      "grad_norm": 0.7445266467259728,
      "learning_rate": 0.00022391839428986482,
      "loss": 2.1094,
      "step": 4372
    },
    {
      "epoch": 2.896026490066225,
      "grad_norm": 0.7983038417800611,
      "learning_rate": 0.00022386030653419098,
      "loss": 2.3906,
      "step": 4373
    },
    {
      "epoch": 2.8966887417218543,
      "grad_norm": 0.8238803405875503,
      "learning_rate": 0.00022380220415332957,
      "loss": 2.3438,
      "step": 4374
    },
    {
      "epoch": 2.8973509933774837,
      "grad_norm": 0.7338443097363743,
      "learning_rate": 0.0002237440871587857,
      "loss": 1.7656,
      "step": 4375
    },
    {
      "epoch": 2.8980132450331126,
      "grad_norm": 0.7664758917022869,
      "learning_rate": 0.0002236859555620671,
      "loss": 1.7344,
      "step": 4376
    },
    {
      "epoch": 2.8986754966887416,
      "grad_norm": 0.7991264301220364,
      "learning_rate": 0.00022362780937468452,
      "loss": 2.2188,
      "step": 4377
    },
    {
      "epoch": 2.899337748344371,
      "grad_norm": 0.7684386742476127,
      "learning_rate": 0.00022356964860815155,
      "loss": 2.0938,
      "step": 4378
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8132228349696328,
      "learning_rate": 0.00022351147327398475,
      "loss": 1.8672,
      "step": 4379
    },
    {
      "epoch": 2.9006622516556293,
      "grad_norm": 0.7545679171857776,
      "learning_rate": 0.00022345328338370342,
      "loss": 2.2188,
      "step": 4380
    },
    {
      "epoch": 2.9013245033112582,
      "grad_norm": 0.8098721462421289,
      "learning_rate": 0.00022339507894882985,
      "loss": 2.0156,
      "step": 4381
    },
    {
      "epoch": 2.901986754966887,
      "grad_norm": 0.7250274957792093,
      "learning_rate": 0.00022333685998088924,
      "loss": 1.9922,
      "step": 4382
    },
    {
      "epoch": 2.9026490066225166,
      "grad_norm": 0.7199057703051756,
      "learning_rate": 0.00022327862649140957,
      "loss": 2.0938,
      "step": 4383
    },
    {
      "epoch": 2.903311258278146,
      "grad_norm": 0.804408619705093,
      "learning_rate": 0.00022322037849192172,
      "loss": 2.2656,
      "step": 4384
    },
    {
      "epoch": 2.903973509933775,
      "grad_norm": 0.7489591907282805,
      "learning_rate": 0.0002231621159939595,
      "loss": 2.1875,
      "step": 4385
    },
    {
      "epoch": 2.904635761589404,
      "grad_norm": 0.7652456168806213,
      "learning_rate": 0.00022310383900905957,
      "loss": 2.1719,
      "step": 4386
    },
    {
      "epoch": 2.9052980132450332,
      "grad_norm": 0.9299591974475737,
      "learning_rate": 0.00022304554754876135,
      "loss": 2.5781,
      "step": 4387
    },
    {
      "epoch": 2.905960264900662,
      "grad_norm": 0.7269200016588775,
      "learning_rate": 0.0002229872416246073,
      "loss": 2.2812,
      "step": 4388
    },
    {
      "epoch": 2.9066225165562916,
      "grad_norm": 0.7246332908206276,
      "learning_rate": 0.00022292892124814264,
      "loss": 2.1406,
      "step": 4389
    },
    {
      "epoch": 2.9072847682119205,
      "grad_norm": 0.750095856647597,
      "learning_rate": 0.00022287058643091548,
      "loss": 2.2031,
      "step": 4390
    },
    {
      "epoch": 2.9079470198675494,
      "grad_norm": 0.7689055995694445,
      "learning_rate": 0.00022281223718447672,
      "loss": 2.2344,
      "step": 4391
    },
    {
      "epoch": 2.908609271523179,
      "grad_norm": 0.7094383988468619,
      "learning_rate": 0.00022275387352038028,
      "loss": 2.0469,
      "step": 4392
    },
    {
      "epoch": 2.909271523178808,
      "grad_norm": 0.7671209839447317,
      "learning_rate": 0.00022269549545018283,
      "loss": 2.3594,
      "step": 4393
    },
    {
      "epoch": 2.909933774834437,
      "grad_norm": 0.8075845535906568,
      "learning_rate": 0.0002226371029854438,
      "loss": 2.1094,
      "step": 4394
    },
    {
      "epoch": 2.910596026490066,
      "grad_norm": 0.7566121971854343,
      "learning_rate": 0.00022257869613772556,
      "loss": 2.0938,
      "step": 4395
    },
    {
      "epoch": 2.9112582781456955,
      "grad_norm": 0.8101089447041504,
      "learning_rate": 0.0002225202749185935,
      "loss": 2.1094,
      "step": 4396
    },
    {
      "epoch": 2.9119205298013244,
      "grad_norm": 0.7826169549376614,
      "learning_rate": 0.00022246183933961555,
      "loss": 2.3906,
      "step": 4397
    },
    {
      "epoch": 2.912582781456954,
      "grad_norm": 0.7770810780765592,
      "learning_rate": 0.00022240338941236263,
      "loss": 2.2031,
      "step": 4398
    },
    {
      "epoch": 2.9132450331125828,
      "grad_norm": 0.7546483542485276,
      "learning_rate": 0.0002223449251484086,
      "loss": 1.7188,
      "step": 4399
    },
    {
      "epoch": 2.9139072847682117,
      "grad_norm": 0.9016660213290708,
      "learning_rate": 0.00022228644655932994,
      "loss": 2.25,
      "step": 4400
    },
    {
      "epoch": 2.914569536423841,
      "grad_norm": 0.7817928476538343,
      "learning_rate": 0.0002222279536567061,
      "loss": 2.1562,
      "step": 4401
    },
    {
      "epoch": 2.91523178807947,
      "grad_norm": 0.7832703143428481,
      "learning_rate": 0.00022216944645211942,
      "loss": 2.1875,
      "step": 4402
    },
    {
      "epoch": 2.9158940397350994,
      "grad_norm": 0.8236398574245598,
      "learning_rate": 0.0002221109249571549,
      "loss": 2.0,
      "step": 4403
    },
    {
      "epoch": 2.9165562913907284,
      "grad_norm": 0.8161552589500917,
      "learning_rate": 0.00022205238918340055,
      "loss": 2.25,
      "step": 4404
    },
    {
      "epoch": 2.9172185430463577,
      "grad_norm": 0.7605307896669022,
      "learning_rate": 0.00022199383914244707,
      "loss": 2.2812,
      "step": 4405
    },
    {
      "epoch": 2.9178807947019867,
      "grad_norm": 0.8444774078955247,
      "learning_rate": 0.00022193527484588806,
      "loss": 2.3125,
      "step": 4406
    },
    {
      "epoch": 2.918543046357616,
      "grad_norm": 0.787730212352518,
      "learning_rate": 0.00022187669630531994,
      "loss": 1.9844,
      "step": 4407
    },
    {
      "epoch": 2.919205298013245,
      "grad_norm": 0.7084323012307221,
      "learning_rate": 0.0002218181035323419,
      "loss": 2.0312,
      "step": 4408
    },
    {
      "epoch": 2.919867549668874,
      "grad_norm": 0.6715426835901356,
      "learning_rate": 0.00022175949653855596,
      "loss": 1.7578,
      "step": 4409
    },
    {
      "epoch": 2.9205298013245033,
      "grad_norm": 0.822227804796344,
      "learning_rate": 0.00022170087533556707,
      "loss": 2.1875,
      "step": 4410
    },
    {
      "epoch": 2.9211920529801323,
      "grad_norm": 0.7979246488949965,
      "learning_rate": 0.00022164223993498277,
      "loss": 2.4219,
      "step": 4411
    },
    {
      "epoch": 2.9218543046357617,
      "grad_norm": 0.7761589318783694,
      "learning_rate": 0.00022158359034841368,
      "loss": 2.1562,
      "step": 4412
    },
    {
      "epoch": 2.9225165562913906,
      "grad_norm": 0.800356755520214,
      "learning_rate": 0.00022152492658747297,
      "loss": 2.2188,
      "step": 4413
    },
    {
      "epoch": 2.92317880794702,
      "grad_norm": 0.6884868023473854,
      "learning_rate": 0.00022146624866377685,
      "loss": 1.875,
      "step": 4414
    },
    {
      "epoch": 2.923841059602649,
      "grad_norm": 0.8360567391001918,
      "learning_rate": 0.00022140755658894409,
      "loss": 2.3438,
      "step": 4415
    },
    {
      "epoch": 2.9245033112582783,
      "grad_norm": 0.8314932754420089,
      "learning_rate": 0.0002213488503745965,
      "loss": 2.2969,
      "step": 4416
    },
    {
      "epoch": 2.9251655629139073,
      "grad_norm": 0.772391265499767,
      "learning_rate": 0.00022129013003235857,
      "loss": 2.2188,
      "step": 4417
    },
    {
      "epoch": 2.9258278145695362,
      "grad_norm": 0.7565697862116512,
      "learning_rate": 0.0002212313955738575,
      "loss": 1.9844,
      "step": 4418
    },
    {
      "epoch": 2.9264900662251656,
      "grad_norm": 0.8194897347181506,
      "learning_rate": 0.00022117264701072354,
      "loss": 2.2969,
      "step": 4419
    },
    {
      "epoch": 2.9271523178807946,
      "grad_norm": 0.7623189541295161,
      "learning_rate": 0.0002211138843545895,
      "loss": 1.9922,
      "step": 4420
    },
    {
      "epoch": 2.927814569536424,
      "grad_norm": 0.8139847421132483,
      "learning_rate": 0.00022105510761709102,
      "loss": 2.1562,
      "step": 4421
    },
    {
      "epoch": 2.928476821192053,
      "grad_norm": 0.7486468355834555,
      "learning_rate": 0.00022099631680986667,
      "loss": 1.9219,
      "step": 4422
    },
    {
      "epoch": 2.9291390728476823,
      "grad_norm": 0.7684198728068913,
      "learning_rate": 0.0002209375119445576,
      "loss": 2.125,
      "step": 4423
    },
    {
      "epoch": 2.929801324503311,
      "grad_norm": 0.7402609538144354,
      "learning_rate": 0.0002208786930328079,
      "loss": 2.1875,
      "step": 4424
    },
    {
      "epoch": 2.9304635761589406,
      "grad_norm": 0.6872091441898548,
      "learning_rate": 0.00022081986008626442,
      "loss": 1.8516,
      "step": 4425
    },
    {
      "epoch": 2.9311258278145695,
      "grad_norm": 0.8060554889522482,
      "learning_rate": 0.00022076101311657665,
      "loss": 2.1406,
      "step": 4426
    },
    {
      "epoch": 2.9317880794701985,
      "grad_norm": 0.782523512720665,
      "learning_rate": 0.00022070215213539708,
      "loss": 2.1094,
      "step": 4427
    },
    {
      "epoch": 2.932450331125828,
      "grad_norm": 0.7797477108745386,
      "learning_rate": 0.0002206432771543808,
      "loss": 2.0156,
      "step": 4428
    },
    {
      "epoch": 2.933112582781457,
      "grad_norm": 0.7406389955235828,
      "learning_rate": 0.00022058438818518575,
      "loss": 2.0938,
      "step": 4429
    },
    {
      "epoch": 2.933774834437086,
      "grad_norm": 0.7917561317572084,
      "learning_rate": 0.00022052548523947258,
      "loss": 2.1875,
      "step": 4430
    },
    {
      "epoch": 2.934437086092715,
      "grad_norm": 0.7673806799573446,
      "learning_rate": 0.00022046656832890485,
      "loss": 1.9531,
      "step": 4431
    },
    {
      "epoch": 2.9350993377483445,
      "grad_norm": 0.7592415231873181,
      "learning_rate": 0.00022040763746514867,
      "loss": 2.25,
      "step": 4432
    },
    {
      "epoch": 2.9357615894039735,
      "grad_norm": 0.68968495370084,
      "learning_rate": 0.0002203486926598731,
      "loss": 1.7891,
      "step": 4433
    },
    {
      "epoch": 2.936423841059603,
      "grad_norm": 0.8189087607148038,
      "learning_rate": 0.00022028973392474986,
      "loss": 2.2031,
      "step": 4434
    },
    {
      "epoch": 2.937086092715232,
      "grad_norm": 0.7820649968477431,
      "learning_rate": 0.00022023076127145345,
      "loss": 2.0781,
      "step": 4435
    },
    {
      "epoch": 2.9377483443708607,
      "grad_norm": 0.8106576243652908,
      "learning_rate": 0.00022017177471166113,
      "loss": 2.3594,
      "step": 4436
    },
    {
      "epoch": 2.93841059602649,
      "grad_norm": 0.7665170914068747,
      "learning_rate": 0.00022011277425705293,
      "loss": 2.25,
      "step": 4437
    },
    {
      "epoch": 2.939072847682119,
      "grad_norm": 0.7489571250743352,
      "learning_rate": 0.00022005375991931162,
      "loss": 2.0938,
      "step": 4438
    },
    {
      "epoch": 2.9397350993377485,
      "grad_norm": 0.7709425541830434,
      "learning_rate": 0.00021999473171012272,
      "loss": 2.0625,
      "step": 4439
    },
    {
      "epoch": 2.9403973509933774,
      "grad_norm": 0.7076182706472636,
      "learning_rate": 0.00021993568964117445,
      "loss": 2.0625,
      "step": 4440
    },
    {
      "epoch": 2.941059602649007,
      "grad_norm": 0.768315105440346,
      "learning_rate": 0.00021987663372415785,
      "loss": 2.0938,
      "step": 4441
    },
    {
      "epoch": 2.9417218543046357,
      "grad_norm": 0.760324880109208,
      "learning_rate": 0.00021981756397076665,
      "loss": 2.1719,
      "step": 4442
    },
    {
      "epoch": 2.942384105960265,
      "grad_norm": 0.7820826706849016,
      "learning_rate": 0.00021975848039269738,
      "loss": 2.0781,
      "step": 4443
    },
    {
      "epoch": 2.943046357615894,
      "grad_norm": 0.8109947859070346,
      "learning_rate": 0.00021969938300164923,
      "loss": 2.1406,
      "step": 4444
    },
    {
      "epoch": 2.943708609271523,
      "grad_norm": 0.786749196896755,
      "learning_rate": 0.00021964027180932416,
      "loss": 2.0625,
      "step": 4445
    },
    {
      "epoch": 2.9443708609271524,
      "grad_norm": 0.7957003400491892,
      "learning_rate": 0.00021958114682742687,
      "loss": 2.3281,
      "step": 4446
    },
    {
      "epoch": 2.9450331125827813,
      "grad_norm": 0.8195127519653472,
      "learning_rate": 0.0002195220080676648,
      "loss": 2.3594,
      "step": 4447
    },
    {
      "epoch": 2.9456953642384107,
      "grad_norm": 0.7735375450136942,
      "learning_rate": 0.00021946285554174806,
      "loss": 2.0781,
      "step": 4448
    },
    {
      "epoch": 2.9463576158940397,
      "grad_norm": 0.7419608431128963,
      "learning_rate": 0.0002194036892613896,
      "loss": 2.1719,
      "step": 4449
    },
    {
      "epoch": 2.9470198675496686,
      "grad_norm": 0.7199208229335173,
      "learning_rate": 0.00021934450923830496,
      "loss": 2.0625,
      "step": 4450
    },
    {
      "epoch": 2.947682119205298,
      "grad_norm": 0.7840832086790385,
      "learning_rate": 0.00021928531548421246,
      "loss": 2.2812,
      "step": 4451
    },
    {
      "epoch": 2.9483443708609274,
      "grad_norm": 0.7604315819547849,
      "learning_rate": 0.0002192261080108332,
      "loss": 1.9609,
      "step": 4452
    },
    {
      "epoch": 2.9490066225165563,
      "grad_norm": 0.7922142817023269,
      "learning_rate": 0.00021916688682989085,
      "loss": 2.2812,
      "step": 4453
    },
    {
      "epoch": 2.9496688741721853,
      "grad_norm": 0.721428609954887,
      "learning_rate": 0.000219107651953112,
      "loss": 2.0469,
      "step": 4454
    },
    {
      "epoch": 2.9503311258278146,
      "grad_norm": 0.7598234997873866,
      "learning_rate": 0.00021904840339222572,
      "loss": 2.1406,
      "step": 4455
    },
    {
      "epoch": 2.9509933774834436,
      "grad_norm": 0.7188071465148412,
      "learning_rate": 0.000218989141158964,
      "loss": 2.0625,
      "step": 4456
    },
    {
      "epoch": 2.951655629139073,
      "grad_norm": 0.743277619605507,
      "learning_rate": 0.00021892986526506135,
      "loss": 2.2969,
      "step": 4457
    },
    {
      "epoch": 2.952317880794702,
      "grad_norm": 0.682162637320931,
      "learning_rate": 0.00021887057572225516,
      "loss": 1.75,
      "step": 4458
    },
    {
      "epoch": 2.952980132450331,
      "grad_norm": 0.7834358923324569,
      "learning_rate": 0.00021881127254228533,
      "loss": 2.0781,
      "step": 4459
    },
    {
      "epoch": 2.9536423841059603,
      "grad_norm": 0.8120382484396265,
      "learning_rate": 0.00021875195573689473,
      "loss": 2.1719,
      "step": 4460
    },
    {
      "epoch": 2.9543046357615896,
      "grad_norm": 0.844829226007287,
      "learning_rate": 0.00021869262531782864,
      "loss": 2.1094,
      "step": 4461
    },
    {
      "epoch": 2.9549668874172186,
      "grad_norm": 0.7011440957275106,
      "learning_rate": 0.0002186332812968352,
      "loss": 1.7969,
      "step": 4462
    },
    {
      "epoch": 2.9556291390728475,
      "grad_norm": 0.7489062805397397,
      "learning_rate": 0.0002185739236856652,
      "loss": 2.0781,
      "step": 4463
    },
    {
      "epoch": 2.956291390728477,
      "grad_norm": 0.7525872556271725,
      "learning_rate": 0.00021851455249607217,
      "loss": 2.0469,
      "step": 4464
    },
    {
      "epoch": 2.956953642384106,
      "grad_norm": 0.7351881522733897,
      "learning_rate": 0.00021845516773981224,
      "loss": 1.7969,
      "step": 4465
    },
    {
      "epoch": 2.9576158940397352,
      "grad_norm": 0.7571555038065063,
      "learning_rate": 0.00021839576942864428,
      "loss": 1.8438,
      "step": 4466
    },
    {
      "epoch": 2.958278145695364,
      "grad_norm": 0.8291127019506365,
      "learning_rate": 0.0002183363575743298,
      "loss": 1.7344,
      "step": 4467
    },
    {
      "epoch": 2.958940397350993,
      "grad_norm": 0.7421690997261874,
      "learning_rate": 0.00021827693218863314,
      "loss": 2.0781,
      "step": 4468
    },
    {
      "epoch": 2.9596026490066225,
      "grad_norm": 0.7572631436657897,
      "learning_rate": 0.0002182174932833211,
      "loss": 2.1562,
      "step": 4469
    },
    {
      "epoch": 2.960264900662252,
      "grad_norm": 0.8328869960295198,
      "learning_rate": 0.0002181580408701633,
      "loss": 2.3594,
      "step": 4470
    },
    {
      "epoch": 2.960927152317881,
      "grad_norm": 0.8274603927309834,
      "learning_rate": 0.00021809857496093199,
      "loss": 2.1406,
      "step": 4471
    },
    {
      "epoch": 2.96158940397351,
      "grad_norm": 0.8031250321268061,
      "learning_rate": 0.00021803909556740214,
      "loss": 2.1719,
      "step": 4472
    },
    {
      "epoch": 2.962251655629139,
      "grad_norm": 0.7429841355674442,
      "learning_rate": 0.00021797960270135132,
      "loss": 2.0781,
      "step": 4473
    },
    {
      "epoch": 2.962913907284768,
      "grad_norm": 0.757421386297501,
      "learning_rate": 0.00021792009637455984,
      "loss": 2.0781,
      "step": 4474
    },
    {
      "epoch": 2.9635761589403975,
      "grad_norm": 0.7973023639367308,
      "learning_rate": 0.0002178605765988105,
      "loss": 2.3906,
      "step": 4475
    },
    {
      "epoch": 2.9642384105960264,
      "grad_norm": 0.794854240603579,
      "learning_rate": 0.00021780104338588906,
      "loss": 2.1562,
      "step": 4476
    },
    {
      "epoch": 2.9649006622516554,
      "grad_norm": 0.7059275976064484,
      "learning_rate": 0.00021774149674758371,
      "loss": 1.7188,
      "step": 4477
    },
    {
      "epoch": 2.9655629139072848,
      "grad_norm": 0.7881509581186316,
      "learning_rate": 0.00021768193669568538,
      "loss": 2.1875,
      "step": 4478
    },
    {
      "epoch": 2.966225165562914,
      "grad_norm": 0.77250826076815,
      "learning_rate": 0.00021762236324198762,
      "loss": 2.0938,
      "step": 4479
    },
    {
      "epoch": 2.966887417218543,
      "grad_norm": 0.7604436987816658,
      "learning_rate": 0.0002175627763982867,
      "loss": 1.9375,
      "step": 4480
    },
    {
      "epoch": 2.967549668874172,
      "grad_norm": 0.7646219472725809,
      "learning_rate": 0.00021750317617638147,
      "loss": 2.2188,
      "step": 4481
    },
    {
      "epoch": 2.9682119205298014,
      "grad_norm": 0.8201232309200533,
      "learning_rate": 0.00021744356258807343,
      "loss": 2.0938,
      "step": 4482
    },
    {
      "epoch": 2.9688741721854304,
      "grad_norm": 0.7506324416273996,
      "learning_rate": 0.00021738393564516682,
      "loss": 2.0156,
      "step": 4483
    },
    {
      "epoch": 2.9695364238410598,
      "grad_norm": 0.726410201903873,
      "learning_rate": 0.00021732429535946838,
      "loss": 2.0312,
      "step": 4484
    },
    {
      "epoch": 2.9701986754966887,
      "grad_norm": 0.7807045033605472,
      "learning_rate": 0.00021726464174278763,
      "loss": 2.0,
      "step": 4485
    },
    {
      "epoch": 2.9708609271523176,
      "grad_norm": 0.7726129997291804,
      "learning_rate": 0.00021720497480693664,
      "loss": 1.7578,
      "step": 4486
    },
    {
      "epoch": 2.971523178807947,
      "grad_norm": 0.8420518665642285,
      "learning_rate": 0.00021714529456373015,
      "loss": 2.0469,
      "step": 4487
    },
    {
      "epoch": 2.9721854304635764,
      "grad_norm": 0.8070035576329213,
      "learning_rate": 0.00021708560102498556,
      "loss": 2.1875,
      "step": 4488
    },
    {
      "epoch": 2.9728476821192054,
      "grad_norm": 0.7524861357345887,
      "learning_rate": 0.00021702589420252288,
      "loss": 2.0469,
      "step": 4489
    },
    {
      "epoch": 2.9735099337748343,
      "grad_norm": 0.7677955518242366,
      "learning_rate": 0.00021696617410816465,
      "loss": 2.0312,
      "step": 4490
    },
    {
      "epoch": 2.9741721854304637,
      "grad_norm": 0.7504578366731383,
      "learning_rate": 0.0002169064407537363,
      "loss": 2.0469,
      "step": 4491
    },
    {
      "epoch": 2.9748344370860926,
      "grad_norm": 0.8033570538067956,
      "learning_rate": 0.00021684669415106554,
      "loss": 2.2656,
      "step": 4492
    },
    {
      "epoch": 2.975496688741722,
      "grad_norm": 0.7635119197812412,
      "learning_rate": 0.000216786934311983,
      "loss": 2.1094,
      "step": 4493
    },
    {
      "epoch": 2.976158940397351,
      "grad_norm": 0.7896517448651882,
      "learning_rate": 0.0002167271612483218,
      "loss": 2.5156,
      "step": 4494
    },
    {
      "epoch": 2.97682119205298,
      "grad_norm": 0.7779784166007824,
      "learning_rate": 0.00021666737497191767,
      "loss": 2.2031,
      "step": 4495
    },
    {
      "epoch": 2.9774834437086093,
      "grad_norm": 0.7818042099696694,
      "learning_rate": 0.00021660757549460893,
      "loss": 2.1406,
      "step": 4496
    },
    {
      "epoch": 2.9781456953642387,
      "grad_norm": 0.8344879510920917,
      "learning_rate": 0.00021654776282823663,
      "loss": 2.0469,
      "step": 4497
    },
    {
      "epoch": 2.9788079470198676,
      "grad_norm": 0.7479499473487909,
      "learning_rate": 0.00021648793698464435,
      "loss": 1.9922,
      "step": 4498
    },
    {
      "epoch": 2.9794701986754966,
      "grad_norm": 0.7103081948807429,
      "learning_rate": 0.00021642809797567828,
      "loss": 2.0781,
      "step": 4499
    },
    {
      "epoch": 2.980132450331126,
      "grad_norm": 0.7847505325973558,
      "learning_rate": 0.00021636824581318726,
      "loss": 1.7891,
      "step": 4500
    },
    {
      "epoch": 2.980794701986755,
      "grad_norm": 0.7530241136721917,
      "learning_rate": 0.0002163083805090227,
      "loss": 2.0781,
      "step": 4501
    },
    {
      "epoch": 2.9814569536423843,
      "grad_norm": 0.8347279485319191,
      "learning_rate": 0.00021624850207503852,
      "loss": 2.3906,
      "step": 4502
    },
    {
      "epoch": 2.982119205298013,
      "grad_norm": 0.7465069649434877,
      "learning_rate": 0.0002161886105230915,
      "loss": 2.0156,
      "step": 4503
    },
    {
      "epoch": 2.982781456953642,
      "grad_norm": 0.7524288090210923,
      "learning_rate": 0.0002161287058650407,
      "loss": 2.0469,
      "step": 4504
    },
    {
      "epoch": 2.9834437086092715,
      "grad_norm": 0.7707613736899056,
      "learning_rate": 0.00021606878811274806,
      "loss": 2.0312,
      "step": 4505
    },
    {
      "epoch": 2.984105960264901,
      "grad_norm": 0.7549465315258881,
      "learning_rate": 0.00021600885727807788,
      "loss": 2.0,
      "step": 4506
    },
    {
      "epoch": 2.98476821192053,
      "grad_norm": 0.793532980187094,
      "learning_rate": 0.0002159489133728972,
      "loss": 2.1562,
      "step": 4507
    },
    {
      "epoch": 2.985430463576159,
      "grad_norm": 0.7851455886902055,
      "learning_rate": 0.00021588895640907557,
      "loss": 1.9922,
      "step": 4508
    },
    {
      "epoch": 2.986092715231788,
      "grad_norm": 0.8073792135091857,
      "learning_rate": 0.00021582898639848524,
      "loss": 2.2656,
      "step": 4509
    },
    {
      "epoch": 2.986754966887417,
      "grad_norm": 0.80474386683991,
      "learning_rate": 0.00021576900335300086,
      "loss": 1.9297,
      "step": 4510
    },
    {
      "epoch": 2.9874172185430465,
      "grad_norm": 0.7546028011285714,
      "learning_rate": 0.00021570900728449976,
      "loss": 2.1406,
      "step": 4511
    },
    {
      "epoch": 2.9880794701986755,
      "grad_norm": 0.7605544670390771,
      "learning_rate": 0.00021564899820486195,
      "loss": 1.8438,
      "step": 4512
    },
    {
      "epoch": 2.9887417218543044,
      "grad_norm": 0.8689870558522502,
      "learning_rate": 0.00021558897612596983,
      "loss": 2.3125,
      "step": 4513
    },
    {
      "epoch": 2.989403973509934,
      "grad_norm": 0.8289683484887721,
      "learning_rate": 0.0002155289410597085,
      "loss": 2.0781,
      "step": 4514
    },
    {
      "epoch": 2.9900662251655628,
      "grad_norm": 0.7891927341014031,
      "learning_rate": 0.00021546889301796556,
      "loss": 2.3906,
      "step": 4515
    },
    {
      "epoch": 2.990728476821192,
      "grad_norm": 0.7293865657255246,
      "learning_rate": 0.0002154088320126312,
      "loss": 1.9844,
      "step": 4516
    },
    {
      "epoch": 2.991390728476821,
      "grad_norm": 0.7793330282525184,
      "learning_rate": 0.00021534875805559825,
      "loss": 2.0938,
      "step": 4517
    },
    {
      "epoch": 2.9920529801324505,
      "grad_norm": 0.7307010919065026,
      "learning_rate": 0.00021528867115876203,
      "loss": 1.8438,
      "step": 4518
    },
    {
      "epoch": 2.9927152317880794,
      "grad_norm": 0.8255527419719854,
      "learning_rate": 0.00021522857133402033,
      "loss": 2.2188,
      "step": 4519
    },
    {
      "epoch": 2.993377483443709,
      "grad_norm": 0.7635729103734045,
      "learning_rate": 0.0002151684585932738,
      "loss": 1.6406,
      "step": 4520
    },
    {
      "epoch": 2.9940397350993377,
      "grad_norm": 0.7534430832238876,
      "learning_rate": 0.00021510833294842522,
      "loss": 2.0938,
      "step": 4521
    },
    {
      "epoch": 2.9947019867549667,
      "grad_norm": 0.7715012642016528,
      "learning_rate": 0.00021504819441138032,
      "loss": 2.1094,
      "step": 4522
    },
    {
      "epoch": 2.995364238410596,
      "grad_norm": 0.8744439778229881,
      "learning_rate": 0.0002149880429940472,
      "loss": 2.3438,
      "step": 4523
    },
    {
      "epoch": 2.996026490066225,
      "grad_norm": 0.7186175424265849,
      "learning_rate": 0.0002149278787083365,
      "loss": 2.0469,
      "step": 4524
    },
    {
      "epoch": 2.9966887417218544,
      "grad_norm": 0.6673242627878568,
      "learning_rate": 0.00021486770156616138,
      "loss": 1.6719,
      "step": 4525
    },
    {
      "epoch": 2.9973509933774833,
      "grad_norm": 0.737447693511495,
      "learning_rate": 0.00021480751157943777,
      "loss": 2.2188,
      "step": 4526
    },
    {
      "epoch": 2.9980132450331127,
      "grad_norm": 0.7752458871831162,
      "learning_rate": 0.00021474730876008377,
      "loss": 2.1406,
      "step": 4527
    },
    {
      "epoch": 2.9986754966887417,
      "grad_norm": 0.7023238930648998,
      "learning_rate": 0.00021468709312002035,
      "loss": 2.1094,
      "step": 4528
    },
    {
      "epoch": 2.999337748344371,
      "grad_norm": 0.7547587852404817,
      "learning_rate": 0.00021462686467117094,
      "loss": 2.125,
      "step": 4529
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7491532614618133,
      "learning_rate": 0.00021456662342546137,
      "loss": 2.0781,
      "step": 4530
    },
    {
      "epoch": 3.0,
      "eval_loss": 2.2021772861480713,
      "eval_runtime": 33.9942,
      "eval_samples_per_second": 9.943,
      "eval_steps_per_second": 9.943,
      "step": 4530
    },
    {
      "epoch": 3.000662251655629,
      "grad_norm": 0.7131464399924161,
      "learning_rate": 0.00021450636939482007,
      "loss": 1.8281,
      "step": 4531
    },
    {
      "epoch": 3.0013245033112583,
      "grad_norm": 0.7315825488786015,
      "learning_rate": 0.0002144461025911782,
      "loss": 1.8438,
      "step": 4532
    },
    {
      "epoch": 3.0019867549668873,
      "grad_norm": 0.7627443914973644,
      "learning_rate": 0.00021438582302646907,
      "loss": 1.8984,
      "step": 4533
    },
    {
      "epoch": 3.0026490066225167,
      "grad_norm": 0.6372938112303863,
      "learning_rate": 0.00021432553071262886,
      "loss": 1.3594,
      "step": 4534
    },
    {
      "epoch": 3.0033112582781456,
      "grad_norm": 0.7312805323907999,
      "learning_rate": 0.0002142652256615961,
      "loss": 1.7031,
      "step": 4535
    },
    {
      "epoch": 3.003973509933775,
      "grad_norm": 0.8102738333912007,
      "learning_rate": 0.0002142049078853119,
      "loss": 1.7188,
      "step": 4536
    },
    {
      "epoch": 3.004635761589404,
      "grad_norm": 0.733655360612381,
      "learning_rate": 0.00021414457739571985,
      "loss": 1.5938,
      "step": 4537
    },
    {
      "epoch": 3.0052980132450333,
      "grad_norm": 0.8222375491316457,
      "learning_rate": 0.00021408423420476606,
      "loss": 1.8047,
      "step": 4538
    },
    {
      "epoch": 3.0059602649006623,
      "grad_norm": 0.9004397607782434,
      "learning_rate": 0.00021402387832439922,
      "loss": 1.75,
      "step": 4539
    },
    {
      "epoch": 3.006622516556291,
      "grad_norm": 0.9069935129797175,
      "learning_rate": 0.0002139635097665704,
      "loss": 1.9453,
      "step": 4540
    },
    {
      "epoch": 3.0072847682119206,
      "grad_norm": 0.8466033725269645,
      "learning_rate": 0.0002139031285432334,
      "loss": 1.6797,
      "step": 4541
    },
    {
      "epoch": 3.0079470198675495,
      "grad_norm": 0.8723546670868306,
      "learning_rate": 0.00021384273466634428,
      "loss": 1.7656,
      "step": 4542
    },
    {
      "epoch": 3.008609271523179,
      "grad_norm": 0.8375098653039175,
      "learning_rate": 0.00021378232814786178,
      "loss": 1.6797,
      "step": 4543
    },
    {
      "epoch": 3.009271523178808,
      "grad_norm": 0.6548608054206255,
      "learning_rate": 0.0002137219089997471,
      "loss": 1.0469,
      "step": 4544
    },
    {
      "epoch": 3.0099337748344372,
      "grad_norm": 0.9014710478179863,
      "learning_rate": 0.00021366147723396385,
      "loss": 1.9531,
      "step": 4545
    },
    {
      "epoch": 3.010596026490066,
      "grad_norm": 0.8049453506205523,
      "learning_rate": 0.00021360103286247826,
      "loss": 1.4375,
      "step": 4546
    },
    {
      "epoch": 3.0112582781456956,
      "grad_norm": 0.8959935929308002,
      "learning_rate": 0.000213540575897259,
      "loss": 1.9688,
      "step": 4547
    },
    {
      "epoch": 3.0119205298013245,
      "grad_norm": 0.7435247641972823,
      "learning_rate": 0.00021348010635027724,
      "loss": 1.3125,
      "step": 4548
    },
    {
      "epoch": 3.0125827814569535,
      "grad_norm": 0.923209184977155,
      "learning_rate": 0.00021341962423350667,
      "loss": 1.8594,
      "step": 4549
    },
    {
      "epoch": 3.013245033112583,
      "grad_norm": 0.9202244523983142,
      "learning_rate": 0.0002133591295589234,
      "loss": 1.9375,
      "step": 4550
    },
    {
      "epoch": 3.013907284768212,
      "grad_norm": 0.844216680828011,
      "learning_rate": 0.00021329862233850606,
      "loss": 1.7031,
      "step": 4551
    },
    {
      "epoch": 3.014569536423841,
      "grad_norm": 0.8039576780111451,
      "learning_rate": 0.00021323810258423585,
      "loss": 1.625,
      "step": 4552
    },
    {
      "epoch": 3.01523178807947,
      "grad_norm": 0.8756934912306449,
      "learning_rate": 0.00021317757030809632,
      "loss": 1.8281,
      "step": 4553
    },
    {
      "epoch": 3.0158940397350995,
      "grad_norm": 0.8519920743999857,
      "learning_rate": 0.00021311702552207352,
      "loss": 1.8672,
      "step": 4554
    },
    {
      "epoch": 3.0165562913907285,
      "grad_norm": 0.8499042001838856,
      "learning_rate": 0.0002130564682381561,
      "loss": 1.5703,
      "step": 4555
    },
    {
      "epoch": 3.017218543046358,
      "grad_norm": 0.8222492742116871,
      "learning_rate": 0.000212995898468335,
      "loss": 1.5312,
      "step": 4556
    },
    {
      "epoch": 3.017880794701987,
      "grad_norm": 0.8052348626261567,
      "learning_rate": 0.00021293531622460386,
      "loss": 1.6094,
      "step": 4557
    },
    {
      "epoch": 3.0185430463576157,
      "grad_norm": 0.8028596088386369,
      "learning_rate": 0.00021287472151895853,
      "loss": 1.6484,
      "step": 4558
    },
    {
      "epoch": 3.019205298013245,
      "grad_norm": 0.8643759578492328,
      "learning_rate": 0.0002128141143633975,
      "loss": 2.0,
      "step": 4559
    },
    {
      "epoch": 3.019867549668874,
      "grad_norm": 0.9014997843574897,
      "learning_rate": 0.00021275349476992166,
      "loss": 1.7969,
      "step": 4560
    },
    {
      "epoch": 3.0205298013245034,
      "grad_norm": 0.7359853546995025,
      "learning_rate": 0.00021269286275053446,
      "loss": 1.3906,
      "step": 4561
    },
    {
      "epoch": 3.0211920529801324,
      "grad_norm": 0.884493549163444,
      "learning_rate": 0.00021263221831724165,
      "loss": 1.8438,
      "step": 4562
    },
    {
      "epoch": 3.0218543046357618,
      "grad_norm": 0.8785724865075689,
      "learning_rate": 0.00021257156148205158,
      "loss": 1.8828,
      "step": 4563
    },
    {
      "epoch": 3.0225165562913907,
      "grad_norm": 0.8061396655751379,
      "learning_rate": 0.00021251089225697497,
      "loss": 1.5469,
      "step": 4564
    },
    {
      "epoch": 3.0231788079470197,
      "grad_norm": 0.8747469273803027,
      "learning_rate": 0.00021245021065402503,
      "loss": 1.6719,
      "step": 4565
    },
    {
      "epoch": 3.023841059602649,
      "grad_norm": 0.9443017460997651,
      "learning_rate": 0.0002123895166852174,
      "loss": 1.9375,
      "step": 4566
    },
    {
      "epoch": 3.024503311258278,
      "grad_norm": 0.7679332518947736,
      "learning_rate": 0.00021232881036257022,
      "loss": 1.4688,
      "step": 4567
    },
    {
      "epoch": 3.0251655629139074,
      "grad_norm": 0.8406131648650397,
      "learning_rate": 0.000212268091698104,
      "loss": 1.6016,
      "step": 4568
    },
    {
      "epoch": 3.0258278145695363,
      "grad_norm": 0.9384893661620362,
      "learning_rate": 0.00021220736070384172,
      "loss": 1.9375,
      "step": 4569
    },
    {
      "epoch": 3.0264900662251657,
      "grad_norm": 0.7703089869103229,
      "learning_rate": 0.00021214661739180884,
      "loss": 1.3906,
      "step": 4570
    },
    {
      "epoch": 3.0271523178807946,
      "grad_norm": 0.8066307329530482,
      "learning_rate": 0.00021208586177403328,
      "loss": 1.4844,
      "step": 4571
    },
    {
      "epoch": 3.027814569536424,
      "grad_norm": 1.1132677778480813,
      "learning_rate": 0.00021202509386254525,
      "loss": 1.9219,
      "step": 4572
    },
    {
      "epoch": 3.028476821192053,
      "grad_norm": 0.890771902168676,
      "learning_rate": 0.0002119643136693776,
      "loss": 1.8359,
      "step": 4573
    },
    {
      "epoch": 3.029139072847682,
      "grad_norm": 0.7472995897435486,
      "learning_rate": 0.0002119035212065654,
      "loss": 1.2656,
      "step": 4574
    },
    {
      "epoch": 3.0298013245033113,
      "grad_norm": 0.8828345557916857,
      "learning_rate": 0.00021184271648614625,
      "loss": 1.7969,
      "step": 4575
    },
    {
      "epoch": 3.0304635761589402,
      "grad_norm": 0.841822873400544,
      "learning_rate": 0.00021178189952016034,
      "loss": 1.6328,
      "step": 4576
    },
    {
      "epoch": 3.0311258278145696,
      "grad_norm": 0.9198778704679523,
      "learning_rate": 0.0002117210703206499,
      "loss": 1.7891,
      "step": 4577
    },
    {
      "epoch": 3.0317880794701986,
      "grad_norm": 0.8884872629891888,
      "learning_rate": 0.00021166022889966004,
      "loss": 1.8438,
      "step": 4578
    },
    {
      "epoch": 3.032450331125828,
      "grad_norm": 0.7570907170547456,
      "learning_rate": 0.00021159937526923786,
      "loss": 1.3281,
      "step": 4579
    },
    {
      "epoch": 3.033112582781457,
      "grad_norm": 0.7928908586603545,
      "learning_rate": 0.00021153850944143323,
      "loss": 1.6953,
      "step": 4580
    },
    {
      "epoch": 3.0337748344370863,
      "grad_norm": 0.8569221305524763,
      "learning_rate": 0.00021147763142829817,
      "loss": 1.7969,
      "step": 4581
    },
    {
      "epoch": 3.0344370860927152,
      "grad_norm": 0.8089661198031223,
      "learning_rate": 0.00021141674124188728,
      "loss": 1.4219,
      "step": 4582
    },
    {
      "epoch": 3.035099337748344,
      "grad_norm": 0.973653013340457,
      "learning_rate": 0.00021135583889425748,
      "loss": 1.8516,
      "step": 4583
    },
    {
      "epoch": 3.0357615894039736,
      "grad_norm": 0.9640560781507477,
      "learning_rate": 0.0002112949243974682,
      "loss": 1.9922,
      "step": 4584
    },
    {
      "epoch": 3.0364238410596025,
      "grad_norm": 0.7471467931816759,
      "learning_rate": 0.00021123399776358114,
      "loss": 1.4141,
      "step": 4585
    },
    {
      "epoch": 3.037086092715232,
      "grad_norm": 0.8675873300238508,
      "learning_rate": 0.00021117305900466052,
      "loss": 1.7422,
      "step": 4586
    },
    {
      "epoch": 3.037748344370861,
      "grad_norm": 0.7666140223996569,
      "learning_rate": 0.0002111121081327729,
      "loss": 1.3984,
      "step": 4587
    },
    {
      "epoch": 3.03841059602649,
      "grad_norm": 0.9673253935338466,
      "learning_rate": 0.00021105114515998722,
      "loss": 1.7578,
      "step": 4588
    },
    {
      "epoch": 3.039072847682119,
      "grad_norm": 0.8238607407814207,
      "learning_rate": 0.00021099017009837485,
      "loss": 1.4141,
      "step": 4589
    },
    {
      "epoch": 3.0397350993377485,
      "grad_norm": 0.799145687643931,
      "learning_rate": 0.0002109291829600096,
      "loss": 1.3594,
      "step": 4590
    },
    {
      "epoch": 3.0403973509933775,
      "grad_norm": 0.9106099219160904,
      "learning_rate": 0.00021086818375696756,
      "loss": 1.6328,
      "step": 4591
    },
    {
      "epoch": 3.0410596026490064,
      "grad_norm": 0.8384710172189688,
      "learning_rate": 0.00021080717250132735,
      "loss": 1.2969,
      "step": 4592
    },
    {
      "epoch": 3.041721854304636,
      "grad_norm": 0.8832237967610664,
      "learning_rate": 0.0002107461492051698,
      "loss": 1.4531,
      "step": 4593
    },
    {
      "epoch": 3.0423841059602648,
      "grad_norm": 0.9935895311907262,
      "learning_rate": 0.00021068511388057837,
      "loss": 1.7266,
      "step": 4594
    },
    {
      "epoch": 3.043046357615894,
      "grad_norm": 0.8499970704099936,
      "learning_rate": 0.0002106240665396385,
      "loss": 1.3984,
      "step": 4595
    },
    {
      "epoch": 3.043708609271523,
      "grad_norm": 0.9769769336712841,
      "learning_rate": 0.0002105630071944386,
      "loss": 1.6328,
      "step": 4596
    },
    {
      "epoch": 3.0443708609271525,
      "grad_norm": 1.0832507525221649,
      "learning_rate": 0.00021050193585706882,
      "loss": 2.0156,
      "step": 4597
    },
    {
      "epoch": 3.0450331125827814,
      "grad_norm": 0.9635711526587869,
      "learning_rate": 0.00021044085253962217,
      "loss": 1.6797,
      "step": 4598
    },
    {
      "epoch": 3.045695364238411,
      "grad_norm": 0.8859375874305845,
      "learning_rate": 0.00021037975725419377,
      "loss": 1.4844,
      "step": 4599
    },
    {
      "epoch": 3.0463576158940397,
      "grad_norm": 0.8787299114979397,
      "learning_rate": 0.0002103186500128812,
      "loss": 1.5859,
      "step": 4600
    },
    {
      "epoch": 3.0470198675496687,
      "grad_norm": 0.8284845545333706,
      "learning_rate": 0.00021025753082778441,
      "loss": 1.6172,
      "step": 4601
    },
    {
      "epoch": 3.047682119205298,
      "grad_norm": 0.900813420277025,
      "learning_rate": 0.00021019639971100567,
      "loss": 1.7031,
      "step": 4602
    },
    {
      "epoch": 3.048344370860927,
      "grad_norm": 0.8666937738309455,
      "learning_rate": 0.00021013525667464965,
      "loss": 1.6016,
      "step": 4603
    },
    {
      "epoch": 3.0490066225165564,
      "grad_norm": 0.7283491076238269,
      "learning_rate": 0.00021007410173082337,
      "loss": 1.3984,
      "step": 4604
    },
    {
      "epoch": 3.0496688741721854,
      "grad_norm": 0.9705945905861213,
      "learning_rate": 0.00021001293489163628,
      "loss": 1.7656,
      "step": 4605
    },
    {
      "epoch": 3.0503311258278147,
      "grad_norm": 0.7777081469681147,
      "learning_rate": 0.0002099517561692,
      "loss": 1.4219,
      "step": 4606
    },
    {
      "epoch": 3.0509933774834437,
      "grad_norm": 0.9182666753062396,
      "learning_rate": 0.00020989056557562871,
      "loss": 1.8359,
      "step": 4607
    },
    {
      "epoch": 3.0516556291390726,
      "grad_norm": 0.9007214848063517,
      "learning_rate": 0.00020982936312303877,
      "loss": 1.6719,
      "step": 4608
    },
    {
      "epoch": 3.052317880794702,
      "grad_norm": 0.9561937945465591,
      "learning_rate": 0.00020976814882354907,
      "loss": 1.625,
      "step": 4609
    },
    {
      "epoch": 3.052980132450331,
      "grad_norm": 0.9039706530854671,
      "learning_rate": 0.00020970692268928063,
      "loss": 1.5781,
      "step": 4610
    },
    {
      "epoch": 3.0536423841059603,
      "grad_norm": 0.8564175378020064,
      "learning_rate": 0.00020964568473235703,
      "loss": 1.4375,
      "step": 4611
    },
    {
      "epoch": 3.0543046357615893,
      "grad_norm": 0.9223629094108217,
      "learning_rate": 0.00020958443496490396,
      "loss": 1.7734,
      "step": 4612
    },
    {
      "epoch": 3.0549668874172187,
      "grad_norm": 0.9474363510730253,
      "learning_rate": 0.00020952317339904974,
      "loss": 1.8359,
      "step": 4613
    },
    {
      "epoch": 3.0556291390728476,
      "grad_norm": 0.9071396190587373,
      "learning_rate": 0.0002094619000469247,
      "loss": 1.7578,
      "step": 4614
    },
    {
      "epoch": 3.056291390728477,
      "grad_norm": 0.8504704858586286,
      "learning_rate": 0.0002094006149206618,
      "loss": 1.4766,
      "step": 4615
    },
    {
      "epoch": 3.056953642384106,
      "grad_norm": 0.7386474272079273,
      "learning_rate": 0.0002093393180323961,
      "loss": 1.2812,
      "step": 4616
    },
    {
      "epoch": 3.057615894039735,
      "grad_norm": 0.8537485674543925,
      "learning_rate": 0.00020927800939426518,
      "loss": 1.7031,
      "step": 4617
    },
    {
      "epoch": 3.0582781456953643,
      "grad_norm": 0.8304351279541017,
      "learning_rate": 0.00020921668901840873,
      "loss": 1.5078,
      "step": 4618
    },
    {
      "epoch": 3.058940397350993,
      "grad_norm": 0.9152434371276499,
      "learning_rate": 0.000209155356916969,
      "loss": 1.5,
      "step": 4619
    },
    {
      "epoch": 3.0596026490066226,
      "grad_norm": 1.0246764449055963,
      "learning_rate": 0.00020909401310209038,
      "loss": 2.125,
      "step": 4620
    },
    {
      "epoch": 3.0602649006622515,
      "grad_norm": 0.8619681759175868,
      "learning_rate": 0.00020903265758591972,
      "loss": 1.4609,
      "step": 4621
    },
    {
      "epoch": 3.060927152317881,
      "grad_norm": 0.986774357536471,
      "learning_rate": 0.00020897129038060602,
      "loss": 1.8984,
      "step": 4622
    },
    {
      "epoch": 3.06158940397351,
      "grad_norm": 0.8832867703865735,
      "learning_rate": 0.0002089099114983008,
      "loss": 1.7422,
      "step": 4623
    },
    {
      "epoch": 3.0622516556291393,
      "grad_norm": 1.1136145066037855,
      "learning_rate": 0.00020884852095115762,
      "loss": 2.0156,
      "step": 4624
    },
    {
      "epoch": 3.062913907284768,
      "grad_norm": 0.7701724040348383,
      "learning_rate": 0.0002087871187513327,
      "loss": 1.2344,
      "step": 4625
    },
    {
      "epoch": 3.063576158940397,
      "grad_norm": 1.0510210699670206,
      "learning_rate": 0.00020872570491098428,
      "loss": 1.8984,
      "step": 4626
    },
    {
      "epoch": 3.0642384105960265,
      "grad_norm": 0.7827687804137577,
      "learning_rate": 0.00020866427944227304,
      "loss": 1.3906,
      "step": 4627
    },
    {
      "epoch": 3.0649006622516555,
      "grad_norm": 0.9443186185881881,
      "learning_rate": 0.00020860284235736188,
      "loss": 2.0156,
      "step": 4628
    },
    {
      "epoch": 3.065562913907285,
      "grad_norm": 0.8229279260942582,
      "learning_rate": 0.00020854139366841606,
      "loss": 1.5938,
      "step": 4629
    },
    {
      "epoch": 3.066225165562914,
      "grad_norm": 0.7990088074615567,
      "learning_rate": 0.00020847993338760318,
      "loss": 1.3672,
      "step": 4630
    },
    {
      "epoch": 3.066887417218543,
      "grad_norm": 0.8550011324768562,
      "learning_rate": 0.000208418461527093,
      "loss": 1.7188,
      "step": 4631
    },
    {
      "epoch": 3.067549668874172,
      "grad_norm": 0.8590525042001437,
      "learning_rate": 0.00020835697809905773,
      "loss": 1.5,
      "step": 4632
    },
    {
      "epoch": 3.0682119205298015,
      "grad_norm": 0.8336930317846128,
      "learning_rate": 0.00020829548311567168,
      "loss": 1.5391,
      "step": 4633
    },
    {
      "epoch": 3.0688741721854305,
      "grad_norm": 1.012098809369901,
      "learning_rate": 0.00020823397658911169,
      "loss": 1.8359,
      "step": 4634
    },
    {
      "epoch": 3.0695364238410594,
      "grad_norm": 0.963966179292962,
      "learning_rate": 0.0002081724585315567,
      "loss": 1.8125,
      "step": 4635
    },
    {
      "epoch": 3.070198675496689,
      "grad_norm": 0.9021894892752844,
      "learning_rate": 0.000208110928955188,
      "loss": 1.7969,
      "step": 4636
    },
    {
      "epoch": 3.0708609271523177,
      "grad_norm": 0.780321266236924,
      "learning_rate": 0.00020804938787218917,
      "loss": 1.1953,
      "step": 4637
    },
    {
      "epoch": 3.071523178807947,
      "grad_norm": 0.9471282440675518,
      "learning_rate": 0.000207987835294746,
      "loss": 1.7656,
      "step": 4638
    },
    {
      "epoch": 3.072185430463576,
      "grad_norm": 0.8910302017289745,
      "learning_rate": 0.00020792627123504667,
      "loss": 1.4531,
      "step": 4639
    },
    {
      "epoch": 3.0728476821192054,
      "grad_norm": 0.9561560252734238,
      "learning_rate": 0.00020786469570528148,
      "loss": 1.7656,
      "step": 4640
    },
    {
      "epoch": 3.0735099337748344,
      "grad_norm": 0.8374932546815229,
      "learning_rate": 0.00020780310871764313,
      "loss": 1.3125,
      "step": 4641
    },
    {
      "epoch": 3.0741721854304638,
      "grad_norm": 0.9404439357163489,
      "learning_rate": 0.00020774151028432662,
      "loss": 1.7812,
      "step": 4642
    },
    {
      "epoch": 3.0748344370860927,
      "grad_norm": 0.810624391049294,
      "learning_rate": 0.00020767990041752906,
      "loss": 1.375,
      "step": 4643
    },
    {
      "epoch": 3.0754966887417217,
      "grad_norm": 0.8775971798246673,
      "learning_rate": 0.00020761827912944993,
      "loss": 1.4531,
      "step": 4644
    },
    {
      "epoch": 3.076158940397351,
      "grad_norm": 0.8160371525473371,
      "learning_rate": 0.00020755664643229099,
      "loss": 1.4453,
      "step": 4645
    },
    {
      "epoch": 3.07682119205298,
      "grad_norm": 0.9290192120049136,
      "learning_rate": 0.00020749500233825617,
      "loss": 1.4922,
      "step": 4646
    },
    {
      "epoch": 3.0774834437086094,
      "grad_norm": 0.9942809664727316,
      "learning_rate": 0.0002074333468595517,
      "loss": 1.8438,
      "step": 4647
    },
    {
      "epoch": 3.0781456953642383,
      "grad_norm": 0.9910832449205357,
      "learning_rate": 0.00020737168000838617,
      "loss": 1.625,
      "step": 4648
    },
    {
      "epoch": 3.0788079470198677,
      "grad_norm": 0.9002561202049684,
      "learning_rate": 0.00020731000179697017,
      "loss": 1.5469,
      "step": 4649
    },
    {
      "epoch": 3.0794701986754967,
      "grad_norm": 0.8982270775150784,
      "learning_rate": 0.00020724831223751684,
      "loss": 1.75,
      "step": 4650
    },
    {
      "epoch": 3.080132450331126,
      "grad_norm": 1.0866889880308153,
      "learning_rate": 0.00020718661134224136,
      "loss": 1.8672,
      "step": 4651
    },
    {
      "epoch": 3.080794701986755,
      "grad_norm": 0.933369095818204,
      "learning_rate": 0.00020712489912336118,
      "loss": 1.6562,
      "step": 4652
    },
    {
      "epoch": 3.081456953642384,
      "grad_norm": 1.0057196713202659,
      "learning_rate": 0.00020706317559309603,
      "loss": 1.7734,
      "step": 4653
    },
    {
      "epoch": 3.0821192052980133,
      "grad_norm": 0.8813842125980536,
      "learning_rate": 0.00020700144076366797,
      "loss": 1.4688,
      "step": 4654
    },
    {
      "epoch": 3.0827814569536423,
      "grad_norm": 0.9609692115211225,
      "learning_rate": 0.00020693969464730108,
      "loss": 1.7188,
      "step": 4655
    },
    {
      "epoch": 3.0834437086092716,
      "grad_norm": 0.8972580430866657,
      "learning_rate": 0.00020687793725622188,
      "loss": 1.5859,
      "step": 4656
    },
    {
      "epoch": 3.0841059602649006,
      "grad_norm": 0.9109002493638131,
      "learning_rate": 0.00020681616860265906,
      "loss": 1.6641,
      "step": 4657
    },
    {
      "epoch": 3.08476821192053,
      "grad_norm": 0.8807301057891866,
      "learning_rate": 0.00020675438869884342,
      "loss": 1.6875,
      "step": 4658
    },
    {
      "epoch": 3.085430463576159,
      "grad_norm": 0.9106604670290821,
      "learning_rate": 0.00020669259755700818,
      "loss": 1.7422,
      "step": 4659
    },
    {
      "epoch": 3.0860927152317883,
      "grad_norm": 0.8628574401047973,
      "learning_rate": 0.00020663079518938867,
      "loss": 1.5703,
      "step": 4660
    },
    {
      "epoch": 3.0867549668874172,
      "grad_norm": 1.0409016032812368,
      "learning_rate": 0.00020656898160822249,
      "loss": 2.0625,
      "step": 4661
    },
    {
      "epoch": 3.087417218543046,
      "grad_norm": 0.8719933766753399,
      "learning_rate": 0.00020650715682574936,
      "loss": 1.6562,
      "step": 4662
    },
    {
      "epoch": 3.0880794701986756,
      "grad_norm": 0.9706143624576691,
      "learning_rate": 0.00020644532085421132,
      "loss": 1.7891,
      "step": 4663
    },
    {
      "epoch": 3.0887417218543045,
      "grad_norm": 0.8872837828024954,
      "learning_rate": 0.00020638347370585265,
      "loss": 1.7344,
      "step": 4664
    },
    {
      "epoch": 3.089403973509934,
      "grad_norm": 0.9966242375420337,
      "learning_rate": 0.00020632161539291982,
      "loss": 1.8281,
      "step": 4665
    },
    {
      "epoch": 3.090066225165563,
      "grad_norm": 0.9232214697770519,
      "learning_rate": 0.0002062597459276614,
      "loss": 1.7266,
      "step": 4666
    },
    {
      "epoch": 3.0907284768211922,
      "grad_norm": 0.8506549686225917,
      "learning_rate": 0.00020619786532232828,
      "loss": 1.5547,
      "step": 4667
    },
    {
      "epoch": 3.091390728476821,
      "grad_norm": 0.8510794844870859,
      "learning_rate": 0.00020613597358917359,
      "loss": 1.4688,
      "step": 4668
    },
    {
      "epoch": 3.0920529801324506,
      "grad_norm": 0.8539137687031026,
      "learning_rate": 0.0002060740707404525,
      "loss": 1.7578,
      "step": 4669
    },
    {
      "epoch": 3.0927152317880795,
      "grad_norm": 0.7947955576786425,
      "learning_rate": 0.0002060121567884225,
      "loss": 1.4688,
      "step": 4670
    },
    {
      "epoch": 3.0933774834437084,
      "grad_norm": 0.7698068357013594,
      "learning_rate": 0.0002059502317453434,
      "loss": 1.3281,
      "step": 4671
    },
    {
      "epoch": 3.094039735099338,
      "grad_norm": 0.9037490549810397,
      "learning_rate": 0.00020588829562347683,
      "loss": 1.6797,
      "step": 4672
    },
    {
      "epoch": 3.0947019867549668,
      "grad_norm": 0.9421652208257584,
      "learning_rate": 0.00020582634843508712,
      "loss": 1.7188,
      "step": 4673
    },
    {
      "epoch": 3.095364238410596,
      "grad_norm": 0.9059003261302294,
      "learning_rate": 0.00020576439019244033,
      "loss": 1.6797,
      "step": 4674
    },
    {
      "epoch": 3.096026490066225,
      "grad_norm": 0.9401419305067541,
      "learning_rate": 0.00020570242090780496,
      "loss": 1.6797,
      "step": 4675
    },
    {
      "epoch": 3.0966887417218545,
      "grad_norm": 0.8776207593802571,
      "learning_rate": 0.00020564044059345163,
      "loss": 1.5625,
      "step": 4676
    },
    {
      "epoch": 3.0973509933774834,
      "grad_norm": 0.8427510199871955,
      "learning_rate": 0.00020557844926165323,
      "loss": 1.4375,
      "step": 4677
    },
    {
      "epoch": 3.0980132450331124,
      "grad_norm": 0.9095356597994694,
      "learning_rate": 0.00020551644692468464,
      "loss": 1.6328,
      "step": 4678
    },
    {
      "epoch": 3.0986754966887418,
      "grad_norm": 1.0074786021084423,
      "learning_rate": 0.00020545443359482307,
      "loss": 1.8594,
      "step": 4679
    },
    {
      "epoch": 3.0993377483443707,
      "grad_norm": 0.7677534564652018,
      "learning_rate": 0.0002053924092843479,
      "loss": 1.3047,
      "step": 4680
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.8658906503570044,
      "learning_rate": 0.0002053303740055407,
      "loss": 1.5156,
      "step": 4681
    },
    {
      "epoch": 3.100662251655629,
      "grad_norm": 0.9739836221842648,
      "learning_rate": 0.000205268327770685,
      "loss": 1.8984,
      "step": 4682
    },
    {
      "epoch": 3.1013245033112584,
      "grad_norm": 0.9467927576376061,
      "learning_rate": 0.00020520627059206686,
      "loss": 1.7656,
      "step": 4683
    },
    {
      "epoch": 3.1019867549668874,
      "grad_norm": 0.8780570662627344,
      "learning_rate": 0.00020514420248197415,
      "loss": 1.6562,
      "step": 4684
    },
    {
      "epoch": 3.1026490066225167,
      "grad_norm": 0.9108640336754732,
      "learning_rate": 0.0002050821234526972,
      "loss": 1.6484,
      "step": 4685
    },
    {
      "epoch": 3.1033112582781457,
      "grad_norm": 0.9921169495990961,
      "learning_rate": 0.00020502003351652827,
      "loss": 1.8125,
      "step": 4686
    },
    {
      "epoch": 3.1039735099337746,
      "grad_norm": 0.9812881323981125,
      "learning_rate": 0.00020495793268576193,
      "loss": 1.7891,
      "step": 4687
    },
    {
      "epoch": 3.104635761589404,
      "grad_norm": 0.9528464449426903,
      "learning_rate": 0.00020489582097269485,
      "loss": 1.8047,
      "step": 4688
    },
    {
      "epoch": 3.105298013245033,
      "grad_norm": 0.8829663942487612,
      "learning_rate": 0.00020483369838962588,
      "loss": 1.7109,
      "step": 4689
    },
    {
      "epoch": 3.1059602649006623,
      "grad_norm": 0.9429364684072947,
      "learning_rate": 0.00020477156494885592,
      "loss": 1.9062,
      "step": 4690
    },
    {
      "epoch": 3.1066225165562913,
      "grad_norm": 0.743973398140737,
      "learning_rate": 0.00020470942066268827,
      "loss": 1.4766,
      "step": 4691
    },
    {
      "epoch": 3.1072847682119207,
      "grad_norm": 0.8439504905370184,
      "learning_rate": 0.00020464726554342798,
      "loss": 1.6484,
      "step": 4692
    },
    {
      "epoch": 3.1079470198675496,
      "grad_norm": 0.8708859799859928,
      "learning_rate": 0.00020458509960338266,
      "loss": 1.6953,
      "step": 4693
    },
    {
      "epoch": 3.108609271523179,
      "grad_norm": 0.8758578885670828,
      "learning_rate": 0.00020452292285486184,
      "loss": 1.8047,
      "step": 4694
    },
    {
      "epoch": 3.109271523178808,
      "grad_norm": 0.8138431188831488,
      "learning_rate": 0.0002044607353101772,
      "loss": 1.5312,
      "step": 4695
    },
    {
      "epoch": 3.109933774834437,
      "grad_norm": 0.7454445131733027,
      "learning_rate": 0.00020439853698164259,
      "loss": 1.4062,
      "step": 4696
    },
    {
      "epoch": 3.1105960264900663,
      "grad_norm": 0.8581127802867281,
      "learning_rate": 0.00020433632788157403,
      "loss": 1.6719,
      "step": 4697
    },
    {
      "epoch": 3.111258278145695,
      "grad_norm": 0.995488758376394,
      "learning_rate": 0.00020427410802228958,
      "loss": 2.0469,
      "step": 4698
    },
    {
      "epoch": 3.1119205298013246,
      "grad_norm": 0.9041309430174965,
      "learning_rate": 0.00020421187741610945,
      "loss": 1.5234,
      "step": 4699
    },
    {
      "epoch": 3.1125827814569536,
      "grad_norm": 0.8969536875233088,
      "learning_rate": 0.00020414963607535616,
      "loss": 1.6328,
      "step": 4700
    },
    {
      "epoch": 3.113245033112583,
      "grad_norm": 0.9864085304226945,
      "learning_rate": 0.00020408738401235405,
      "loss": 1.9453,
      "step": 4701
    },
    {
      "epoch": 3.113907284768212,
      "grad_norm": 1.00770183925664,
      "learning_rate": 0.00020402512123942989,
      "loss": 1.9922,
      "step": 4702
    },
    {
      "epoch": 3.1145695364238413,
      "grad_norm": 0.9464208092604262,
      "learning_rate": 0.00020396284776891228,
      "loss": 1.7812,
      "step": 4703
    },
    {
      "epoch": 3.11523178807947,
      "grad_norm": 0.9373626234723549,
      "learning_rate": 0.0002039005636131322,
      "loss": 1.6641,
      "step": 4704
    },
    {
      "epoch": 3.115894039735099,
      "grad_norm": 0.8730635001507833,
      "learning_rate": 0.00020383826878442253,
      "loss": 1.5078,
      "step": 4705
    },
    {
      "epoch": 3.1165562913907285,
      "grad_norm": 0.9852556376590144,
      "learning_rate": 0.00020377596329511843,
      "loss": 1.8125,
      "step": 4706
    },
    {
      "epoch": 3.1172185430463575,
      "grad_norm": 0.9307282502607125,
      "learning_rate": 0.00020371364715755707,
      "loss": 1.9297,
      "step": 4707
    },
    {
      "epoch": 3.117880794701987,
      "grad_norm": 0.9247479426248992,
      "learning_rate": 0.00020365132038407775,
      "loss": 1.9141,
      "step": 4708
    },
    {
      "epoch": 3.118543046357616,
      "grad_norm": 0.9707305264364012,
      "learning_rate": 0.00020358898298702193,
      "loss": 1.8359,
      "step": 4709
    },
    {
      "epoch": 3.119205298013245,
      "grad_norm": 0.8132560641434111,
      "learning_rate": 0.0002035266349787331,
      "loss": 1.625,
      "step": 4710
    },
    {
      "epoch": 3.119867549668874,
      "grad_norm": 0.983083712009347,
      "learning_rate": 0.0002034642763715568,
      "loss": 1.8828,
      "step": 4711
    },
    {
      "epoch": 3.120529801324503,
      "grad_norm": 0.8893176483203356,
      "learning_rate": 0.00020340190717784099,
      "loss": 1.6172,
      "step": 4712
    },
    {
      "epoch": 3.1211920529801325,
      "grad_norm": 0.9781223684790548,
      "learning_rate": 0.0002033395274099352,
      "loss": 1.9219,
      "step": 4713
    },
    {
      "epoch": 3.1218543046357614,
      "grad_norm": 0.8302557960962095,
      "learning_rate": 0.0002032771370801915,
      "loss": 1.5391,
      "step": 4714
    },
    {
      "epoch": 3.122516556291391,
      "grad_norm": 0.939152866317544,
      "learning_rate": 0.00020321473620096386,
      "loss": 1.9062,
      "step": 4715
    },
    {
      "epoch": 3.1231788079470197,
      "grad_norm": 0.8070751496203296,
      "learning_rate": 0.00020315232478460837,
      "loss": 1.4688,
      "step": 4716
    },
    {
      "epoch": 3.123841059602649,
      "grad_norm": 0.8605305543412525,
      "learning_rate": 0.00020308990284348315,
      "loss": 1.6719,
      "step": 4717
    },
    {
      "epoch": 3.124503311258278,
      "grad_norm": 1.488738921551124,
      "learning_rate": 0.0002030274703899486,
      "loss": 1.5703,
      "step": 4718
    },
    {
      "epoch": 3.1251655629139075,
      "grad_norm": 0.9093921971000104,
      "learning_rate": 0.00020296502743636692,
      "loss": 1.7109,
      "step": 4719
    },
    {
      "epoch": 3.1258278145695364,
      "grad_norm": 0.8938371419431302,
      "learning_rate": 0.00020290257399510262,
      "loss": 1.7344,
      "step": 4720
    },
    {
      "epoch": 3.1264900662251653,
      "grad_norm": 0.8135085818442961,
      "learning_rate": 0.00020284011007852212,
      "loss": 1.3984,
      "step": 4721
    },
    {
      "epoch": 3.1271523178807947,
      "grad_norm": 0.9250476231939163,
      "learning_rate": 0.00020277763569899403,
      "loss": 1.7891,
      "step": 4722
    },
    {
      "epoch": 3.1278145695364237,
      "grad_norm": 1.0292301280420675,
      "learning_rate": 0.000202715150868889,
      "loss": 1.7812,
      "step": 4723
    },
    {
      "epoch": 3.128476821192053,
      "grad_norm": 0.8435942079855326,
      "learning_rate": 0.0002026526556005798,
      "loss": 1.4766,
      "step": 4724
    },
    {
      "epoch": 3.129139072847682,
      "grad_norm": 0.8479999093634796,
      "learning_rate": 0.00020259014990644107,
      "loss": 1.4531,
      "step": 4725
    },
    {
      "epoch": 3.1298013245033114,
      "grad_norm": 0.9081233401945791,
      "learning_rate": 0.00020252763379884978,
      "loss": 1.5391,
      "step": 4726
    },
    {
      "epoch": 3.1304635761589403,
      "grad_norm": 0.8418863332855729,
      "learning_rate": 0.00020246510729018477,
      "loss": 1.4297,
      "step": 4727
    },
    {
      "epoch": 3.1311258278145697,
      "grad_norm": 0.9531005290172077,
      "learning_rate": 0.00020240257039282698,
      "loss": 1.6562,
      "step": 4728
    },
    {
      "epoch": 3.1317880794701987,
      "grad_norm": 0.9498994217447143,
      "learning_rate": 0.00020234002311915955,
      "loss": 1.7031,
      "step": 4729
    },
    {
      "epoch": 3.1324503311258276,
      "grad_norm": 0.9122095478336788,
      "learning_rate": 0.00020227746548156744,
      "loss": 1.6016,
      "step": 4730
    },
    {
      "epoch": 3.133112582781457,
      "grad_norm": 0.8667728389547882,
      "learning_rate": 0.00020221489749243783,
      "loss": 1.6562,
      "step": 4731
    },
    {
      "epoch": 3.133774834437086,
      "grad_norm": 0.8730820325350501,
      "learning_rate": 0.00020215231916415988,
      "loss": 1.5625,
      "step": 4732
    },
    {
      "epoch": 3.1344370860927153,
      "grad_norm": 0.8724884287334662,
      "learning_rate": 0.00020208973050912487,
      "loss": 1.5391,
      "step": 4733
    },
    {
      "epoch": 3.1350993377483443,
      "grad_norm": 0.8976583527234036,
      "learning_rate": 0.00020202713153972597,
      "loss": 1.5312,
      "step": 4734
    },
    {
      "epoch": 3.1357615894039736,
      "grad_norm": 1.0609405924304658,
      "learning_rate": 0.00020196452226835864,
      "loss": 2.1406,
      "step": 4735
    },
    {
      "epoch": 3.1364238410596026,
      "grad_norm": 0.9866013172667113,
      "learning_rate": 0.00020190190270742006,
      "loss": 1.9062,
      "step": 4736
    },
    {
      "epoch": 3.137086092715232,
      "grad_norm": 1.0038374344526333,
      "learning_rate": 0.00020183927286930974,
      "loss": 1.8672,
      "step": 4737
    },
    {
      "epoch": 3.137748344370861,
      "grad_norm": 0.926491647796575,
      "learning_rate": 0.00020177663276642907,
      "loss": 1.6641,
      "step": 4738
    },
    {
      "epoch": 3.13841059602649,
      "grad_norm": 0.9604939795846617,
      "learning_rate": 0.00020171398241118155,
      "loss": 1.7812,
      "step": 4739
    },
    {
      "epoch": 3.1390728476821192,
      "grad_norm": 0.7811555641962808,
      "learning_rate": 0.00020165132181597256,
      "loss": 1.4062,
      "step": 4740
    },
    {
      "epoch": 3.139735099337748,
      "grad_norm": 0.8726268745451699,
      "learning_rate": 0.00020158865099320975,
      "loss": 1.625,
      "step": 4741
    },
    {
      "epoch": 3.1403973509933776,
      "grad_norm": 0.9145235732264867,
      "learning_rate": 0.00020152596995530255,
      "loss": 1.7031,
      "step": 4742
    },
    {
      "epoch": 3.1410596026490065,
      "grad_norm": 0.949005804819896,
      "learning_rate": 0.0002014632787146626,
      "loss": 1.9062,
      "step": 4743
    },
    {
      "epoch": 3.141721854304636,
      "grad_norm": 0.9539903473673941,
      "learning_rate": 0.0002014005772837035,
      "loss": 1.8516,
      "step": 4744
    },
    {
      "epoch": 3.142384105960265,
      "grad_norm": 0.78762212482454,
      "learning_rate": 0.00020133786567484076,
      "loss": 1.3125,
      "step": 4745
    },
    {
      "epoch": 3.1430463576158942,
      "grad_norm": 0.8680657528549294,
      "learning_rate": 0.00020127514390049203,
      "loss": 1.5781,
      "step": 4746
    },
    {
      "epoch": 3.143708609271523,
      "grad_norm": 0.9599630427991712,
      "learning_rate": 0.000201212411973077,
      "loss": 1.7188,
      "step": 4747
    },
    {
      "epoch": 3.144370860927152,
      "grad_norm": 1.0601707812336376,
      "learning_rate": 0.00020114966990501722,
      "loss": 1.8125,
      "step": 4748
    },
    {
      "epoch": 3.1450331125827815,
      "grad_norm": 0.9135483200718882,
      "learning_rate": 0.00020108691770873646,
      "loss": 1.5312,
      "step": 4749
    },
    {
      "epoch": 3.1456953642384105,
      "grad_norm": 0.9371944369931705,
      "learning_rate": 0.00020102415539666023,
      "loss": 1.7812,
      "step": 4750
    },
    {
      "epoch": 3.14635761589404,
      "grad_norm": 0.9380896744405436,
      "learning_rate": 0.00020096138298121628,
      "loss": 1.6719,
      "step": 4751
    },
    {
      "epoch": 3.147019867549669,
      "grad_norm": 1.0001339059853116,
      "learning_rate": 0.00020089860047483427,
      "loss": 1.7422,
      "step": 4752
    },
    {
      "epoch": 3.147682119205298,
      "grad_norm": 0.943004838918222,
      "learning_rate": 0.00020083580788994582,
      "loss": 1.5859,
      "step": 4753
    },
    {
      "epoch": 3.148344370860927,
      "grad_norm": 0.9761495831594614,
      "learning_rate": 0.00020077300523898458,
      "loss": 1.7422,
      "step": 4754
    },
    {
      "epoch": 3.1490066225165565,
      "grad_norm": 0.9450951009824943,
      "learning_rate": 0.00020071019253438621,
      "loss": 1.5938,
      "step": 4755
    },
    {
      "epoch": 3.1496688741721854,
      "grad_norm": 0.9445845698512724,
      "learning_rate": 0.00020064736978858832,
      "loss": 1.7109,
      "step": 4756
    },
    {
      "epoch": 3.1503311258278144,
      "grad_norm": 0.8032985757460158,
      "learning_rate": 0.00020058453701403062,
      "loss": 1.3906,
      "step": 4757
    },
    {
      "epoch": 3.1509933774834438,
      "grad_norm": 0.9568367853722892,
      "learning_rate": 0.00020052169422315466,
      "loss": 1.5312,
      "step": 4758
    },
    {
      "epoch": 3.1516556291390727,
      "grad_norm": 0.9080326510309713,
      "learning_rate": 0.00020045884142840396,
      "loss": 1.5469,
      "step": 4759
    },
    {
      "epoch": 3.152317880794702,
      "grad_norm": 0.9668411272037408,
      "learning_rate": 0.00020039597864222422,
      "loss": 1.6953,
      "step": 4760
    },
    {
      "epoch": 3.152980132450331,
      "grad_norm": 0.951547293110932,
      "learning_rate": 0.00020033310587706294,
      "loss": 1.5938,
      "step": 4761
    },
    {
      "epoch": 3.1536423841059604,
      "grad_norm": 0.8661842339621213,
      "learning_rate": 0.00020027022314536962,
      "loss": 1.4453,
      "step": 4762
    },
    {
      "epoch": 3.1543046357615894,
      "grad_norm": 0.9888373280493069,
      "learning_rate": 0.0002002073304595958,
      "loss": 1.6484,
      "step": 4763
    },
    {
      "epoch": 3.1549668874172188,
      "grad_norm": 0.9665658068523724,
      "learning_rate": 0.00020014442783219496,
      "loss": 1.6484,
      "step": 4764
    },
    {
      "epoch": 3.1556291390728477,
      "grad_norm": 0.9573797818516224,
      "learning_rate": 0.00020008151527562247,
      "loss": 1.6719,
      "step": 4765
    },
    {
      "epoch": 3.1562913907284766,
      "grad_norm": 0.8399673916289535,
      "learning_rate": 0.00020001859280233583,
      "loss": 1.3047,
      "step": 4766
    },
    {
      "epoch": 3.156953642384106,
      "grad_norm": 0.9025927986643074,
      "learning_rate": 0.00019995566042479437,
      "loss": 1.5859,
      "step": 4767
    },
    {
      "epoch": 3.157615894039735,
      "grad_norm": 0.9912661131344919,
      "learning_rate": 0.00019989271815545943,
      "loss": 1.7656,
      "step": 4768
    },
    {
      "epoch": 3.1582781456953644,
      "grad_norm": 0.875713571107221,
      "learning_rate": 0.00019982976600679424,
      "loss": 1.4688,
      "step": 4769
    },
    {
      "epoch": 3.1589403973509933,
      "grad_norm": 0.8986893238647694,
      "learning_rate": 0.00019976680399126417,
      "loss": 1.5703,
      "step": 4770
    },
    {
      "epoch": 3.1596026490066227,
      "grad_norm": 0.9639289276921501,
      "learning_rate": 0.0001997038321213363,
      "loss": 1.6641,
      "step": 4771
    },
    {
      "epoch": 3.1602649006622516,
      "grad_norm": 0.8914560321893221,
      "learning_rate": 0.00019964085040947988,
      "loss": 1.6094,
      "step": 4772
    },
    {
      "epoch": 3.160927152317881,
      "grad_norm": 0.9993083039168085,
      "learning_rate": 0.00019957785886816592,
      "loss": 1.7656,
      "step": 4773
    },
    {
      "epoch": 3.16158940397351,
      "grad_norm": 0.8849672561838005,
      "learning_rate": 0.00019951485750986753,
      "loss": 1.5938,
      "step": 4774
    },
    {
      "epoch": 3.162251655629139,
      "grad_norm": 1.2790459220560648,
      "learning_rate": 0.00019945184634705967,
      "loss": 1.8359,
      "step": 4775
    },
    {
      "epoch": 3.1629139072847683,
      "grad_norm": 0.8416389287272015,
      "learning_rate": 0.00019938882539221936,
      "loss": 1.4062,
      "step": 4776
    },
    {
      "epoch": 3.1635761589403972,
      "grad_norm": 0.9469331724518958,
      "learning_rate": 0.00019932579465782528,
      "loss": 1.7344,
      "step": 4777
    },
    {
      "epoch": 3.1642384105960266,
      "grad_norm": 0.9152258051597071,
      "learning_rate": 0.00019926275415635847,
      "loss": 1.6094,
      "step": 4778
    },
    {
      "epoch": 3.1649006622516556,
      "grad_norm": 0.9925922054764895,
      "learning_rate": 0.00019919970390030146,
      "loss": 1.875,
      "step": 4779
    },
    {
      "epoch": 3.165562913907285,
      "grad_norm": 0.8513714862427627,
      "learning_rate": 0.00019913664390213909,
      "loss": 1.6875,
      "step": 4780
    },
    {
      "epoch": 3.166225165562914,
      "grad_norm": 0.8285205626472368,
      "learning_rate": 0.00019907357417435788,
      "loss": 1.4453,
      "step": 4781
    },
    {
      "epoch": 3.1668874172185433,
      "grad_norm": 0.9512535072851382,
      "learning_rate": 0.00019901049472944635,
      "loss": 1.75,
      "step": 4782
    },
    {
      "epoch": 3.167549668874172,
      "grad_norm": 1.0026635643529767,
      "learning_rate": 0.00019894740557989496,
      "loss": 1.8828,
      "step": 4783
    },
    {
      "epoch": 3.168211920529801,
      "grad_norm": 0.9973596357614611,
      "learning_rate": 0.0001988843067381962,
      "loss": 1.8672,
      "step": 4784
    },
    {
      "epoch": 3.1688741721854305,
      "grad_norm": 1.017454552496059,
      "learning_rate": 0.00019882119821684416,
      "loss": 1.7578,
      "step": 4785
    },
    {
      "epoch": 3.1695364238410595,
      "grad_norm": 0.951756919833561,
      "learning_rate": 0.00019875808002833522,
      "loss": 1.8594,
      "step": 4786
    },
    {
      "epoch": 3.170198675496689,
      "grad_norm": 0.8446962452081238,
      "learning_rate": 0.0001986949521851674,
      "loss": 1.625,
      "step": 4787
    },
    {
      "epoch": 3.170860927152318,
      "grad_norm": 0.9277982495163843,
      "learning_rate": 0.0001986318146998408,
      "loss": 1.8125,
      "step": 4788
    },
    {
      "epoch": 3.171523178807947,
      "grad_norm": 0.8757118669745091,
      "learning_rate": 0.00019856866758485734,
      "loss": 1.5391,
      "step": 4789
    },
    {
      "epoch": 3.172185430463576,
      "grad_norm": 0.9288180630813149,
      "learning_rate": 0.00019850551085272085,
      "loss": 1.7109,
      "step": 4790
    },
    {
      "epoch": 3.172847682119205,
      "grad_norm": 0.9044629273475588,
      "learning_rate": 0.00019844234451593716,
      "loss": 1.5859,
      "step": 4791
    },
    {
      "epoch": 3.1735099337748345,
      "grad_norm": 0.9763842476388541,
      "learning_rate": 0.00019837916858701386,
      "loss": 1.6406,
      "step": 4792
    },
    {
      "epoch": 3.1741721854304634,
      "grad_norm": 0.9899297525287235,
      "learning_rate": 0.00019831598307846056,
      "loss": 1.7969,
      "step": 4793
    },
    {
      "epoch": 3.174834437086093,
      "grad_norm": 0.9008710636987658,
      "learning_rate": 0.00019825278800278864,
      "loss": 1.4297,
      "step": 4794
    },
    {
      "epoch": 3.1754966887417218,
      "grad_norm": 0.9612942292523843,
      "learning_rate": 0.0001981895833725115,
      "loss": 1.6562,
      "step": 4795
    },
    {
      "epoch": 3.176158940397351,
      "grad_norm": 0.9149025571555022,
      "learning_rate": 0.00019812636920014438,
      "loss": 1.5547,
      "step": 4796
    },
    {
      "epoch": 3.17682119205298,
      "grad_norm": 0.9577596693959226,
      "learning_rate": 0.00019806314549820442,
      "loss": 1.6172,
      "step": 4797
    },
    {
      "epoch": 3.1774834437086095,
      "grad_norm": 0.9773002412922633,
      "learning_rate": 0.0001979999122792106,
      "loss": 1.6797,
      "step": 4798
    },
    {
      "epoch": 3.1781456953642384,
      "grad_norm": 0.9553474025727391,
      "learning_rate": 0.0001979366695556839,
      "loss": 1.5781,
      "step": 4799
    },
    {
      "epoch": 3.1788079470198674,
      "grad_norm": 0.9840394769545622,
      "learning_rate": 0.000197873417340147,
      "loss": 1.6641,
      "step": 4800
    },
    {
      "epoch": 3.1794701986754967,
      "grad_norm": 0.8771929524206465,
      "learning_rate": 0.00019781015564512464,
      "loss": 1.4922,
      "step": 4801
    },
    {
      "epoch": 3.1801324503311257,
      "grad_norm": 0.8007004073227335,
      "learning_rate": 0.00019774688448314336,
      "loss": 1.3984,
      "step": 4802
    },
    {
      "epoch": 3.180794701986755,
      "grad_norm": 0.9827267918991061,
      "learning_rate": 0.0001976836038667315,
      "loss": 1.9297,
      "step": 4803
    },
    {
      "epoch": 3.181456953642384,
      "grad_norm": 1.198526836331655,
      "learning_rate": 0.00019762031380841943,
      "loss": 1.6797,
      "step": 4804
    },
    {
      "epoch": 3.1821192052980134,
      "grad_norm": 0.9509696245265932,
      "learning_rate": 0.0001975570143207393,
      "loss": 1.7266,
      "step": 4805
    },
    {
      "epoch": 3.1827814569536423,
      "grad_norm": 0.9104250293045867,
      "learning_rate": 0.00019749370541622503,
      "loss": 1.7031,
      "step": 4806
    },
    {
      "epoch": 3.1834437086092717,
      "grad_norm": 0.8189778202197172,
      "learning_rate": 0.00019743038710741268,
      "loss": 1.4844,
      "step": 4807
    },
    {
      "epoch": 3.1841059602649007,
      "grad_norm": 0.8190698815898578,
      "learning_rate": 0.00019736705940683985,
      "loss": 1.5469,
      "step": 4808
    },
    {
      "epoch": 3.1847682119205296,
      "grad_norm": 0.9366395358577256,
      "learning_rate": 0.00019730372232704626,
      "loss": 1.6875,
      "step": 4809
    },
    {
      "epoch": 3.185430463576159,
      "grad_norm": 0.9400742396935188,
      "learning_rate": 0.0001972403758805733,
      "loss": 1.7188,
      "step": 4810
    },
    {
      "epoch": 3.186092715231788,
      "grad_norm": 0.9493753619356124,
      "learning_rate": 0.00019717702007996432,
      "loss": 1.625,
      "step": 4811
    },
    {
      "epoch": 3.1867549668874173,
      "grad_norm": 0.9398627229211451,
      "learning_rate": 0.0001971136549377645,
      "loss": 1.5781,
      "step": 4812
    },
    {
      "epoch": 3.1874172185430463,
      "grad_norm": 0.8928465578514219,
      "learning_rate": 0.0001970502804665209,
      "loss": 1.4922,
      "step": 4813
    },
    {
      "epoch": 3.1880794701986757,
      "grad_norm": 0.9960723709747267,
      "learning_rate": 0.0001969868966787823,
      "loss": 1.6641,
      "step": 4814
    },
    {
      "epoch": 3.1887417218543046,
      "grad_norm": 1.0131731735292189,
      "learning_rate": 0.00019692350358709948,
      "loss": 1.7969,
      "step": 4815
    },
    {
      "epoch": 3.1894039735099335,
      "grad_norm": 0.9652077287315285,
      "learning_rate": 0.00019686010120402501,
      "loss": 1.5,
      "step": 4816
    },
    {
      "epoch": 3.190066225165563,
      "grad_norm": 1.0335359633173542,
      "learning_rate": 0.00019679668954211327,
      "loss": 1.8047,
      "step": 4817
    },
    {
      "epoch": 3.190728476821192,
      "grad_norm": 1.0330314983797757,
      "learning_rate": 0.00019673326861392048,
      "loss": 1.8438,
      "step": 4818
    },
    {
      "epoch": 3.1913907284768213,
      "grad_norm": 1.0485981242686409,
      "learning_rate": 0.00019666983843200472,
      "loss": 2.0156,
      "step": 4819
    },
    {
      "epoch": 3.19205298013245,
      "grad_norm": 1.017382082816668,
      "learning_rate": 0.0001966063990089259,
      "loss": 1.7812,
      "step": 4820
    },
    {
      "epoch": 3.1927152317880796,
      "grad_norm": 0.7998309235828397,
      "learning_rate": 0.00019654295035724577,
      "loss": 1.4844,
      "step": 4821
    },
    {
      "epoch": 3.1933774834437085,
      "grad_norm": 0.7705291381186155,
      "learning_rate": 0.0001964794924895279,
      "loss": 1.3281,
      "step": 4822
    },
    {
      "epoch": 3.194039735099338,
      "grad_norm": 0.9038680546967081,
      "learning_rate": 0.00019641602541833759,
      "loss": 1.4688,
      "step": 4823
    },
    {
      "epoch": 3.194701986754967,
      "grad_norm": 0.922974383023784,
      "learning_rate": 0.0001963525491562421,
      "loss": 1.7969,
      "step": 4824
    },
    {
      "epoch": 3.195364238410596,
      "grad_norm": 0.9750477134622206,
      "learning_rate": 0.0001962890637158105,
      "loss": 1.6953,
      "step": 4825
    },
    {
      "epoch": 3.196026490066225,
      "grad_norm": 0.9725357180803608,
      "learning_rate": 0.0001962255691096136,
      "loss": 1.6953,
      "step": 4826
    },
    {
      "epoch": 3.196688741721854,
      "grad_norm": 0.9464508723550148,
      "learning_rate": 0.000196162065350224,
      "loss": 1.7422,
      "step": 4827
    },
    {
      "epoch": 3.1973509933774835,
      "grad_norm": 0.9764154860961823,
      "learning_rate": 0.0001960985524502163,
      "loss": 1.875,
      "step": 4828
    },
    {
      "epoch": 3.1980132450331125,
      "grad_norm": 0.9582627945669965,
      "learning_rate": 0.00019603503042216666,
      "loss": 1.6797,
      "step": 4829
    },
    {
      "epoch": 3.198675496688742,
      "grad_norm": 0.9346992952607505,
      "learning_rate": 0.00019597149927865325,
      "loss": 1.5859,
      "step": 4830
    },
    {
      "epoch": 3.199337748344371,
      "grad_norm": 0.9215032683868045,
      "learning_rate": 0.00019590795903225592,
      "loss": 1.6406,
      "step": 4831
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.8530564093005187,
      "learning_rate": 0.0001958444096955564,
      "loss": 1.4375,
      "step": 4832
    },
    {
      "epoch": 3.200662251655629,
      "grad_norm": 0.9649602027120932,
      "learning_rate": 0.0001957808512811382,
      "loss": 1.7656,
      "step": 4833
    },
    {
      "epoch": 3.201324503311258,
      "grad_norm": 0.8746722212516136,
      "learning_rate": 0.00019571728380158652,
      "loss": 1.4844,
      "step": 4834
    },
    {
      "epoch": 3.2019867549668874,
      "grad_norm": 1.1491697946409087,
      "learning_rate": 0.0001956537072694885,
      "loss": 1.7109,
      "step": 4835
    },
    {
      "epoch": 3.2026490066225164,
      "grad_norm": 0.9876531246334254,
      "learning_rate": 0.00019559012169743315,
      "loss": 1.5703,
      "step": 4836
    },
    {
      "epoch": 3.203311258278146,
      "grad_norm": 0.9253618621675296,
      "learning_rate": 0.00019552652709801092,
      "loss": 1.5,
      "step": 4837
    },
    {
      "epoch": 3.2039735099337747,
      "grad_norm": 0.9842876306317241,
      "learning_rate": 0.00019546292348381444,
      "loss": 1.8359,
      "step": 4838
    },
    {
      "epoch": 3.204635761589404,
      "grad_norm": 0.9477480348255984,
      "learning_rate": 0.0001953993108674379,
      "loss": 1.6562,
      "step": 4839
    },
    {
      "epoch": 3.205298013245033,
      "grad_norm": 0.9077160498000002,
      "learning_rate": 0.00019533568926147728,
      "loss": 1.5781,
      "step": 4840
    },
    {
      "epoch": 3.2059602649006624,
      "grad_norm": 0.9533461266022754,
      "learning_rate": 0.00019527205867853043,
      "loss": 1.7109,
      "step": 4841
    },
    {
      "epoch": 3.2066225165562914,
      "grad_norm": 7.053061669844967,
      "learning_rate": 0.000195208419131197,
      "loss": 1.5078,
      "step": 4842
    },
    {
      "epoch": 3.2072847682119203,
      "grad_norm": 0.9584480507770043,
      "learning_rate": 0.0001951447706320782,
      "loss": 1.7656,
      "step": 4843
    },
    {
      "epoch": 3.2079470198675497,
      "grad_norm": 0.9185393859922065,
      "learning_rate": 0.0001950811131937773,
      "loss": 1.6484,
      "step": 4844
    },
    {
      "epoch": 3.2086092715231787,
      "grad_norm": 0.9244197866272448,
      "learning_rate": 0.00019501744682889915,
      "loss": 1.5391,
      "step": 4845
    },
    {
      "epoch": 3.209271523178808,
      "grad_norm": 0.865482180345394,
      "learning_rate": 0.0001949537715500504,
      "loss": 1.6172,
      "step": 4846
    },
    {
      "epoch": 3.209933774834437,
      "grad_norm": 0.7972364654115054,
      "learning_rate": 0.00019489008736983953,
      "loss": 1.4375,
      "step": 4847
    },
    {
      "epoch": 3.2105960264900664,
      "grad_norm": 0.9292399844848587,
      "learning_rate": 0.0001948263943008767,
      "loss": 1.6562,
      "step": 4848
    },
    {
      "epoch": 3.2112582781456953,
      "grad_norm": 0.9460124097628286,
      "learning_rate": 0.00019476269235577388,
      "loss": 1.6484,
      "step": 4849
    },
    {
      "epoch": 3.2119205298013247,
      "grad_norm": 1.0260904401138844,
      "learning_rate": 0.00019469898154714474,
      "loss": 1.8984,
      "step": 4850
    },
    {
      "epoch": 3.2125827814569536,
      "grad_norm": 0.9544954952010148,
      "learning_rate": 0.00019463526188760483,
      "loss": 1.7188,
      "step": 4851
    },
    {
      "epoch": 3.2132450331125826,
      "grad_norm": 0.9408366216465428,
      "learning_rate": 0.00019457153338977136,
      "loss": 1.6797,
      "step": 4852
    },
    {
      "epoch": 3.213907284768212,
      "grad_norm": 0.952872436657239,
      "learning_rate": 0.00019450779606626327,
      "loss": 1.5234,
      "step": 4853
    },
    {
      "epoch": 3.214569536423841,
      "grad_norm": 0.9296375596840306,
      "learning_rate": 0.0001944440499297013,
      "loss": 1.6797,
      "step": 4854
    },
    {
      "epoch": 3.2152317880794703,
      "grad_norm": 1.1188789260980818,
      "learning_rate": 0.0001943802949927079,
      "loss": 1.8672,
      "step": 4855
    },
    {
      "epoch": 3.2158940397350992,
      "grad_norm": 1.0000600783203344,
      "learning_rate": 0.00019431653126790726,
      "loss": 1.6172,
      "step": 4856
    },
    {
      "epoch": 3.2165562913907286,
      "grad_norm": 0.8707557676502273,
      "learning_rate": 0.00019425275876792545,
      "loss": 1.5156,
      "step": 4857
    },
    {
      "epoch": 3.2172185430463576,
      "grad_norm": 0.9132577220630183,
      "learning_rate": 0.00019418897750538998,
      "loss": 1.4766,
      "step": 4858
    },
    {
      "epoch": 3.217880794701987,
      "grad_norm": 0.9260817991994212,
      "learning_rate": 0.00019412518749293038,
      "loss": 1.6562,
      "step": 4859
    },
    {
      "epoch": 3.218543046357616,
      "grad_norm": 0.9708688863840168,
      "learning_rate": 0.00019406138874317774,
      "loss": 1.7891,
      "step": 4860
    },
    {
      "epoch": 3.219205298013245,
      "grad_norm": 1.041455062859568,
      "learning_rate": 0.00019399758126876503,
      "loss": 1.8359,
      "step": 4861
    },
    {
      "epoch": 3.2198675496688742,
      "grad_norm": 0.8751066218062601,
      "learning_rate": 0.00019393376508232679,
      "loss": 1.5078,
      "step": 4862
    },
    {
      "epoch": 3.220529801324503,
      "grad_norm": 0.8636007075942181,
      "learning_rate": 0.0001938699401964994,
      "loss": 1.4141,
      "step": 4863
    },
    {
      "epoch": 3.2211920529801326,
      "grad_norm": 1.027546781883396,
      "learning_rate": 0.00019380610662392085,
      "loss": 1.7578,
      "step": 4864
    },
    {
      "epoch": 3.2218543046357615,
      "grad_norm": 0.868202554367135,
      "learning_rate": 0.00019374226437723105,
      "loss": 1.5312,
      "step": 4865
    },
    {
      "epoch": 3.222516556291391,
      "grad_norm": 1.055183901110289,
      "learning_rate": 0.00019367841346907134,
      "loss": 1.8047,
      "step": 4866
    },
    {
      "epoch": 3.22317880794702,
      "grad_norm": 1.018924114009745,
      "learning_rate": 0.00019361455391208503,
      "loss": 1.7656,
      "step": 4867
    },
    {
      "epoch": 3.223841059602649,
      "grad_norm": 1.0403146046277116,
      "learning_rate": 0.00019355068571891702,
      "loss": 2.0156,
      "step": 4868
    },
    {
      "epoch": 3.224503311258278,
      "grad_norm": 0.9036726133713865,
      "learning_rate": 0.00019348680890221397,
      "loss": 1.5547,
      "step": 4869
    },
    {
      "epoch": 3.225165562913907,
      "grad_norm": 0.8921035589227718,
      "learning_rate": 0.00019342292347462414,
      "loss": 1.7031,
      "step": 4870
    },
    {
      "epoch": 3.2258278145695365,
      "grad_norm": 0.9171277730736541,
      "learning_rate": 0.00019335902944879772,
      "loss": 1.8438,
      "step": 4871
    },
    {
      "epoch": 3.2264900662251654,
      "grad_norm": 0.9393102022935743,
      "learning_rate": 0.0001932951268373863,
      "loss": 2.0469,
      "step": 4872
    },
    {
      "epoch": 3.227152317880795,
      "grad_norm": 0.8816303229639194,
      "learning_rate": 0.00019323121565304342,
      "loss": 1.75,
      "step": 4873
    },
    {
      "epoch": 3.2278145695364238,
      "grad_norm": 0.9452546115838456,
      "learning_rate": 0.00019316729590842423,
      "loss": 2.0,
      "step": 4874
    },
    {
      "epoch": 3.228476821192053,
      "grad_norm": 0.7631561615764948,
      "learning_rate": 0.00019310336761618554,
      "loss": 1.4609,
      "step": 4875
    },
    {
      "epoch": 3.229139072847682,
      "grad_norm": 0.816914132084238,
      "learning_rate": 0.0001930394307889859,
      "loss": 1.6094,
      "step": 4876
    },
    {
      "epoch": 3.2298013245033115,
      "grad_norm": 0.8185009443276723,
      "learning_rate": 0.00019297548543948555,
      "loss": 1.6953,
      "step": 4877
    },
    {
      "epoch": 3.2304635761589404,
      "grad_norm": 0.9336564258908694,
      "learning_rate": 0.00019291153158034635,
      "loss": 1.9766,
      "step": 4878
    },
    {
      "epoch": 3.2311258278145694,
      "grad_norm": 0.9364367756976485,
      "learning_rate": 0.00019284756922423196,
      "loss": 1.6875,
      "step": 4879
    },
    {
      "epoch": 3.2317880794701987,
      "grad_norm": 0.9525706359250368,
      "learning_rate": 0.00019278359838380762,
      "loss": 1.8516,
      "step": 4880
    },
    {
      "epoch": 3.2324503311258277,
      "grad_norm": 0.9759224580732908,
      "learning_rate": 0.00019271961907174032,
      "loss": 1.9219,
      "step": 4881
    },
    {
      "epoch": 3.233112582781457,
      "grad_norm": 0.9100532853953787,
      "learning_rate": 0.00019265563130069867,
      "loss": 1.7344,
      "step": 4882
    },
    {
      "epoch": 3.233774834437086,
      "grad_norm": 0.8145191419596924,
      "learning_rate": 0.00019259163508335297,
      "loss": 1.4453,
      "step": 4883
    },
    {
      "epoch": 3.2344370860927154,
      "grad_norm": 0.8990821578691979,
      "learning_rate": 0.00019252763043237526,
      "loss": 1.4922,
      "step": 4884
    },
    {
      "epoch": 3.2350993377483444,
      "grad_norm": 1.0125083867996316,
      "learning_rate": 0.0001924636173604391,
      "loss": 1.9688,
      "step": 4885
    },
    {
      "epoch": 3.2357615894039737,
      "grad_norm": 1.0307807269538847,
      "learning_rate": 0.00019239959588021998,
      "loss": 1.8828,
      "step": 4886
    },
    {
      "epoch": 3.2364238410596027,
      "grad_norm": 1.0119292700872482,
      "learning_rate": 0.00019233556600439466,
      "loss": 1.8281,
      "step": 4887
    },
    {
      "epoch": 3.2370860927152316,
      "grad_norm": 0.9640582265653731,
      "learning_rate": 0.000192271527745642,
      "loss": 1.75,
      "step": 4888
    },
    {
      "epoch": 3.237748344370861,
      "grad_norm": 0.9785363733124639,
      "learning_rate": 0.00019220748111664217,
      "loss": 1.7422,
      "step": 4889
    },
    {
      "epoch": 3.23841059602649,
      "grad_norm": 0.8596427652146855,
      "learning_rate": 0.00019214342613007722,
      "loss": 1.6172,
      "step": 4890
    },
    {
      "epoch": 3.2390728476821193,
      "grad_norm": 0.888870099348229,
      "learning_rate": 0.0001920793627986307,
      "loss": 1.7109,
      "step": 4891
    },
    {
      "epoch": 3.2397350993377483,
      "grad_norm": 0.9808046427816547,
      "learning_rate": 0.00019201529113498797,
      "loss": 1.6719,
      "step": 4892
    },
    {
      "epoch": 3.2403973509933777,
      "grad_norm": 0.8934102956514723,
      "learning_rate": 0.00019195121115183584,
      "loss": 1.7812,
      "step": 4893
    },
    {
      "epoch": 3.2410596026490066,
      "grad_norm": 0.9304499820236418,
      "learning_rate": 0.00019188712286186303,
      "loss": 1.8359,
      "step": 4894
    },
    {
      "epoch": 3.241721854304636,
      "grad_norm": 0.9409039542920326,
      "learning_rate": 0.0001918230262777596,
      "loss": 1.6719,
      "step": 4895
    },
    {
      "epoch": 3.242384105960265,
      "grad_norm": 0.9542723786867001,
      "learning_rate": 0.00019175892141221748,
      "loss": 1.7891,
      "step": 4896
    },
    {
      "epoch": 3.243046357615894,
      "grad_norm": 0.9535822380521166,
      "learning_rate": 0.00019169480827793022,
      "loss": 1.7266,
      "step": 4897
    },
    {
      "epoch": 3.2437086092715233,
      "grad_norm": 0.9435524082120601,
      "learning_rate": 0.0001916306868875929,
      "loss": 1.7188,
      "step": 4898
    },
    {
      "epoch": 3.244370860927152,
      "grad_norm": 0.9908033868567754,
      "learning_rate": 0.00019156655725390226,
      "loss": 2.0156,
      "step": 4899
    },
    {
      "epoch": 3.2450331125827816,
      "grad_norm": 1.0063589420231038,
      "learning_rate": 0.00019150241938955677,
      "loss": 1.8594,
      "step": 4900
    },
    {
      "epoch": 3.2456953642384105,
      "grad_norm": 0.8974342434456295,
      "learning_rate": 0.0001914382733072564,
      "loss": 1.6094,
      "step": 4901
    },
    {
      "epoch": 3.24635761589404,
      "grad_norm": 0.9373137441108849,
      "learning_rate": 0.00019137411901970288,
      "loss": 1.9297,
      "step": 4902
    },
    {
      "epoch": 3.247019867549669,
      "grad_norm": 0.9536229261645489,
      "learning_rate": 0.00019130995653959942,
      "loss": 1.7578,
      "step": 4903
    },
    {
      "epoch": 3.247682119205298,
      "grad_norm": 0.9185011624949272,
      "learning_rate": 0.000191245785879651,
      "loss": 1.7422,
      "step": 4904
    },
    {
      "epoch": 3.248344370860927,
      "grad_norm": 0.7804035594968938,
      "learning_rate": 0.00019118160705256407,
      "loss": 1.3281,
      "step": 4905
    },
    {
      "epoch": 3.249006622516556,
      "grad_norm": 0.8481157369974262,
      "learning_rate": 0.00019111742007104686,
      "loss": 1.4688,
      "step": 4906
    },
    {
      "epoch": 3.2496688741721855,
      "grad_norm": 0.885660031580941,
      "learning_rate": 0.000191053224947809,
      "loss": 1.4844,
      "step": 4907
    },
    {
      "epoch": 3.2503311258278145,
      "grad_norm": 1.0517835858787712,
      "learning_rate": 0.000190989021695562,
      "loss": 1.8672,
      "step": 4908
    },
    {
      "epoch": 3.250993377483444,
      "grad_norm": 0.8942583848239999,
      "learning_rate": 0.0001909248103270188,
      "loss": 1.6094,
      "step": 4909
    },
    {
      "epoch": 3.251655629139073,
      "grad_norm": 0.9875719823218626,
      "learning_rate": 0.00019086059085489393,
      "loss": 1.7266,
      "step": 4910
    },
    {
      "epoch": 3.2523178807947017,
      "grad_norm": 0.9240386474487559,
      "learning_rate": 0.0001907963632919036,
      "loss": 1.7266,
      "step": 4911
    },
    {
      "epoch": 3.252980132450331,
      "grad_norm": 0.8921565048279185,
      "learning_rate": 0.00019073212765076565,
      "loss": 1.6094,
      "step": 4912
    },
    {
      "epoch": 3.25364238410596,
      "grad_norm": 0.9130339713731725,
      "learning_rate": 0.0001906678839441995,
      "loss": 1.6406,
      "step": 4913
    },
    {
      "epoch": 3.2543046357615895,
      "grad_norm": 1.0209129976535385,
      "learning_rate": 0.00019060363218492598,
      "loss": 1.8906,
      "step": 4914
    },
    {
      "epoch": 3.2549668874172184,
      "grad_norm": 0.9560073960212097,
      "learning_rate": 0.0001905393723856679,
      "loss": 1.6641,
      "step": 4915
    },
    {
      "epoch": 3.255629139072848,
      "grad_norm": 1.0409209189966806,
      "learning_rate": 0.0001904751045591492,
      "loss": 1.875,
      "step": 4916
    },
    {
      "epoch": 3.2562913907284767,
      "grad_norm": 0.8815186191235089,
      "learning_rate": 0.00019041082871809582,
      "loss": 1.625,
      "step": 4917
    },
    {
      "epoch": 3.256953642384106,
      "grad_norm": 0.9658119113973413,
      "learning_rate": 0.00019034654487523508,
      "loss": 1.625,
      "step": 4918
    },
    {
      "epoch": 3.257615894039735,
      "grad_norm": 0.9434182707562291,
      "learning_rate": 0.00019028225304329588,
      "loss": 1.7812,
      "step": 4919
    },
    {
      "epoch": 3.258278145695364,
      "grad_norm": 0.9784490209456781,
      "learning_rate": 0.00019021795323500873,
      "loss": 1.7656,
      "step": 4920
    },
    {
      "epoch": 3.2589403973509934,
      "grad_norm": 0.7987677486872828,
      "learning_rate": 0.0001901536454631058,
      "loss": 1.2734,
      "step": 4921
    },
    {
      "epoch": 3.2596026490066223,
      "grad_norm": 1.2162966490012375,
      "learning_rate": 0.00019008932974032066,
      "loss": 1.5078,
      "step": 4922
    },
    {
      "epoch": 3.2602649006622517,
      "grad_norm": 1.0779355197509937,
      "learning_rate": 0.0001900250060793887,
      "loss": 1.8359,
      "step": 4923
    },
    {
      "epoch": 3.2609271523178807,
      "grad_norm": 1.0704686680880526,
      "learning_rate": 0.0001899606744930466,
      "loss": 1.8984,
      "step": 4924
    },
    {
      "epoch": 3.26158940397351,
      "grad_norm": 0.9026616374879265,
      "learning_rate": 0.00018989633499403284,
      "loss": 1.625,
      "step": 4925
    },
    {
      "epoch": 3.262251655629139,
      "grad_norm": 1.049832751285563,
      "learning_rate": 0.00018983198759508737,
      "loss": 1.75,
      "step": 4926
    },
    {
      "epoch": 3.2629139072847684,
      "grad_norm": 0.8690844459982967,
      "learning_rate": 0.00018976763230895167,
      "loss": 1.4219,
      "step": 4927
    },
    {
      "epoch": 3.2635761589403973,
      "grad_norm": 0.9560964681118392,
      "learning_rate": 0.00018970326914836884,
      "loss": 1.5859,
      "step": 4928
    },
    {
      "epoch": 3.2642384105960263,
      "grad_norm": 0.9878997188060116,
      "learning_rate": 0.0001896388981260836,
      "loss": 1.8047,
      "step": 4929
    },
    {
      "epoch": 3.2649006622516556,
      "grad_norm": 0.9672244983198531,
      "learning_rate": 0.00018957451925484203,
      "loss": 1.6797,
      "step": 4930
    },
    {
      "epoch": 3.2655629139072846,
      "grad_norm": 0.9635703770644034,
      "learning_rate": 0.00018951013254739195,
      "loss": 1.8047,
      "step": 4931
    },
    {
      "epoch": 3.266225165562914,
      "grad_norm": 0.9019249306808154,
      "learning_rate": 0.00018944573801648267,
      "loss": 1.625,
      "step": 4932
    },
    {
      "epoch": 3.266887417218543,
      "grad_norm": 1.025381551015555,
      "learning_rate": 0.00018938133567486501,
      "loss": 1.8516,
      "step": 4933
    },
    {
      "epoch": 3.2675496688741723,
      "grad_norm": 1.003662708592068,
      "learning_rate": 0.00018931692553529142,
      "loss": 1.8594,
      "step": 4934
    },
    {
      "epoch": 3.2682119205298013,
      "grad_norm": 1.0074175809342523,
      "learning_rate": 0.00018925250761051582,
      "loss": 1.8672,
      "step": 4935
    },
    {
      "epoch": 3.2688741721854306,
      "grad_norm": 0.8485323414378746,
      "learning_rate": 0.00018918808191329366,
      "loss": 1.4922,
      "step": 4936
    },
    {
      "epoch": 3.2695364238410596,
      "grad_norm": 1.02746066209781,
      "learning_rate": 0.00018912364845638204,
      "loss": 2.0156,
      "step": 4937
    },
    {
      "epoch": 3.2701986754966885,
      "grad_norm": 1.036354689109393,
      "learning_rate": 0.0001890592072525395,
      "loss": 2.0156,
      "step": 4938
    },
    {
      "epoch": 3.270860927152318,
      "grad_norm": 0.8701163608368833,
      "learning_rate": 0.0001889947583145261,
      "loss": 1.6094,
      "step": 4939
    },
    {
      "epoch": 3.271523178807947,
      "grad_norm": 0.8703096632149003,
      "learning_rate": 0.00018893030165510353,
      "loss": 1.5312,
      "step": 4940
    },
    {
      "epoch": 3.2721854304635762,
      "grad_norm": 0.9407916867551112,
      "learning_rate": 0.00018886583728703487,
      "loss": 1.7109,
      "step": 4941
    },
    {
      "epoch": 3.272847682119205,
      "grad_norm": 0.9323798438595865,
      "learning_rate": 0.00018880136522308484,
      "loss": 1.6719,
      "step": 4942
    },
    {
      "epoch": 3.2735099337748346,
      "grad_norm": 0.9184087167045276,
      "learning_rate": 0.00018873688547601966,
      "loss": 1.5391,
      "step": 4943
    },
    {
      "epoch": 3.2741721854304635,
      "grad_norm": 0.8492989005262241,
      "learning_rate": 0.00018867239805860707,
      "loss": 1.4375,
      "step": 4944
    },
    {
      "epoch": 3.274834437086093,
      "grad_norm": 1.0555789870980972,
      "learning_rate": 0.0001886079029836163,
      "loss": 1.8281,
      "step": 4945
    },
    {
      "epoch": 3.275496688741722,
      "grad_norm": 0.952399223300368,
      "learning_rate": 0.00018854340026381808,
      "loss": 1.5859,
      "step": 4946
    },
    {
      "epoch": 3.276158940397351,
      "grad_norm": 0.9521886941476095,
      "learning_rate": 0.00018847888991198477,
      "loss": 1.6406,
      "step": 4947
    },
    {
      "epoch": 3.27682119205298,
      "grad_norm": 1.0454151990248994,
      "learning_rate": 0.00018841437194089006,
      "loss": 1.8281,
      "step": 4948
    },
    {
      "epoch": 3.277483443708609,
      "grad_norm": 1.0383014133253852,
      "learning_rate": 0.00018834984636330936,
      "loss": 1.8281,
      "step": 4949
    },
    {
      "epoch": 3.2781456953642385,
      "grad_norm": 0.9871398546157205,
      "learning_rate": 0.0001882853131920194,
      "loss": 1.7734,
      "step": 4950
    },
    {
      "epoch": 3.2788079470198674,
      "grad_norm": 0.8235358582185949,
      "learning_rate": 0.00018822077243979844,
      "loss": 1.4375,
      "step": 4951
    },
    {
      "epoch": 3.279470198675497,
      "grad_norm": 0.8652282082550177,
      "learning_rate": 0.0001881562241194264,
      "loss": 1.5234,
      "step": 4952
    },
    {
      "epoch": 3.2801324503311258,
      "grad_norm": 0.8796517015548668,
      "learning_rate": 0.00018809166824368452,
      "loss": 1.5938,
      "step": 4953
    },
    {
      "epoch": 3.280794701986755,
      "grad_norm": 0.8918248184392334,
      "learning_rate": 0.00018802710482535562,
      "loss": 1.6484,
      "step": 4954
    },
    {
      "epoch": 3.281456953642384,
      "grad_norm": 0.9027303931693318,
      "learning_rate": 0.00018796253387722402,
      "loss": 1.6406,
      "step": 4955
    },
    {
      "epoch": 3.282119205298013,
      "grad_norm": 0.9102768669930302,
      "learning_rate": 0.00018789795541207548,
      "loss": 1.7109,
      "step": 4956
    },
    {
      "epoch": 3.2827814569536424,
      "grad_norm": 0.8809029837344639,
      "learning_rate": 0.00018783336944269723,
      "loss": 1.5469,
      "step": 4957
    },
    {
      "epoch": 3.2834437086092714,
      "grad_norm": 0.8798101989204969,
      "learning_rate": 0.00018776877598187817,
      "loss": 1.6016,
      "step": 4958
    },
    {
      "epoch": 3.2841059602649008,
      "grad_norm": 0.9967923148959271,
      "learning_rate": 0.0001877041750424084,
      "loss": 1.8438,
      "step": 4959
    },
    {
      "epoch": 3.2847682119205297,
      "grad_norm": 0.9344468974112051,
      "learning_rate": 0.00018763956663707976,
      "loss": 1.8203,
      "step": 4960
    },
    {
      "epoch": 3.285430463576159,
      "grad_norm": 1.0223131026847796,
      "learning_rate": 0.00018757495077868536,
      "loss": 1.8203,
      "step": 4961
    },
    {
      "epoch": 3.286092715231788,
      "grad_norm": 0.943235459345669,
      "learning_rate": 0.0001875103274800199,
      "loss": 1.5781,
      "step": 4962
    },
    {
      "epoch": 3.2867549668874174,
      "grad_norm": 0.8851655098542037,
      "learning_rate": 0.0001874456967538796,
      "loss": 1.5625,
      "step": 4963
    },
    {
      "epoch": 3.2874172185430464,
      "grad_norm": 0.9680180748877525,
      "learning_rate": 0.00018738105861306208,
      "loss": 1.5312,
      "step": 4964
    },
    {
      "epoch": 3.2880794701986753,
      "grad_norm": 0.9269914596287598,
      "learning_rate": 0.00018731641307036628,
      "loss": 1.7188,
      "step": 4965
    },
    {
      "epoch": 3.2887417218543047,
      "grad_norm": 0.8303317714731553,
      "learning_rate": 0.00018725176013859295,
      "loss": 1.4062,
      "step": 4966
    },
    {
      "epoch": 3.2894039735099336,
      "grad_norm": 0.8663931058472737,
      "learning_rate": 0.00018718709983054404,
      "loss": 1.5156,
      "step": 4967
    },
    {
      "epoch": 3.290066225165563,
      "grad_norm": 0.903862199090456,
      "learning_rate": 0.000187122432159023,
      "loss": 1.5234,
      "step": 4968
    },
    {
      "epoch": 3.290728476821192,
      "grad_norm": 0.9895259617786025,
      "learning_rate": 0.00018705775713683483,
      "loss": 1.7188,
      "step": 4969
    },
    {
      "epoch": 3.2913907284768213,
      "grad_norm": 0.9833364379227905,
      "learning_rate": 0.00018699307477678588,
      "loss": 1.7188,
      "step": 4970
    },
    {
      "epoch": 3.2920529801324503,
      "grad_norm": 0.940702014814214,
      "learning_rate": 0.000186928385091684,
      "loss": 1.75,
      "step": 4971
    },
    {
      "epoch": 3.2927152317880797,
      "grad_norm": 0.9986038652067661,
      "learning_rate": 0.0001868636880943385,
      "loss": 1.6094,
      "step": 4972
    },
    {
      "epoch": 3.2933774834437086,
      "grad_norm": 0.9622274174238085,
      "learning_rate": 0.00018679898379756014,
      "loss": 1.6328,
      "step": 4973
    },
    {
      "epoch": 3.2940397350993376,
      "grad_norm": 1.042295730814121,
      "learning_rate": 0.0001867342722141611,
      "loss": 1.8125,
      "step": 4974
    },
    {
      "epoch": 3.294701986754967,
      "grad_norm": 1.0940875053193069,
      "learning_rate": 0.00018666955335695502,
      "loss": 2.0156,
      "step": 4975
    },
    {
      "epoch": 3.295364238410596,
      "grad_norm": 1.0595773742522536,
      "learning_rate": 0.00018660482723875698,
      "loss": 1.8281,
      "step": 4976
    },
    {
      "epoch": 3.2960264900662253,
      "grad_norm": 0.8607914897807912,
      "learning_rate": 0.00018654009387238349,
      "loss": 1.5078,
      "step": 4977
    },
    {
      "epoch": 3.296688741721854,
      "grad_norm": 0.9447447461810932,
      "learning_rate": 0.00018647535327065247,
      "loss": 1.7031,
      "step": 4978
    },
    {
      "epoch": 3.2973509933774836,
      "grad_norm": 0.9803853128274533,
      "learning_rate": 0.00018641060544638332,
      "loss": 1.8125,
      "step": 4979
    },
    {
      "epoch": 3.2980132450331126,
      "grad_norm": 0.9776586241238924,
      "learning_rate": 0.00018634585041239684,
      "loss": 1.8281,
      "step": 4980
    },
    {
      "epoch": 3.298675496688742,
      "grad_norm": 0.8806533123957256,
      "learning_rate": 0.0001862810881815154,
      "loss": 1.8203,
      "step": 4981
    },
    {
      "epoch": 3.299337748344371,
      "grad_norm": 0.821539990769179,
      "learning_rate": 0.00018621631876656242,
      "loss": 1.375,
      "step": 4982
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.954521860008864,
      "learning_rate": 0.00018615154218036317,
      "loss": 1.8984,
      "step": 4983
    },
    {
      "epoch": 3.300662251655629,
      "grad_norm": 0.9274993237580643,
      "learning_rate": 0.00018608675843574413,
      "loss": 1.7891,
      "step": 4984
    },
    {
      "epoch": 3.301324503311258,
      "grad_norm": 0.8524896713381565,
      "learning_rate": 0.00018602196754553314,
      "loss": 1.5625,
      "step": 4985
    },
    {
      "epoch": 3.3019867549668875,
      "grad_norm": 0.8518998129475894,
      "learning_rate": 0.0001859571695225596,
      "loss": 1.4844,
      "step": 4986
    },
    {
      "epoch": 3.3026490066225165,
      "grad_norm": 0.985641121847216,
      "learning_rate": 0.00018589236437965435,
      "loss": 1.8516,
      "step": 4987
    },
    {
      "epoch": 3.303311258278146,
      "grad_norm": 0.9162685674783501,
      "learning_rate": 0.00018582755212964935,
      "loss": 1.6641,
      "step": 4988
    },
    {
      "epoch": 3.303973509933775,
      "grad_norm": 0.8836736930529665,
      "learning_rate": 0.00018576273278537837,
      "loss": 1.6484,
      "step": 4989
    },
    {
      "epoch": 3.304635761589404,
      "grad_norm": 0.7997712518198936,
      "learning_rate": 0.00018569790635967624,
      "loss": 1.3672,
      "step": 4990
    },
    {
      "epoch": 3.305298013245033,
      "grad_norm": 0.9483711102590494,
      "learning_rate": 0.0001856330728653794,
      "loss": 1.5078,
      "step": 4991
    },
    {
      "epoch": 3.305960264900662,
      "grad_norm": 0.8960855638603586,
      "learning_rate": 0.00018556823231532556,
      "loss": 1.625,
      "step": 4992
    },
    {
      "epoch": 3.3066225165562915,
      "grad_norm": 1.0905356204584076,
      "learning_rate": 0.00018550338472235407,
      "loss": 1.8984,
      "step": 4993
    },
    {
      "epoch": 3.3072847682119204,
      "grad_norm": 1.049593545490903,
      "learning_rate": 0.00018543853009930525,
      "loss": 1.7109,
      "step": 4994
    },
    {
      "epoch": 3.30794701986755,
      "grad_norm": 0.9577978630492061,
      "learning_rate": 0.00018537366845902125,
      "loss": 1.7344,
      "step": 4995
    },
    {
      "epoch": 3.3086092715231787,
      "grad_norm": 0.9651349669733016,
      "learning_rate": 0.0001853087998143453,
      "loss": 1.7266,
      "step": 4996
    },
    {
      "epoch": 3.309271523178808,
      "grad_norm": 0.9237230639236741,
      "learning_rate": 0.00018524392417812222,
      "loss": 1.7109,
      "step": 4997
    },
    {
      "epoch": 3.309933774834437,
      "grad_norm": 0.9145469599746658,
      "learning_rate": 0.00018517904156319808,
      "loss": 1.6641,
      "step": 4998
    },
    {
      "epoch": 3.3105960264900665,
      "grad_norm": 0.9335655630951304,
      "learning_rate": 0.0001851141519824204,
      "loss": 1.7812,
      "step": 4999
    },
    {
      "epoch": 3.3112582781456954,
      "grad_norm": 0.9339079484818458,
      "learning_rate": 0.00018504925544863806,
      "loss": 1.75,
      "step": 5000
    },
    {
      "epoch": 3.3119205298013243,
      "grad_norm": 0.9574596902728908,
      "learning_rate": 0.0001849843519747013,
      "loss": 1.8047,
      "step": 5001
    },
    {
      "epoch": 3.3125827814569537,
      "grad_norm": 0.9622339369788938,
      "learning_rate": 0.0001849194415734618,
      "loss": 1.8281,
      "step": 5002
    },
    {
      "epoch": 3.3132450331125827,
      "grad_norm": 1.0108825522391516,
      "learning_rate": 0.0001848545242577725,
      "loss": 1.8047,
      "step": 5003
    },
    {
      "epoch": 3.313907284768212,
      "grad_norm": 0.9059615359449151,
      "learning_rate": 0.00018478960004048785,
      "loss": 1.7266,
      "step": 5004
    },
    {
      "epoch": 3.314569536423841,
      "grad_norm": 0.9031193684197732,
      "learning_rate": 0.00018472466893446347,
      "loss": 1.7109,
      "step": 5005
    },
    {
      "epoch": 3.3152317880794704,
      "grad_norm": 0.8582760031650488,
      "learning_rate": 0.00018465973095255656,
      "loss": 1.5547,
      "step": 5006
    },
    {
      "epoch": 3.3158940397350993,
      "grad_norm": 0.8941598387946792,
      "learning_rate": 0.00018459478610762558,
      "loss": 1.5312,
      "step": 5007
    },
    {
      "epoch": 3.3165562913907287,
      "grad_norm": 0.9574608327299471,
      "learning_rate": 0.00018452983441253034,
      "loss": 1.8281,
      "step": 5008
    },
    {
      "epoch": 3.3172185430463577,
      "grad_norm": 0.8337914799636005,
      "learning_rate": 0.000184464875880132,
      "loss": 1.5391,
      "step": 5009
    },
    {
      "epoch": 3.3178807947019866,
      "grad_norm": 0.9120848933452215,
      "learning_rate": 0.00018439991052329315,
      "loss": 1.6953,
      "step": 5010
    },
    {
      "epoch": 3.318543046357616,
      "grad_norm": 0.9481178581214507,
      "learning_rate": 0.00018433493835487755,
      "loss": 1.6484,
      "step": 5011
    },
    {
      "epoch": 3.319205298013245,
      "grad_norm": 0.880103436228372,
      "learning_rate": 0.00018426995938775057,
      "loss": 1.4844,
      "step": 5012
    },
    {
      "epoch": 3.3198675496688743,
      "grad_norm": 0.9941958887331018,
      "learning_rate": 0.00018420497363477874,
      "loss": 1.6875,
      "step": 5013
    },
    {
      "epoch": 3.3205298013245033,
      "grad_norm": 0.9811146122730885,
      "learning_rate": 0.00018413998110882997,
      "loss": 1.6328,
      "step": 5014
    },
    {
      "epoch": 3.321192052980132,
      "grad_norm": 0.8690144622538178,
      "learning_rate": 0.0001840749818227735,
      "loss": 1.5469,
      "step": 5015
    },
    {
      "epoch": 3.3218543046357616,
      "grad_norm": 1.0421103787115655,
      "learning_rate": 0.00018400997578948004,
      "loss": 1.7188,
      "step": 5016
    },
    {
      "epoch": 3.322516556291391,
      "grad_norm": 0.9776795557393647,
      "learning_rate": 0.00018394496302182143,
      "loss": 1.7344,
      "step": 5017
    },
    {
      "epoch": 3.32317880794702,
      "grad_norm": 0.8167499405481764,
      "learning_rate": 0.00018387994353267096,
      "loss": 1.3828,
      "step": 5018
    },
    {
      "epoch": 3.323841059602649,
      "grad_norm": 1.0179867303789103,
      "learning_rate": 0.00018381491733490332,
      "loss": 1.7422,
      "step": 5019
    },
    {
      "epoch": 3.3245033112582782,
      "grad_norm": 0.8771750048790115,
      "learning_rate": 0.00018374988444139431,
      "loss": 1.5156,
      "step": 5020
    },
    {
      "epoch": 3.325165562913907,
      "grad_norm": 1.0817401158917053,
      "learning_rate": 0.00018368484486502124,
      "loss": 2.1094,
      "step": 5021
    },
    {
      "epoch": 3.3258278145695366,
      "grad_norm": 0.9188472060843526,
      "learning_rate": 0.00018361979861866278,
      "loss": 1.8125,
      "step": 5022
    },
    {
      "epoch": 3.3264900662251655,
      "grad_norm": 0.8255974302938327,
      "learning_rate": 0.00018355474571519868,
      "loss": 1.5234,
      "step": 5023
    },
    {
      "epoch": 3.3271523178807945,
      "grad_norm": 0.88238951083261,
      "learning_rate": 0.00018348968616751024,
      "loss": 1.6094,
      "step": 5024
    },
    {
      "epoch": 3.327814569536424,
      "grad_norm": 0.8026687114985431,
      "learning_rate": 0.00018342461998848003,
      "loss": 1.4609,
      "step": 5025
    },
    {
      "epoch": 3.328476821192053,
      "grad_norm": 1.1027280149205263,
      "learning_rate": 0.0001833595471909918,
      "loss": 2.1562,
      "step": 5026
    },
    {
      "epoch": 3.329139072847682,
      "grad_norm": 0.9345685778614579,
      "learning_rate": 0.0001832944677879308,
      "loss": 1.6094,
      "step": 5027
    },
    {
      "epoch": 3.329801324503311,
      "grad_norm": 0.9346015360278925,
      "learning_rate": 0.0001832293817921835,
      "loss": 1.8672,
      "step": 5028
    },
    {
      "epoch": 3.3304635761589405,
      "grad_norm": 0.9400707482711643,
      "learning_rate": 0.00018316428921663763,
      "loss": 1.8359,
      "step": 5029
    },
    {
      "epoch": 3.3311258278145695,
      "grad_norm": 0.987169996816523,
      "learning_rate": 0.0001830991900741823,
      "loss": 2.0625,
      "step": 5030
    },
    {
      "epoch": 3.331788079470199,
      "grad_norm": 0.9834782079646976,
      "learning_rate": 0.00018303408437770783,
      "loss": 1.8047,
      "step": 5031
    },
    {
      "epoch": 3.332450331125828,
      "grad_norm": 0.9914640845494984,
      "learning_rate": 0.00018296897214010595,
      "loss": 1.8203,
      "step": 5032
    },
    {
      "epoch": 3.3331125827814567,
      "grad_norm": 0.8976719812447882,
      "learning_rate": 0.0001829038533742696,
      "loss": 1.6016,
      "step": 5033
    },
    {
      "epoch": 3.333774834437086,
      "grad_norm": 0.7800758534756888,
      "learning_rate": 0.00018283872809309303,
      "loss": 1.3984,
      "step": 5034
    },
    {
      "epoch": 3.334437086092715,
      "grad_norm": 0.8929383388945731,
      "learning_rate": 0.00018277359630947186,
      "loss": 1.6797,
      "step": 5035
    },
    {
      "epoch": 3.3350993377483444,
      "grad_norm": 0.9837256524164946,
      "learning_rate": 0.00018270845803630286,
      "loss": 1.7109,
      "step": 5036
    },
    {
      "epoch": 3.3357615894039734,
      "grad_norm": 0.9833429636484584,
      "learning_rate": 0.00018264331328648415,
      "loss": 1.8359,
      "step": 5037
    },
    {
      "epoch": 3.3364238410596028,
      "grad_norm": 0.9992347770867407,
      "learning_rate": 0.0001825781620729152,
      "loss": 1.6172,
      "step": 5038
    },
    {
      "epoch": 3.3370860927152317,
      "grad_norm": 0.992486990087195,
      "learning_rate": 0.00018251300440849666,
      "loss": 1.7109,
      "step": 5039
    },
    {
      "epoch": 3.337748344370861,
      "grad_norm": 0.948886943612022,
      "learning_rate": 0.0001824478403061305,
      "loss": 1.6641,
      "step": 5040
    },
    {
      "epoch": 3.33841059602649,
      "grad_norm": 0.8743937954867568,
      "learning_rate": 0.00018238266977871996,
      "loss": 1.3516,
      "step": 5041
    },
    {
      "epoch": 3.339072847682119,
      "grad_norm": 0.948388915531691,
      "learning_rate": 0.00018231749283916952,
      "loss": 1.5547,
      "step": 5042
    },
    {
      "epoch": 3.3397350993377484,
      "grad_norm": 1.084187145765075,
      "learning_rate": 0.00018225230950038503,
      "loss": 1.8516,
      "step": 5043
    },
    {
      "epoch": 3.3403973509933773,
      "grad_norm": 0.9419551206500083,
      "learning_rate": 0.00018218711977527346,
      "loss": 1.5859,
      "step": 5044
    },
    {
      "epoch": 3.3410596026490067,
      "grad_norm": 1.0638050780973562,
      "learning_rate": 0.0001821219236767432,
      "loss": 1.9609,
      "step": 5045
    },
    {
      "epoch": 3.3417218543046356,
      "grad_norm": 0.9221734484489416,
      "learning_rate": 0.00018205672121770372,
      "loss": 1.7344,
      "step": 5046
    },
    {
      "epoch": 3.342384105960265,
      "grad_norm": 0.8810407645169892,
      "learning_rate": 0.00018199151241106597,
      "loss": 1.4453,
      "step": 5047
    },
    {
      "epoch": 3.343046357615894,
      "grad_norm": 1.031099908174573,
      "learning_rate": 0.000181926297269742,
      "loss": 1.8438,
      "step": 5048
    },
    {
      "epoch": 3.3437086092715234,
      "grad_norm": 0.8860284071959196,
      "learning_rate": 0.00018186107580664515,
      "loss": 1.5859,
      "step": 5049
    },
    {
      "epoch": 3.3443708609271523,
      "grad_norm": 0.9649460197630083,
      "learning_rate": 0.00018179584803468995,
      "loss": 1.7266,
      "step": 5050
    },
    {
      "epoch": 3.3450331125827812,
      "grad_norm": 0.9124202943425093,
      "learning_rate": 0.0001817306139667924,
      "loss": 1.5859,
      "step": 5051
    },
    {
      "epoch": 3.3456953642384106,
      "grad_norm": 0.9037979728665405,
      "learning_rate": 0.00018166537361586943,
      "loss": 1.5781,
      "step": 5052
    },
    {
      "epoch": 3.3463576158940396,
      "grad_norm": 1.0290571187327713,
      "learning_rate": 0.0001816001269948395,
      "loss": 1.8984,
      "step": 5053
    },
    {
      "epoch": 3.347019867549669,
      "grad_norm": 1.0449616638772539,
      "learning_rate": 0.00018153487411662214,
      "loss": 1.7656,
      "step": 5054
    },
    {
      "epoch": 3.347682119205298,
      "grad_norm": 1.0316468345319405,
      "learning_rate": 0.00018146961499413816,
      "loss": 1.9844,
      "step": 5055
    },
    {
      "epoch": 3.3483443708609273,
      "grad_norm": 0.8319563223450991,
      "learning_rate": 0.0001814043496403096,
      "loss": 1.375,
      "step": 5056
    },
    {
      "epoch": 3.3490066225165562,
      "grad_norm": 0.9343107230003916,
      "learning_rate": 0.00018133907806805984,
      "loss": 1.7344,
      "step": 5057
    },
    {
      "epoch": 3.3496688741721856,
      "grad_norm": 1.022443823532356,
      "learning_rate": 0.00018127380029031324,
      "loss": 1.7344,
      "step": 5058
    },
    {
      "epoch": 3.3503311258278146,
      "grad_norm": 1.01762922722523,
      "learning_rate": 0.0001812085163199957,
      "loss": 1.7734,
      "step": 5059
    },
    {
      "epoch": 3.3509933774834435,
      "grad_norm": 0.9968573220230004,
      "learning_rate": 0.00018114322617003413,
      "loss": 1.5469,
      "step": 5060
    },
    {
      "epoch": 3.351655629139073,
      "grad_norm": 0.9751367972087437,
      "learning_rate": 0.00018107792985335672,
      "loss": 1.7656,
      "step": 5061
    },
    {
      "epoch": 3.352317880794702,
      "grad_norm": 0.9640972340488173,
      "learning_rate": 0.0001810126273828929,
      "loss": 1.7578,
      "step": 5062
    },
    {
      "epoch": 3.352980132450331,
      "grad_norm": 0.9802542266216752,
      "learning_rate": 0.00018094731877157333,
      "loss": 1.7422,
      "step": 5063
    },
    {
      "epoch": 3.35364238410596,
      "grad_norm": 0.9388307632989934,
      "learning_rate": 0.00018088200403232983,
      "loss": 1.6484,
      "step": 5064
    },
    {
      "epoch": 3.3543046357615895,
      "grad_norm": 0.9609260919424933,
      "learning_rate": 0.0001808166831780955,
      "loss": 1.7188,
      "step": 5065
    },
    {
      "epoch": 3.3549668874172185,
      "grad_norm": 1.0610135633655273,
      "learning_rate": 0.00018075135622180457,
      "loss": 1.8828,
      "step": 5066
    },
    {
      "epoch": 3.355629139072848,
      "grad_norm": 0.8991894359222248,
      "learning_rate": 0.00018068602317639255,
      "loss": 1.5,
      "step": 5067
    },
    {
      "epoch": 3.356291390728477,
      "grad_norm": 1.0465344277467956,
      "learning_rate": 0.00018062068405479614,
      "loss": 1.8125,
      "step": 5068
    },
    {
      "epoch": 3.3569536423841058,
      "grad_norm": 1.0180421713451087,
      "learning_rate": 0.00018055533886995325,
      "loss": 1.6641,
      "step": 5069
    },
    {
      "epoch": 3.357615894039735,
      "grad_norm": 0.960124802528361,
      "learning_rate": 0.00018048998763480295,
      "loss": 1.6719,
      "step": 5070
    },
    {
      "epoch": 3.358278145695364,
      "grad_norm": 0.9502964674719154,
      "learning_rate": 0.00018042463036228556,
      "loss": 1.6875,
      "step": 5071
    },
    {
      "epoch": 3.3589403973509935,
      "grad_norm": 1.0749719657070982,
      "learning_rate": 0.00018035926706534255,
      "loss": 1.9766,
      "step": 5072
    },
    {
      "epoch": 3.3596026490066224,
      "grad_norm": 0.8406925620681531,
      "learning_rate": 0.00018029389775691657,
      "loss": 1.4375,
      "step": 5073
    },
    {
      "epoch": 3.360264900662252,
      "grad_norm": 0.9780892615733453,
      "learning_rate": 0.00018022852244995157,
      "loss": 1.875,
      "step": 5074
    },
    {
      "epoch": 3.3609271523178808,
      "grad_norm": 0.8682228548621989,
      "learning_rate": 0.00018016314115739253,
      "loss": 1.5391,
      "step": 5075
    },
    {
      "epoch": 3.36158940397351,
      "grad_norm": 0.9309727136669333,
      "learning_rate": 0.00018009775389218573,
      "loss": 1.7734,
      "step": 5076
    },
    {
      "epoch": 3.362251655629139,
      "grad_norm": 0.8893599767171835,
      "learning_rate": 0.00018003236066727862,
      "loss": 1.7266,
      "step": 5077
    },
    {
      "epoch": 3.362913907284768,
      "grad_norm": 0.9350769335309351,
      "learning_rate": 0.00017996696149561977,
      "loss": 1.8594,
      "step": 5078
    },
    {
      "epoch": 3.3635761589403974,
      "grad_norm": 0.8558348576534965,
      "learning_rate": 0.000179901556390159,
      "loss": 1.5234,
      "step": 5079
    },
    {
      "epoch": 3.3642384105960264,
      "grad_norm": 0.9071987013320861,
      "learning_rate": 0.00017983614536384727,
      "loss": 1.6797,
      "step": 5080
    },
    {
      "epoch": 3.3649006622516557,
      "grad_norm": 0.8668205201882796,
      "learning_rate": 0.00017977072842963665,
      "loss": 1.6562,
      "step": 5081
    },
    {
      "epoch": 3.3655629139072847,
      "grad_norm": 0.9364387986560406,
      "learning_rate": 0.00017970530560048052,
      "loss": 1.8047,
      "step": 5082
    },
    {
      "epoch": 3.366225165562914,
      "grad_norm": 0.941052400467578,
      "learning_rate": 0.00017963987688933333,
      "loss": 1.6953,
      "step": 5083
    },
    {
      "epoch": 3.366887417218543,
      "grad_norm": 0.9277656406501404,
      "learning_rate": 0.0001795744423091507,
      "loss": 1.8203,
      "step": 5084
    },
    {
      "epoch": 3.3675496688741724,
      "grad_norm": 0.9449287247129142,
      "learning_rate": 0.0001795090018728894,
      "loss": 1.8828,
      "step": 5085
    },
    {
      "epoch": 3.3682119205298013,
      "grad_norm": 0.9672444783816346,
      "learning_rate": 0.0001794435555935075,
      "loss": 1.8594,
      "step": 5086
    },
    {
      "epoch": 3.3688741721854303,
      "grad_norm": 0.9957896559548454,
      "learning_rate": 0.00017937810348396394,
      "loss": 1.9453,
      "step": 5087
    },
    {
      "epoch": 3.3695364238410597,
      "grad_norm": 0.986517051079045,
      "learning_rate": 0.0001793126455572191,
      "loss": 1.7812,
      "step": 5088
    },
    {
      "epoch": 3.3701986754966886,
      "grad_norm": 0.8636512093455239,
      "learning_rate": 0.00017924718182623443,
      "loss": 1.6094,
      "step": 5089
    },
    {
      "epoch": 3.370860927152318,
      "grad_norm": 0.9110108067689228,
      "learning_rate": 0.00017918171230397244,
      "loss": 1.6797,
      "step": 5090
    },
    {
      "epoch": 3.371523178807947,
      "grad_norm": 1.165778743061743,
      "learning_rate": 0.00017911623700339683,
      "loss": 1.8828,
      "step": 5091
    },
    {
      "epoch": 3.3721854304635763,
      "grad_norm": 0.9212506783652876,
      "learning_rate": 0.0001790507559374725,
      "loss": 1.6484,
      "step": 5092
    },
    {
      "epoch": 3.3728476821192053,
      "grad_norm": 0.8754487821143622,
      "learning_rate": 0.0001789852691191654,
      "loss": 1.625,
      "step": 5093
    },
    {
      "epoch": 3.3735099337748347,
      "grad_norm": 0.9159833008898494,
      "learning_rate": 0.0001789197765614428,
      "loss": 1.6562,
      "step": 5094
    },
    {
      "epoch": 3.3741721854304636,
      "grad_norm": 0.8783513636468169,
      "learning_rate": 0.0001788542782772728,
      "loss": 1.5234,
      "step": 5095
    },
    {
      "epoch": 3.3748344370860925,
      "grad_norm": 0.9197782442711238,
      "learning_rate": 0.00017878877427962494,
      "loss": 1.6328,
      "step": 5096
    },
    {
      "epoch": 3.375496688741722,
      "grad_norm": 0.9554336499177547,
      "learning_rate": 0.0001787232645814697,
      "loss": 1.6641,
      "step": 5097
    },
    {
      "epoch": 3.376158940397351,
      "grad_norm": 0.9125179128302443,
      "learning_rate": 0.0001786577491957788,
      "loss": 1.5938,
      "step": 5098
    },
    {
      "epoch": 3.3768211920529803,
      "grad_norm": 1.0084577877741088,
      "learning_rate": 0.00017859222813552497,
      "loss": 1.6562,
      "step": 5099
    },
    {
      "epoch": 3.377483443708609,
      "grad_norm": 1.0410624958083203,
      "learning_rate": 0.0001785267014136822,
      "loss": 1.8828,
      "step": 5100
    },
    {
      "epoch": 3.3781456953642386,
      "grad_norm": 1.0131893977822088,
      "learning_rate": 0.0001784611690432255,
      "loss": 1.7109,
      "step": 5101
    },
    {
      "epoch": 3.3788079470198675,
      "grad_norm": 0.9458835609804696,
      "learning_rate": 0.000178395631037131,
      "loss": 1.5859,
      "step": 5102
    },
    {
      "epoch": 3.379470198675497,
      "grad_norm": 1.0156015632569515,
      "learning_rate": 0.00017833008740837606,
      "loss": 1.8984,
      "step": 5103
    },
    {
      "epoch": 3.380132450331126,
      "grad_norm": 0.9756965648383552,
      "learning_rate": 0.00017826453816993898,
      "loss": 1.7031,
      "step": 5104
    },
    {
      "epoch": 3.380794701986755,
      "grad_norm": 0.8956392749755535,
      "learning_rate": 0.00017819898333479932,
      "loss": 1.5703,
      "step": 5105
    },
    {
      "epoch": 3.381456953642384,
      "grad_norm": 0.9963481266964629,
      "learning_rate": 0.00017813342291593765,
      "loss": 1.8594,
      "step": 5106
    },
    {
      "epoch": 3.382119205298013,
      "grad_norm": 1.0091932522462537,
      "learning_rate": 0.00017806785692633573,
      "loss": 1.9844,
      "step": 5107
    },
    {
      "epoch": 3.3827814569536425,
      "grad_norm": 0.8878873652868379,
      "learning_rate": 0.0001780022853789763,
      "loss": 1.6562,
      "step": 5108
    },
    {
      "epoch": 3.3834437086092715,
      "grad_norm": 0.9670764897878065,
      "learning_rate": 0.00017793670828684339,
      "loss": 1.875,
      "step": 5109
    },
    {
      "epoch": 3.384105960264901,
      "grad_norm": 0.8255076547604595,
      "learning_rate": 0.00017787112566292193,
      "loss": 1.6094,
      "step": 5110
    },
    {
      "epoch": 3.38476821192053,
      "grad_norm": 0.8818807173340291,
      "learning_rate": 0.00017780553752019804,
      "loss": 1.6641,
      "step": 5111
    },
    {
      "epoch": 3.385430463576159,
      "grad_norm": 0.9090337399738706,
      "learning_rate": 0.00017773994387165895,
      "loss": 1.625,
      "step": 5112
    },
    {
      "epoch": 3.386092715231788,
      "grad_norm": 0.8490336544998451,
      "learning_rate": 0.00017767434473029298,
      "loss": 1.4453,
      "step": 5113
    },
    {
      "epoch": 3.386754966887417,
      "grad_norm": 0.9893413967551508,
      "learning_rate": 0.0001776087401090894,
      "loss": 1.8125,
      "step": 5114
    },
    {
      "epoch": 3.3874172185430464,
      "grad_norm": 0.9641637277538008,
      "learning_rate": 0.00017754313002103883,
      "loss": 1.6406,
      "step": 5115
    },
    {
      "epoch": 3.3880794701986754,
      "grad_norm": 0.9318992993731228,
      "learning_rate": 0.00017747751447913269,
      "loss": 1.6016,
      "step": 5116
    },
    {
      "epoch": 3.388741721854305,
      "grad_norm": 0.9101877559318672,
      "learning_rate": 0.00017741189349636374,
      "loss": 1.5547,
      "step": 5117
    },
    {
      "epoch": 3.3894039735099337,
      "grad_norm": 0.974597500722578,
      "learning_rate": 0.00017734626708572554,
      "loss": 1.7422,
      "step": 5118
    },
    {
      "epoch": 3.390066225165563,
      "grad_norm": 1.0820275836193503,
      "learning_rate": 0.000177280635260213,
      "loss": 1.7812,
      "step": 5119
    },
    {
      "epoch": 3.390728476821192,
      "grad_norm": 0.9185536801502157,
      "learning_rate": 0.00017721499803282186,
      "loss": 1.5469,
      "step": 5120
    },
    {
      "epoch": 3.3913907284768214,
      "grad_norm": 1.0103915650128934,
      "learning_rate": 0.00017714935541654912,
      "loss": 1.7812,
      "step": 5121
    },
    {
      "epoch": 3.3920529801324504,
      "grad_norm": 1.054022166284303,
      "learning_rate": 0.00017708370742439272,
      "loss": 1.9688,
      "step": 5122
    },
    {
      "epoch": 3.3927152317880793,
      "grad_norm": 0.9219987100440208,
      "learning_rate": 0.0001770180540693518,
      "loss": 1.625,
      "step": 5123
    },
    {
      "epoch": 3.3933774834437087,
      "grad_norm": 1.0058784411001633,
      "learning_rate": 0.0001769523953644264,
      "loss": 1.7188,
      "step": 5124
    },
    {
      "epoch": 3.3940397350993377,
      "grad_norm": 0.9165941477436577,
      "learning_rate": 0.0001768867313226177,
      "loss": 1.6484,
      "step": 5125
    },
    {
      "epoch": 3.394701986754967,
      "grad_norm": 0.9888203613782737,
      "learning_rate": 0.000176821061956928,
      "loss": 1.7422,
      "step": 5126
    },
    {
      "epoch": 3.395364238410596,
      "grad_norm": 0.9129513486955185,
      "learning_rate": 0.00017675538728036052,
      "loss": 1.6328,
      "step": 5127
    },
    {
      "epoch": 3.396026490066225,
      "grad_norm": 0.8710059058136744,
      "learning_rate": 0.0001766897073059196,
      "loss": 1.5703,
      "step": 5128
    },
    {
      "epoch": 3.3966887417218543,
      "grad_norm": 0.8423861063369497,
      "learning_rate": 0.00017662402204661064,
      "loss": 1.4844,
      "step": 5129
    },
    {
      "epoch": 3.3973509933774833,
      "grad_norm": 1.0017613191912975,
      "learning_rate": 0.0001765583315154401,
      "loss": 1.7422,
      "step": 5130
    },
    {
      "epoch": 3.3980132450331126,
      "grad_norm": 1.0220260167071629,
      "learning_rate": 0.00017649263572541542,
      "loss": 1.7344,
      "step": 5131
    },
    {
      "epoch": 3.3986754966887416,
      "grad_norm": 1.013103145234632,
      "learning_rate": 0.00017642693468954517,
      "loss": 1.8516,
      "step": 5132
    },
    {
      "epoch": 3.399337748344371,
      "grad_norm": 0.9769217142885969,
      "learning_rate": 0.00017636122842083882,
      "loss": 1.7891,
      "step": 5133
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0383267855881078,
      "learning_rate": 0.00017629551693230708,
      "loss": 1.8906,
      "step": 5134
    },
    {
      "epoch": 3.4006622516556293,
      "grad_norm": 0.9269581507474266,
      "learning_rate": 0.00017622980023696152,
      "loss": 1.5234,
      "step": 5135
    },
    {
      "epoch": 3.4013245033112582,
      "grad_norm": 0.9226033334110777,
      "learning_rate": 0.00017616407834781476,
      "loss": 1.5938,
      "step": 5136
    },
    {
      "epoch": 3.401986754966887,
      "grad_norm": 0.9081231875994198,
      "learning_rate": 0.0001760983512778805,
      "loss": 1.5859,
      "step": 5137
    },
    {
      "epoch": 3.4026490066225166,
      "grad_norm": 1.0738185585901567,
      "learning_rate": 0.00017603261904017357,
      "loss": 1.9531,
      "step": 5138
    },
    {
      "epoch": 3.4033112582781455,
      "grad_norm": 0.8832305354231387,
      "learning_rate": 0.00017596688164770954,
      "loss": 1.5156,
      "step": 5139
    },
    {
      "epoch": 3.403973509933775,
      "grad_norm": 1.0400026640035083,
      "learning_rate": 0.00017590113911350528,
      "loss": 1.8984,
      "step": 5140
    },
    {
      "epoch": 3.404635761589404,
      "grad_norm": 0.8944807683302377,
      "learning_rate": 0.00017583539145057852,
      "loss": 1.6016,
      "step": 5141
    },
    {
      "epoch": 3.4052980132450332,
      "grad_norm": 0.9260783917645355,
      "learning_rate": 0.00017576963867194805,
      "loss": 1.6797,
      "step": 5142
    },
    {
      "epoch": 3.405960264900662,
      "grad_norm": 0.8835608765662691,
      "learning_rate": 0.00017570388079063366,
      "loss": 1.6641,
      "step": 5143
    },
    {
      "epoch": 3.4066225165562916,
      "grad_norm": 0.8816293366083017,
      "learning_rate": 0.00017563811781965628,
      "loss": 1.5156,
      "step": 5144
    },
    {
      "epoch": 3.4072847682119205,
      "grad_norm": 0.8811289852256803,
      "learning_rate": 0.00017557234977203754,
      "loss": 1.5859,
      "step": 5145
    },
    {
      "epoch": 3.4079470198675494,
      "grad_norm": 1.0573460701966,
      "learning_rate": 0.00017550657666080046,
      "loss": 1.8672,
      "step": 5146
    },
    {
      "epoch": 3.408609271523179,
      "grad_norm": 0.8225149793388778,
      "learning_rate": 0.0001754407984989687,
      "loss": 1.4375,
      "step": 5147
    },
    {
      "epoch": 3.4092715231788078,
      "grad_norm": 0.7935449828389942,
      "learning_rate": 0.00017537501529956716,
      "loss": 1.4062,
      "step": 5148
    },
    {
      "epoch": 3.409933774834437,
      "grad_norm": 1.0137609996665957,
      "learning_rate": 0.00017530922707562172,
      "loss": 1.7344,
      "step": 5149
    },
    {
      "epoch": 3.410596026490066,
      "grad_norm": 0.9846339830912962,
      "learning_rate": 0.00017524343384015914,
      "loss": 1.6953,
      "step": 5150
    },
    {
      "epoch": 3.4112582781456955,
      "grad_norm": 0.8512413079628269,
      "learning_rate": 0.00017517763560620725,
      "loss": 1.4375,
      "step": 5151
    },
    {
      "epoch": 3.4119205298013244,
      "grad_norm": 1.0022531353037152,
      "learning_rate": 0.00017511183238679492,
      "loss": 1.7969,
      "step": 5152
    },
    {
      "epoch": 3.412582781456954,
      "grad_norm": 0.9778620353582559,
      "learning_rate": 0.00017504602419495178,
      "loss": 1.7812,
      "step": 5153
    },
    {
      "epoch": 3.4132450331125828,
      "grad_norm": 0.8350125507052268,
      "learning_rate": 0.00017498021104370878,
      "loss": 1.3594,
      "step": 5154
    },
    {
      "epoch": 3.4139072847682117,
      "grad_norm": 0.9690444733007956,
      "learning_rate": 0.00017491439294609756,
      "loss": 1.6641,
      "step": 5155
    },
    {
      "epoch": 3.414569536423841,
      "grad_norm": 0.9627171325812037,
      "learning_rate": 0.00017484856991515095,
      "loss": 1.6875,
      "step": 5156
    },
    {
      "epoch": 3.41523178807947,
      "grad_norm": 0.851727492951156,
      "learning_rate": 0.00017478274196390257,
      "loss": 1.5625,
      "step": 5157
    },
    {
      "epoch": 3.4158940397350994,
      "grad_norm": 1.0453021078697722,
      "learning_rate": 0.00017471690910538716,
      "loss": 2.0312,
      "step": 5158
    },
    {
      "epoch": 3.4165562913907284,
      "grad_norm": 0.9830777749073366,
      "learning_rate": 0.0001746510713526404,
      "loss": 1.6016,
      "step": 5159
    },
    {
      "epoch": 3.4172185430463577,
      "grad_norm": 0.9971480024196377,
      "learning_rate": 0.0001745852287186989,
      "loss": 1.875,
      "step": 5160
    },
    {
      "epoch": 3.4178807947019867,
      "grad_norm": 0.9942757053527151,
      "learning_rate": 0.00017451938121660024,
      "loss": 1.9219,
      "step": 5161
    },
    {
      "epoch": 3.418543046357616,
      "grad_norm": 1.0050534685934427,
      "learning_rate": 0.000174453528859383,
      "loss": 1.9375,
      "step": 5162
    },
    {
      "epoch": 3.419205298013245,
      "grad_norm": 0.9301307388383542,
      "learning_rate": 0.0001743876716600867,
      "loss": 1.7969,
      "step": 5163
    },
    {
      "epoch": 3.419867549668874,
      "grad_norm": 1.0910609991504707,
      "learning_rate": 0.0001743218096317518,
      "loss": 1.9375,
      "step": 5164
    },
    {
      "epoch": 3.4205298013245033,
      "grad_norm": 0.9431145080024445,
      "learning_rate": 0.00017425594278741975,
      "loss": 1.6094,
      "step": 5165
    },
    {
      "epoch": 3.4211920529801323,
      "grad_norm": 0.9253719953549182,
      "learning_rate": 0.00017419007114013294,
      "loss": 1.7578,
      "step": 5166
    },
    {
      "epoch": 3.4218543046357617,
      "grad_norm": 0.8843643292777894,
      "learning_rate": 0.00017412419470293474,
      "loss": 1.7109,
      "step": 5167
    },
    {
      "epoch": 3.4225165562913906,
      "grad_norm": 0.8593766949201391,
      "learning_rate": 0.00017405831348886938,
      "loss": 1.7109,
      "step": 5168
    },
    {
      "epoch": 3.42317880794702,
      "grad_norm": 0.9578827339813413,
      "learning_rate": 0.00017399242751098214,
      "loss": 1.8047,
      "step": 5169
    },
    {
      "epoch": 3.423841059602649,
      "grad_norm": 0.8219353680274456,
      "learning_rate": 0.00017392653678231922,
      "loss": 1.4297,
      "step": 5170
    },
    {
      "epoch": 3.4245033112582783,
      "grad_norm": 0.937618782490708,
      "learning_rate": 0.00017386064131592768,
      "loss": 1.7188,
      "step": 5171
    },
    {
      "epoch": 3.4251655629139073,
      "grad_norm": 0.8926713982758337,
      "learning_rate": 0.0001737947411248556,
      "loss": 1.5625,
      "step": 5172
    },
    {
      "epoch": 3.4258278145695362,
      "grad_norm": 0.8275875525709161,
      "learning_rate": 0.00017372883622215202,
      "loss": 1.4219,
      "step": 5173
    },
    {
      "epoch": 3.4264900662251656,
      "grad_norm": 0.9529976715954195,
      "learning_rate": 0.00017366292662086675,
      "loss": 1.6719,
      "step": 5174
    },
    {
      "epoch": 3.4271523178807946,
      "grad_norm": 0.9124424848710261,
      "learning_rate": 0.0001735970123340508,
      "loss": 1.5781,
      "step": 5175
    },
    {
      "epoch": 3.427814569536424,
      "grad_norm": 0.9632432252481405,
      "learning_rate": 0.00017353109337475583,
      "loss": 1.6484,
      "step": 5176
    },
    {
      "epoch": 3.428476821192053,
      "grad_norm": 0.962037185201734,
      "learning_rate": 0.00017346516975603462,
      "loss": 1.6953,
      "step": 5177
    },
    {
      "epoch": 3.4291390728476823,
      "grad_norm": 1.0312288289682048,
      "learning_rate": 0.00017339924149094077,
      "loss": 1.7891,
      "step": 5178
    },
    {
      "epoch": 3.429801324503311,
      "grad_norm": 0.9986741960623842,
      "learning_rate": 0.00017333330859252885,
      "loss": 1.6719,
      "step": 5179
    },
    {
      "epoch": 3.4304635761589406,
      "grad_norm": 0.9407574450157301,
      "learning_rate": 0.00017326737107385432,
      "loss": 1.6562,
      "step": 5180
    },
    {
      "epoch": 3.4311258278145695,
      "grad_norm": 1.1083206945456445,
      "learning_rate": 0.00017320142894797357,
      "loss": 1.9844,
      "step": 5181
    },
    {
      "epoch": 3.4317880794701985,
      "grad_norm": 0.9130389249446954,
      "learning_rate": 0.00017313548222794384,
      "loss": 1.5625,
      "step": 5182
    },
    {
      "epoch": 3.432450331125828,
      "grad_norm": 0.9985952866795484,
      "learning_rate": 0.00017306953092682345,
      "loss": 1.7969,
      "step": 5183
    },
    {
      "epoch": 3.433112582781457,
      "grad_norm": 0.9345722211104205,
      "learning_rate": 0.00017300357505767145,
      "loss": 1.6797,
      "step": 5184
    },
    {
      "epoch": 3.433774834437086,
      "grad_norm": 0.9942148398657046,
      "learning_rate": 0.00017293761463354782,
      "loss": 1.8125,
      "step": 5185
    },
    {
      "epoch": 3.434437086092715,
      "grad_norm": 0.9175002432868609,
      "learning_rate": 0.00017287164966751356,
      "loss": 1.7812,
      "step": 5186
    },
    {
      "epoch": 3.4350993377483445,
      "grad_norm": 0.8078161634006769,
      "learning_rate": 0.00017280568017263045,
      "loss": 1.4531,
      "step": 5187
    },
    {
      "epoch": 3.4357615894039735,
      "grad_norm": 0.9462517171189211,
      "learning_rate": 0.00017273970616196117,
      "loss": 1.7344,
      "step": 5188
    },
    {
      "epoch": 3.436423841059603,
      "grad_norm": 0.9940237190080945,
      "learning_rate": 0.00017267372764856941,
      "loss": 1.7812,
      "step": 5189
    },
    {
      "epoch": 3.437086092715232,
      "grad_norm": 0.8548880566761075,
      "learning_rate": 0.0001726077446455196,
      "loss": 1.5156,
      "step": 5190
    },
    {
      "epoch": 3.4377483443708607,
      "grad_norm": 0.9833874378071286,
      "learning_rate": 0.00017254175716587718,
      "loss": 1.9766,
      "step": 5191
    },
    {
      "epoch": 3.43841059602649,
      "grad_norm": 0.9044997317576567,
      "learning_rate": 0.00017247576522270842,
      "loss": 1.6016,
      "step": 5192
    },
    {
      "epoch": 3.439072847682119,
      "grad_norm": 0.9245921145990159,
      "learning_rate": 0.00017240976882908045,
      "loss": 1.6328,
      "step": 5193
    },
    {
      "epoch": 3.4397350993377485,
      "grad_norm": 0.8612237107283524,
      "learning_rate": 0.00017234376799806132,
      "loss": 1.3438,
      "step": 5194
    },
    {
      "epoch": 3.4403973509933774,
      "grad_norm": 1.0544629423580019,
      "learning_rate": 0.00017227776274271995,
      "loss": 1.9141,
      "step": 5195
    },
    {
      "epoch": 3.441059602649007,
      "grad_norm": 0.9283501469804176,
      "learning_rate": 0.00017221175307612625,
      "loss": 1.5703,
      "step": 5196
    },
    {
      "epoch": 3.4417218543046357,
      "grad_norm": 0.9652977305624869,
      "learning_rate": 0.00017214573901135066,
      "loss": 1.6484,
      "step": 5197
    },
    {
      "epoch": 3.442384105960265,
      "grad_norm": 1.0340862167885854,
      "learning_rate": 0.00017207972056146494,
      "loss": 1.8203,
      "step": 5198
    },
    {
      "epoch": 3.443046357615894,
      "grad_norm": 0.9776205777261562,
      "learning_rate": 0.00017201369773954138,
      "loss": 1.7109,
      "step": 5199
    },
    {
      "epoch": 3.443708609271523,
      "grad_norm": 0.992653793883981,
      "learning_rate": 0.0001719476705586533,
      "loss": 1.8828,
      "step": 5200
    },
    {
      "epoch": 3.4443708609271524,
      "grad_norm": 0.9493260288157698,
      "learning_rate": 0.00017188163903187478,
      "loss": 1.6328,
      "step": 5201
    },
    {
      "epoch": 3.4450331125827813,
      "grad_norm": 1.1148339790373614,
      "learning_rate": 0.00017181560317228097,
      "loss": 1.7656,
      "step": 5202
    },
    {
      "epoch": 3.4456953642384107,
      "grad_norm": 0.9340067036068305,
      "learning_rate": 0.00017174956299294752,
      "loss": 1.7578,
      "step": 5203
    },
    {
      "epoch": 3.4463576158940397,
      "grad_norm": 0.9394075405131492,
      "learning_rate": 0.00017168351850695133,
      "loss": 1.6562,
      "step": 5204
    },
    {
      "epoch": 3.447019867549669,
      "grad_norm": 0.9813082441785328,
      "learning_rate": 0.00017161746972736981,
      "loss": 1.5078,
      "step": 5205
    },
    {
      "epoch": 3.447682119205298,
      "grad_norm": 0.9214886875005742,
      "learning_rate": 0.0001715514166672815,
      "loss": 1.5781,
      "step": 5206
    },
    {
      "epoch": 3.4483443708609274,
      "grad_norm": 0.9286119216719376,
      "learning_rate": 0.0001714853593397656,
      "loss": 1.7812,
      "step": 5207
    },
    {
      "epoch": 3.4490066225165563,
      "grad_norm": 0.9926078091220786,
      "learning_rate": 0.00017141929775790223,
      "loss": 1.9609,
      "step": 5208
    },
    {
      "epoch": 3.4496688741721853,
      "grad_norm": 0.9505181992828995,
      "learning_rate": 0.00017135323193477226,
      "loss": 1.7344,
      "step": 5209
    },
    {
      "epoch": 3.4503311258278146,
      "grad_norm": 0.8921344413666605,
      "learning_rate": 0.0001712871618834577,
      "loss": 1.5312,
      "step": 5210
    },
    {
      "epoch": 3.4509933774834436,
      "grad_norm": 0.9719192118167114,
      "learning_rate": 0.00017122108761704092,
      "loss": 1.7109,
      "step": 5211
    },
    {
      "epoch": 3.451655629139073,
      "grad_norm": 0.9122406435169303,
      "learning_rate": 0.00017115500914860555,
      "loss": 1.6797,
      "step": 5212
    },
    {
      "epoch": 3.452317880794702,
      "grad_norm": 0.9868875464505232,
      "learning_rate": 0.00017108892649123585,
      "loss": 1.7266,
      "step": 5213
    },
    {
      "epoch": 3.4529801324503313,
      "grad_norm": 0.9799445831968059,
      "learning_rate": 0.0001710228396580169,
      "loss": 1.8203,
      "step": 5214
    },
    {
      "epoch": 3.4536423841059603,
      "grad_norm": 0.9462054016228678,
      "learning_rate": 0.00017095674866203468,
      "loss": 1.7188,
      "step": 5215
    },
    {
      "epoch": 3.4543046357615896,
      "grad_norm": 0.9829135808918863,
      "learning_rate": 0.000170890653516376,
      "loss": 1.8203,
      "step": 5216
    },
    {
      "epoch": 3.4549668874172186,
      "grad_norm": 0.9758074887367897,
      "learning_rate": 0.00017082455423412838,
      "loss": 1.8203,
      "step": 5217
    },
    {
      "epoch": 3.4556291390728475,
      "grad_norm": 0.8931033412785244,
      "learning_rate": 0.0001707584508283803,
      "loss": 1.625,
      "step": 5218
    },
    {
      "epoch": 3.456291390728477,
      "grad_norm": 0.8894440899934166,
      "learning_rate": 0.00017069234331222095,
      "loss": 1.6094,
      "step": 5219
    },
    {
      "epoch": 3.456953642384106,
      "grad_norm": 0.9354605166093602,
      "learning_rate": 0.00017062623169874038,
      "loss": 1.5859,
      "step": 5220
    },
    {
      "epoch": 3.4576158940397352,
      "grad_norm": 0.8595179034065843,
      "learning_rate": 0.0001705601160010295,
      "loss": 1.5859,
      "step": 5221
    },
    {
      "epoch": 3.458278145695364,
      "grad_norm": 0.9767206693405877,
      "learning_rate": 0.00017049399623217994,
      "loss": 1.7109,
      "step": 5222
    },
    {
      "epoch": 3.4589403973509936,
      "grad_norm": 0.9874006782889105,
      "learning_rate": 0.00017042787240528412,
      "loss": 1.7734,
      "step": 5223
    },
    {
      "epoch": 3.4596026490066225,
      "grad_norm": 1.0040262306897425,
      "learning_rate": 0.0001703617445334354,
      "loss": 1.7188,
      "step": 5224
    },
    {
      "epoch": 3.460264900662252,
      "grad_norm": 0.9445113353407504,
      "learning_rate": 0.00017029561262972784,
      "loss": 1.7422,
      "step": 5225
    },
    {
      "epoch": 3.460927152317881,
      "grad_norm": 0.9494424848810213,
      "learning_rate": 0.00017022947670725626,
      "loss": 1.5781,
      "step": 5226
    },
    {
      "epoch": 3.46158940397351,
      "grad_norm": 0.9718869461790585,
      "learning_rate": 0.00017016333677911637,
      "loss": 1.6797,
      "step": 5227
    },
    {
      "epoch": 3.462251655629139,
      "grad_norm": 0.9498185102336367,
      "learning_rate": 0.00017009719285840467,
      "loss": 1.7422,
      "step": 5228
    },
    {
      "epoch": 3.462913907284768,
      "grad_norm": 0.8194112847223924,
      "learning_rate": 0.00017003104495821834,
      "loss": 1.4062,
      "step": 5229
    },
    {
      "epoch": 3.4635761589403975,
      "grad_norm": 0.9250246431600145,
      "learning_rate": 0.00016996489309165544,
      "loss": 1.5234,
      "step": 5230
    },
    {
      "epoch": 3.4642384105960264,
      "grad_norm": 1.0198790943345215,
      "learning_rate": 0.00016989873727181492,
      "loss": 1.9297,
      "step": 5231
    },
    {
      "epoch": 3.4649006622516554,
      "grad_norm": 0.9842678224561896,
      "learning_rate": 0.00016983257751179617,
      "loss": 1.6875,
      "step": 5232
    },
    {
      "epoch": 3.4655629139072848,
      "grad_norm": 0.8330272488671963,
      "learning_rate": 0.00016976641382469978,
      "loss": 1.3672,
      "step": 5233
    },
    {
      "epoch": 3.466225165562914,
      "grad_norm": 0.9792468844088593,
      "learning_rate": 0.0001697002462236268,
      "loss": 1.7031,
      "step": 5234
    },
    {
      "epoch": 3.466887417218543,
      "grad_norm": 1.002084953663449,
      "learning_rate": 0.00016963407472167922,
      "loss": 1.9062,
      "step": 5235
    },
    {
      "epoch": 3.467549668874172,
      "grad_norm": 1.0603825795026105,
      "learning_rate": 0.00016956789933195976,
      "loss": 2.0156,
      "step": 5236
    },
    {
      "epoch": 3.4682119205298014,
      "grad_norm": 0.8962450222557493,
      "learning_rate": 0.00016950172006757188,
      "loss": 1.6094,
      "step": 5237
    },
    {
      "epoch": 3.4688741721854304,
      "grad_norm": 0.9944748601312259,
      "learning_rate": 0.0001694355369416198,
      "loss": 1.8359,
      "step": 5238
    },
    {
      "epoch": 3.4695364238410598,
      "grad_norm": 0.928621485633392,
      "learning_rate": 0.0001693693499672087,
      "loss": 1.7656,
      "step": 5239
    },
    {
      "epoch": 3.4701986754966887,
      "grad_norm": 0.9554765753132255,
      "learning_rate": 0.00016930315915744413,
      "loss": 1.7891,
      "step": 5240
    },
    {
      "epoch": 3.4708609271523176,
      "grad_norm": 0.8309663100079769,
      "learning_rate": 0.0001692369645254328,
      "loss": 1.4844,
      "step": 5241
    },
    {
      "epoch": 3.471523178807947,
      "grad_norm": 0.8459899206018462,
      "learning_rate": 0.00016917076608428196,
      "loss": 1.6328,
      "step": 5242
    },
    {
      "epoch": 3.472185430463576,
      "grad_norm": 0.9249338020546705,
      "learning_rate": 0.00016910456384709962,
      "loss": 1.7109,
      "step": 5243
    },
    {
      "epoch": 3.4728476821192054,
      "grad_norm": 0.9420128900420639,
      "learning_rate": 0.0001690383578269946,
      "loss": 1.7344,
      "step": 5244
    },
    {
      "epoch": 3.4735099337748343,
      "grad_norm": 0.9154703425442362,
      "learning_rate": 0.0001689721480370765,
      "loss": 1.8516,
      "step": 5245
    },
    {
      "epoch": 3.4741721854304637,
      "grad_norm": 0.9115509457711516,
      "learning_rate": 0.00016890593449045554,
      "loss": 1.5547,
      "step": 5246
    },
    {
      "epoch": 3.4748344370860926,
      "grad_norm": 0.9195113380865291,
      "learning_rate": 0.00016883971720024283,
      "loss": 1.6328,
      "step": 5247
    },
    {
      "epoch": 3.475496688741722,
      "grad_norm": 1.0300629991813193,
      "learning_rate": 0.00016877349617955013,
      "loss": 1.9609,
      "step": 5248
    },
    {
      "epoch": 3.476158940397351,
      "grad_norm": 0.9200860492406189,
      "learning_rate": 0.00016870727144148995,
      "loss": 1.75,
      "step": 5249
    },
    {
      "epoch": 3.47682119205298,
      "grad_norm": 1.0466688254251262,
      "learning_rate": 0.00016864104299917556,
      "loss": 1.9141,
      "step": 5250
    },
    {
      "epoch": 3.4774834437086093,
      "grad_norm": 0.9571153679732961,
      "learning_rate": 0.00016857481086572092,
      "loss": 1.7422,
      "step": 5251
    },
    {
      "epoch": 3.4781456953642382,
      "grad_norm": 1.005558675131107,
      "learning_rate": 0.0001685085750542408,
      "loss": 1.9688,
      "step": 5252
    },
    {
      "epoch": 3.4788079470198676,
      "grad_norm": 0.8650513105357885,
      "learning_rate": 0.00016844233557785057,
      "loss": 1.5234,
      "step": 5253
    },
    {
      "epoch": 3.4794701986754966,
      "grad_norm": 0.8834245347261634,
      "learning_rate": 0.00016837609244966652,
      "loss": 1.7109,
      "step": 5254
    },
    {
      "epoch": 3.480132450331126,
      "grad_norm": 0.9949494662009583,
      "learning_rate": 0.00016830984568280547,
      "loss": 1.8047,
      "step": 5255
    },
    {
      "epoch": 3.480794701986755,
      "grad_norm": 0.9711205775089836,
      "learning_rate": 0.00016824359529038507,
      "loss": 1.8984,
      "step": 5256
    },
    {
      "epoch": 3.4814569536423843,
      "grad_norm": 0.8947138616107438,
      "learning_rate": 0.00016817734128552364,
      "loss": 1.6094,
      "step": 5257
    },
    {
      "epoch": 3.482119205298013,
      "grad_norm": 0.8553176349555313,
      "learning_rate": 0.00016811108368134026,
      "loss": 1.6875,
      "step": 5258
    },
    {
      "epoch": 3.482781456953642,
      "grad_norm": 0.895860607198179,
      "learning_rate": 0.0001680448224909546,
      "loss": 1.7109,
      "step": 5259
    },
    {
      "epoch": 3.4834437086092715,
      "grad_norm": 0.9551462077379637,
      "learning_rate": 0.00016797855772748732,
      "loss": 1.6875,
      "step": 5260
    },
    {
      "epoch": 3.4841059602649005,
      "grad_norm": 0.9585888795253112,
      "learning_rate": 0.00016791228940405938,
      "loss": 1.875,
      "step": 5261
    },
    {
      "epoch": 3.48476821192053,
      "grad_norm": 0.8771332527273271,
      "learning_rate": 0.00016784601753379286,
      "loss": 1.5312,
      "step": 5262
    },
    {
      "epoch": 3.485430463576159,
      "grad_norm": 0.8746484470103016,
      "learning_rate": 0.00016777974212981022,
      "loss": 1.5859,
      "step": 5263
    },
    {
      "epoch": 3.486092715231788,
      "grad_norm": 0.9598038288933977,
      "learning_rate": 0.00016771346320523485,
      "loss": 1.8281,
      "step": 5264
    },
    {
      "epoch": 3.486754966887417,
      "grad_norm": 0.9392528817749131,
      "learning_rate": 0.00016764718077319068,
      "loss": 1.6641,
      "step": 5265
    },
    {
      "epoch": 3.4874172185430465,
      "grad_norm": 0.9503590662789897,
      "learning_rate": 0.00016758089484680236,
      "loss": 1.5625,
      "step": 5266
    },
    {
      "epoch": 3.4880794701986755,
      "grad_norm": 1.0724450285528642,
      "learning_rate": 0.0001675146054391953,
      "loss": 1.9297,
      "step": 5267
    },
    {
      "epoch": 3.4887417218543044,
      "grad_norm": 0.9400965140843992,
      "learning_rate": 0.00016744831256349565,
      "loss": 1.6641,
      "step": 5268
    },
    {
      "epoch": 3.489403973509934,
      "grad_norm": 0.9756525924498628,
      "learning_rate": 0.00016738201623282996,
      "loss": 1.6406,
      "step": 5269
    },
    {
      "epoch": 3.4900662251655628,
      "grad_norm": 0.969980022614054,
      "learning_rate": 0.00016731571646032585,
      "loss": 1.5781,
      "step": 5270
    },
    {
      "epoch": 3.490728476821192,
      "grad_norm": 0.8971813730908372,
      "learning_rate": 0.0001672494132591113,
      "loss": 1.5234,
      "step": 5271
    },
    {
      "epoch": 3.491390728476821,
      "grad_norm": 0.9517732045252427,
      "learning_rate": 0.00016718310664231523,
      "loss": 1.7188,
      "step": 5272
    },
    {
      "epoch": 3.4920529801324505,
      "grad_norm": 0.9126961195759552,
      "learning_rate": 0.00016711679662306702,
      "loss": 1.5,
      "step": 5273
    },
    {
      "epoch": 3.4927152317880794,
      "grad_norm": 0.9204594854959108,
      "learning_rate": 0.00016705048321449687,
      "loss": 1.6016,
      "step": 5274
    },
    {
      "epoch": 3.493377483443709,
      "grad_norm": 0.9655831877074936,
      "learning_rate": 0.0001669841664297355,
      "loss": 1.5234,
      "step": 5275
    },
    {
      "epoch": 3.4940397350993377,
      "grad_norm": 0.9461758735625779,
      "learning_rate": 0.00016691784628191452,
      "loss": 1.7734,
      "step": 5276
    },
    {
      "epoch": 3.4947019867549667,
      "grad_norm": 0.9420351792903754,
      "learning_rate": 0.00016685152278416598,
      "loss": 1.625,
      "step": 5277
    },
    {
      "epoch": 3.495364238410596,
      "grad_norm": 0.8992624539972025,
      "learning_rate": 0.00016678519594962276,
      "loss": 1.6094,
      "step": 5278
    },
    {
      "epoch": 3.496026490066225,
      "grad_norm": 1.012086155907385,
      "learning_rate": 0.00016671886579141832,
      "loss": 1.7109,
      "step": 5279
    },
    {
      "epoch": 3.4966887417218544,
      "grad_norm": 0.9666842373900281,
      "learning_rate": 0.00016665253232268676,
      "loss": 1.5938,
      "step": 5280
    },
    {
      "epoch": 3.4973509933774833,
      "grad_norm": 1.0646123567243013,
      "learning_rate": 0.00016658619555656287,
      "loss": 1.9609,
      "step": 5281
    },
    {
      "epoch": 3.4980132450331127,
      "grad_norm": 0.9925466748554195,
      "learning_rate": 0.00016651985550618213,
      "loss": 1.7734,
      "step": 5282
    },
    {
      "epoch": 3.4986754966887417,
      "grad_norm": 0.9918108701806201,
      "learning_rate": 0.0001664535121846806,
      "loss": 1.8281,
      "step": 5283
    },
    {
      "epoch": 3.499337748344371,
      "grad_norm": 0.9261380881931536,
      "learning_rate": 0.000166387165605195,
      "loss": 1.6562,
      "step": 5284
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.8741274561890802,
      "learning_rate": 0.00016632081578086275,
      "loss": 1.5078,
      "step": 5285
    },
    {
      "epoch": 3.500662251655629,
      "grad_norm": 1.0060911809836188,
      "learning_rate": 0.00016625446272482188,
      "loss": 1.8281,
      "step": 5286
    },
    {
      "epoch": 3.5013245033112583,
      "grad_norm": 0.9569351129318284,
      "learning_rate": 0.00016618810645021103,
      "loss": 1.6875,
      "step": 5287
    },
    {
      "epoch": 3.5019867549668873,
      "grad_norm": 0.9321104000352731,
      "learning_rate": 0.00016612174697016948,
      "loss": 1.6172,
      "step": 5288
    },
    {
      "epoch": 3.5026490066225167,
      "grad_norm": 0.9651616673486624,
      "learning_rate": 0.00016605538429783728,
      "loss": 1.8516,
      "step": 5289
    },
    {
      "epoch": 3.5033112582781456,
      "grad_norm": 0.933445569041152,
      "learning_rate": 0.00016598901844635484,
      "loss": 1.6953,
      "step": 5290
    },
    {
      "epoch": 3.503973509933775,
      "grad_norm": 0.901934301030569,
      "learning_rate": 0.00016592264942886348,
      "loss": 1.6328,
      "step": 5291
    },
    {
      "epoch": 3.504635761589404,
      "grad_norm": 0.8250077250213438,
      "learning_rate": 0.00016585627725850498,
      "loss": 1.5234,
      "step": 5292
    },
    {
      "epoch": 3.5052980132450333,
      "grad_norm": 0.8320573737159552,
      "learning_rate": 0.0001657899019484218,
      "loss": 1.4531,
      "step": 5293
    },
    {
      "epoch": 3.5059602649006623,
      "grad_norm": 0.9353399210580502,
      "learning_rate": 0.00016572352351175697,
      "loss": 1.6797,
      "step": 5294
    },
    {
      "epoch": 3.506622516556291,
      "grad_norm": 1.003243494706625,
      "learning_rate": 0.00016565714196165428,
      "loss": 1.8516,
      "step": 5295
    },
    {
      "epoch": 3.5072847682119206,
      "grad_norm": 0.9401139521924904,
      "learning_rate": 0.0001655907573112579,
      "loss": 1.8203,
      "step": 5296
    },
    {
      "epoch": 3.5079470198675495,
      "grad_norm": 1.018118386901662,
      "learning_rate": 0.00016552436957371292,
      "loss": 1.8438,
      "step": 5297
    },
    {
      "epoch": 3.508609271523179,
      "grad_norm": 0.9213721937148905,
      "learning_rate": 0.0001654579787621647,
      "loss": 1.5859,
      "step": 5298
    },
    {
      "epoch": 3.509271523178808,
      "grad_norm": 0.8482252684075946,
      "learning_rate": 0.0001653915848897595,
      "loss": 1.4844,
      "step": 5299
    },
    {
      "epoch": 3.5099337748344372,
      "grad_norm": 0.9403256391577341,
      "learning_rate": 0.00016532518796964408,
      "loss": 1.6562,
      "step": 5300
    },
    {
      "epoch": 3.510596026490066,
      "grad_norm": 1.0416924243612968,
      "learning_rate": 0.0001652587880149657,
      "loss": 2.0469,
      "step": 5301
    },
    {
      "epoch": 3.5112582781456956,
      "grad_norm": 0.9416614362980242,
      "learning_rate": 0.00016519238503887235,
      "loss": 1.6875,
      "step": 5302
    },
    {
      "epoch": 3.5119205298013245,
      "grad_norm": 0.9406254505140788,
      "learning_rate": 0.0001651259790545126,
      "loss": 1.7188,
      "step": 5303
    },
    {
      "epoch": 3.5125827814569535,
      "grad_norm": 0.9774319316061385,
      "learning_rate": 0.0001650595700750356,
      "loss": 1.8906,
      "step": 5304
    },
    {
      "epoch": 3.513245033112583,
      "grad_norm": 0.9867103464541125,
      "learning_rate": 0.00016499315811359104,
      "loss": 1.7031,
      "step": 5305
    },
    {
      "epoch": 3.513907284768212,
      "grad_norm": 0.9589847928558287,
      "learning_rate": 0.00016492674318332933,
      "loss": 1.7188,
      "step": 5306
    },
    {
      "epoch": 3.514569536423841,
      "grad_norm": 0.8966152008905183,
      "learning_rate": 0.00016486032529740128,
      "loss": 1.5859,
      "step": 5307
    },
    {
      "epoch": 3.51523178807947,
      "grad_norm": 0.7386399050134128,
      "learning_rate": 0.00016479390446895852,
      "loss": 1.3359,
      "step": 5308
    },
    {
      "epoch": 3.515894039735099,
      "grad_norm": 0.823873002611102,
      "learning_rate": 0.00016472748071115302,
      "loss": 1.4219,
      "step": 5309
    },
    {
      "epoch": 3.5165562913907285,
      "grad_norm": 0.9215419416073001,
      "learning_rate": 0.00016466105403713748,
      "loss": 1.625,
      "step": 5310
    },
    {
      "epoch": 3.517218543046358,
      "grad_norm": 0.9229146485782297,
      "learning_rate": 0.00016459462446006515,
      "loss": 1.6094,
      "step": 5311
    },
    {
      "epoch": 3.517880794701987,
      "grad_norm": 1.0142740466052895,
      "learning_rate": 0.00016452819199308993,
      "loss": 1.7969,
      "step": 5312
    },
    {
      "epoch": 3.5185430463576157,
      "grad_norm": 1.0373773911987614,
      "learning_rate": 0.0001644617566493661,
      "loss": 1.7734,
      "step": 5313
    },
    {
      "epoch": 3.519205298013245,
      "grad_norm": 1.03877460628737,
      "learning_rate": 0.00016439531844204862,
      "loss": 2.0156,
      "step": 5314
    },
    {
      "epoch": 3.519867549668874,
      "grad_norm": 0.9573624856950872,
      "learning_rate": 0.00016432887738429308,
      "loss": 1.7578,
      "step": 5315
    },
    {
      "epoch": 3.5205298013245034,
      "grad_norm": 0.972271586718286,
      "learning_rate": 0.00016426243348925554,
      "loss": 1.8594,
      "step": 5316
    },
    {
      "epoch": 3.5211920529801324,
      "grad_norm": 1.0027442672793094,
      "learning_rate": 0.0001641959867700927,
      "loss": 1.7188,
      "step": 5317
    },
    {
      "epoch": 3.5218543046357613,
      "grad_norm": 0.9117405587575127,
      "learning_rate": 0.00016412953723996166,
      "loss": 1.6875,
      "step": 5318
    },
    {
      "epoch": 3.5225165562913907,
      "grad_norm": 0.9127624287298804,
      "learning_rate": 0.00016406308491202028,
      "loss": 1.5469,
      "step": 5319
    },
    {
      "epoch": 3.52317880794702,
      "grad_norm": 0.8596008269240704,
      "learning_rate": 0.0001639966297994269,
      "loss": 1.5234,
      "step": 5320
    },
    {
      "epoch": 3.523841059602649,
      "grad_norm": 0.9400024357350738,
      "learning_rate": 0.00016393017191534024,
      "loss": 1.7109,
      "step": 5321
    },
    {
      "epoch": 3.524503311258278,
      "grad_norm": 0.9319391549211633,
      "learning_rate": 0.00016386371127291994,
      "loss": 1.5234,
      "step": 5322
    },
    {
      "epoch": 3.5251655629139074,
      "grad_norm": 0.9176192087423477,
      "learning_rate": 0.00016379724788532585,
      "loss": 1.6641,
      "step": 5323
    },
    {
      "epoch": 3.5258278145695363,
      "grad_norm": 0.9478734226947579,
      "learning_rate": 0.00016373078176571846,
      "loss": 1.6719,
      "step": 5324
    },
    {
      "epoch": 3.5264900662251657,
      "grad_norm": 0.9740121291970824,
      "learning_rate": 0.0001636643129272589,
      "loss": 1.8203,
      "step": 5325
    },
    {
      "epoch": 3.5271523178807946,
      "grad_norm": 0.9670634309004048,
      "learning_rate": 0.00016359784138310872,
      "loss": 1.8281,
      "step": 5326
    },
    {
      "epoch": 3.5278145695364236,
      "grad_norm": 1.1150831716696257,
      "learning_rate": 0.00016353136714643003,
      "loss": 1.8906,
      "step": 5327
    },
    {
      "epoch": 3.528476821192053,
      "grad_norm": 1.0700857593354138,
      "learning_rate": 0.0001634648902303855,
      "loss": 1.7578,
      "step": 5328
    },
    {
      "epoch": 3.5291390728476824,
      "grad_norm": 1.0058119883684582,
      "learning_rate": 0.0001633984106481384,
      "loss": 1.8516,
      "step": 5329
    },
    {
      "epoch": 3.5298013245033113,
      "grad_norm": 1.0434992788607969,
      "learning_rate": 0.00016333192841285233,
      "loss": 2.0625,
      "step": 5330
    },
    {
      "epoch": 3.5304635761589402,
      "grad_norm": 0.9012628605402628,
      "learning_rate": 0.00016326544353769158,
      "loss": 1.6562,
      "step": 5331
    },
    {
      "epoch": 3.5311258278145696,
      "grad_norm": 0.8322996338329111,
      "learning_rate": 0.00016319895603582096,
      "loss": 1.4688,
      "step": 5332
    },
    {
      "epoch": 3.5317880794701986,
      "grad_norm": 0.9008661495937892,
      "learning_rate": 0.00016313246592040571,
      "loss": 1.6797,
      "step": 5333
    },
    {
      "epoch": 3.532450331125828,
      "grad_norm": 0.9187321145513938,
      "learning_rate": 0.00016306597320461165,
      "loss": 1.75,
      "step": 5334
    },
    {
      "epoch": 3.533112582781457,
      "grad_norm": 0.9478436478403233,
      "learning_rate": 0.0001629994779016051,
      "loss": 1.8125,
      "step": 5335
    },
    {
      "epoch": 3.533774834437086,
      "grad_norm": 0.9718965375396881,
      "learning_rate": 0.00016293298002455287,
      "loss": 1.6406,
      "step": 5336
    },
    {
      "epoch": 3.5344370860927152,
      "grad_norm": 0.8292936538070134,
      "learning_rate": 0.00016286647958662235,
      "loss": 1.3906,
      "step": 5337
    },
    {
      "epoch": 3.5350993377483446,
      "grad_norm": 0.9983472822966541,
      "learning_rate": 0.00016279997660098138,
      "loss": 1.7422,
      "step": 5338
    },
    {
      "epoch": 3.5357615894039736,
      "grad_norm": 0.9525656553610059,
      "learning_rate": 0.00016273347108079823,
      "loss": 1.7656,
      "step": 5339
    },
    {
      "epoch": 3.5364238410596025,
      "grad_norm": 0.9207529051251476,
      "learning_rate": 0.0001626669630392419,
      "loss": 1.6641,
      "step": 5340
    },
    {
      "epoch": 3.537086092715232,
      "grad_norm": 0.8906035639535697,
      "learning_rate": 0.00016260045248948162,
      "loss": 1.6094,
      "step": 5341
    },
    {
      "epoch": 3.537748344370861,
      "grad_norm": 0.9652991120405814,
      "learning_rate": 0.0001625339394446873,
      "loss": 1.7422,
      "step": 5342
    },
    {
      "epoch": 3.53841059602649,
      "grad_norm": 0.9870217413823394,
      "learning_rate": 0.00016246742391802928,
      "loss": 1.7344,
      "step": 5343
    },
    {
      "epoch": 3.539072847682119,
      "grad_norm": 0.9189682503745589,
      "learning_rate": 0.0001624009059226784,
      "loss": 1.6094,
      "step": 5344
    },
    {
      "epoch": 3.539735099337748,
      "grad_norm": 1.0390735357272625,
      "learning_rate": 0.00016233438547180596,
      "loss": 1.7734,
      "step": 5345
    },
    {
      "epoch": 3.5403973509933775,
      "grad_norm": 1.0241092171949788,
      "learning_rate": 0.00016226786257858377,
      "loss": 1.7422,
      "step": 5346
    },
    {
      "epoch": 3.541059602649007,
      "grad_norm": 0.922730632725712,
      "learning_rate": 0.00016220133725618416,
      "loss": 1.5469,
      "step": 5347
    },
    {
      "epoch": 3.541721854304636,
      "grad_norm": 0.9616496064607529,
      "learning_rate": 0.00016213480951777988,
      "loss": 1.6641,
      "step": 5348
    },
    {
      "epoch": 3.5423841059602648,
      "grad_norm": 1.083854411177325,
      "learning_rate": 0.00016206827937654422,
      "loss": 2.1094,
      "step": 5349
    },
    {
      "epoch": 3.543046357615894,
      "grad_norm": 0.9185236640848786,
      "learning_rate": 0.00016200174684565086,
      "loss": 1.6406,
      "step": 5350
    },
    {
      "epoch": 3.543708609271523,
      "grad_norm": 0.9771546885209988,
      "learning_rate": 0.00016193521193827405,
      "loss": 1.75,
      "step": 5351
    },
    {
      "epoch": 3.5443708609271525,
      "grad_norm": 1.0834340016136035,
      "learning_rate": 0.00016186867466758841,
      "loss": 2.0469,
      "step": 5352
    },
    {
      "epoch": 3.5450331125827814,
      "grad_norm": 0.8663187373358745,
      "learning_rate": 0.00016180213504676913,
      "loss": 1.5,
      "step": 5353
    },
    {
      "epoch": 3.5456953642384104,
      "grad_norm": 0.927913074240894,
      "learning_rate": 0.0001617355930889918,
      "loss": 1.75,
      "step": 5354
    },
    {
      "epoch": 3.5463576158940397,
      "grad_norm": 0.9486252595529084,
      "learning_rate": 0.00016166904880743254,
      "loss": 1.6094,
      "step": 5355
    },
    {
      "epoch": 3.547019867549669,
      "grad_norm": 0.965354088632048,
      "learning_rate": 0.00016160250221526775,
      "loss": 1.875,
      "step": 5356
    },
    {
      "epoch": 3.547682119205298,
      "grad_norm": 0.8338428783457262,
      "learning_rate": 0.0001615359533256746,
      "loss": 1.5234,
      "step": 5357
    },
    {
      "epoch": 3.548344370860927,
      "grad_norm": 0.9130006045409188,
      "learning_rate": 0.00016146940215183036,
      "loss": 1.7812,
      "step": 5358
    },
    {
      "epoch": 3.5490066225165564,
      "grad_norm": 1.0145154158734289,
      "learning_rate": 0.00016140284870691301,
      "loss": 2.0312,
      "step": 5359
    },
    {
      "epoch": 3.5496688741721854,
      "grad_norm": 0.8876100411554546,
      "learning_rate": 0.0001613362930041009,
      "loss": 1.6719,
      "step": 5360
    },
    {
      "epoch": 3.5503311258278147,
      "grad_norm": 0.9418101170194497,
      "learning_rate": 0.00016126973505657284,
      "loss": 1.6797,
      "step": 5361
    },
    {
      "epoch": 3.5509933774834437,
      "grad_norm": 0.9490533025987453,
      "learning_rate": 0.00016120317487750802,
      "loss": 1.6641,
      "step": 5362
    },
    {
      "epoch": 3.5516556291390726,
      "grad_norm": 0.874417609386111,
      "learning_rate": 0.0001611366124800861,
      "loss": 1.5469,
      "step": 5363
    },
    {
      "epoch": 3.552317880794702,
      "grad_norm": 0.9573053560230277,
      "learning_rate": 0.00016107004787748727,
      "loss": 1.6797,
      "step": 5364
    },
    {
      "epoch": 3.5529801324503314,
      "grad_norm": 1.0297934741878627,
      "learning_rate": 0.00016100348108289206,
      "loss": 1.7812,
      "step": 5365
    },
    {
      "epoch": 3.5536423841059603,
      "grad_norm": 0.8841883924055438,
      "learning_rate": 0.00016093691210948137,
      "loss": 1.5938,
      "step": 5366
    },
    {
      "epoch": 3.5543046357615893,
      "grad_norm": 0.9894702599579677,
      "learning_rate": 0.00016087034097043678,
      "loss": 1.8438,
      "step": 5367
    },
    {
      "epoch": 3.5549668874172187,
      "grad_norm": 0.990306523487853,
      "learning_rate": 0.00016080376767893997,
      "loss": 1.7969,
      "step": 5368
    },
    {
      "epoch": 3.5556291390728476,
      "grad_norm": 0.9899658849023713,
      "learning_rate": 0.00016073719224817337,
      "loss": 1.7891,
      "step": 5369
    },
    {
      "epoch": 3.556291390728477,
      "grad_norm": 0.8311432536383253,
      "learning_rate": 0.00016067061469131956,
      "loss": 1.4141,
      "step": 5370
    },
    {
      "epoch": 3.556953642384106,
      "grad_norm": 0.7984629533567666,
      "learning_rate": 0.00016060403502156172,
      "loss": 1.4375,
      "step": 5371
    },
    {
      "epoch": 3.557615894039735,
      "grad_norm": 0.967955308474612,
      "learning_rate": 0.0001605374532520834,
      "loss": 1.7031,
      "step": 5372
    },
    {
      "epoch": 3.5582781456953643,
      "grad_norm": 0.9761527886101768,
      "learning_rate": 0.00016047086939606852,
      "loss": 1.7422,
      "step": 5373
    },
    {
      "epoch": 3.558940397350993,
      "grad_norm": 0.9340661407231386,
      "learning_rate": 0.00016040428346670145,
      "loss": 1.6094,
      "step": 5374
    },
    {
      "epoch": 3.5596026490066226,
      "grad_norm": 0.9783495099741326,
      "learning_rate": 0.000160337695477167,
      "loss": 1.7344,
      "step": 5375
    },
    {
      "epoch": 3.5602649006622515,
      "grad_norm": 0.9359563624768892,
      "learning_rate": 0.00016027110544065024,
      "loss": 1.5859,
      "step": 5376
    },
    {
      "epoch": 3.560927152317881,
      "grad_norm": 0.9450339624509777,
      "learning_rate": 0.0001602045133703369,
      "loss": 1.6016,
      "step": 5377
    },
    {
      "epoch": 3.56158940397351,
      "grad_norm": 0.9628700639638467,
      "learning_rate": 0.00016013791927941293,
      "loss": 1.6719,
      "step": 5378
    },
    {
      "epoch": 3.5622516556291393,
      "grad_norm": 1.0251631613773962,
      "learning_rate": 0.00016007132318106472,
      "loss": 1.8594,
      "step": 5379
    },
    {
      "epoch": 3.562913907284768,
      "grad_norm": 0.9996603208388756,
      "learning_rate": 0.00016000472508847903,
      "loss": 1.5703,
      "step": 5380
    },
    {
      "epoch": 3.563576158940397,
      "grad_norm": 0.8910190273271855,
      "learning_rate": 0.0001599381250148431,
      "loss": 1.5,
      "step": 5381
    },
    {
      "epoch": 3.5642384105960265,
      "grad_norm": 0.9299813235358839,
      "learning_rate": 0.00015987152297334447,
      "loss": 1.5391,
      "step": 5382
    },
    {
      "epoch": 3.5649006622516555,
      "grad_norm": 1.0491675738589996,
      "learning_rate": 0.0001598049189771711,
      "loss": 1.6797,
      "step": 5383
    },
    {
      "epoch": 3.565562913907285,
      "grad_norm": 0.975461702275182,
      "learning_rate": 0.00015973831303951142,
      "loss": 1.5938,
      "step": 5384
    },
    {
      "epoch": 3.566225165562914,
      "grad_norm": 1.0353038114152482,
      "learning_rate": 0.00015967170517355405,
      "loss": 1.7031,
      "step": 5385
    },
    {
      "epoch": 3.566887417218543,
      "grad_norm": 0.9648528974811715,
      "learning_rate": 0.0001596050953924882,
      "loss": 1.5625,
      "step": 5386
    },
    {
      "epoch": 3.567549668874172,
      "grad_norm": 0.9084991596426929,
      "learning_rate": 0.00015953848370950334,
      "loss": 1.5781,
      "step": 5387
    },
    {
      "epoch": 3.5682119205298015,
      "grad_norm": 0.9983643976619423,
      "learning_rate": 0.00015947187013778933,
      "loss": 1.6797,
      "step": 5388
    },
    {
      "epoch": 3.5688741721854305,
      "grad_norm": 1.0358335739726412,
      "learning_rate": 0.00015940525469053645,
      "loss": 1.7266,
      "step": 5389
    },
    {
      "epoch": 3.5695364238410594,
      "grad_norm": 0.8919323314061441,
      "learning_rate": 0.00015933863738093533,
      "loss": 1.4141,
      "step": 5390
    },
    {
      "epoch": 3.570198675496689,
      "grad_norm": 0.9381757175654588,
      "learning_rate": 0.00015927201822217696,
      "loss": 1.4922,
      "step": 5391
    },
    {
      "epoch": 3.5708609271523177,
      "grad_norm": 1.0589568980792632,
      "learning_rate": 0.00015920539722745264,
      "loss": 1.8125,
      "step": 5392
    },
    {
      "epoch": 3.571523178807947,
      "grad_norm": 1.0028699686031068,
      "learning_rate": 0.00015913877440995415,
      "loss": 1.7344,
      "step": 5393
    },
    {
      "epoch": 3.572185430463576,
      "grad_norm": 1.036546700263265,
      "learning_rate": 0.00015907214978287356,
      "loss": 1.7734,
      "step": 5394
    },
    {
      "epoch": 3.5728476821192054,
      "grad_norm": 0.9379696953110752,
      "learning_rate": 0.00015900552335940327,
      "loss": 1.5625,
      "step": 5395
    },
    {
      "epoch": 3.5735099337748344,
      "grad_norm": 0.9273931399306699,
      "learning_rate": 0.00015893889515273618,
      "loss": 1.5234,
      "step": 5396
    },
    {
      "epoch": 3.5741721854304638,
      "grad_norm": 0.9561556912096334,
      "learning_rate": 0.0001588722651760653,
      "loss": 1.625,
      "step": 5397
    },
    {
      "epoch": 3.5748344370860927,
      "grad_norm": 0.9592435514233916,
      "learning_rate": 0.00015880563344258423,
      "loss": 1.7109,
      "step": 5398
    },
    {
      "epoch": 3.5754966887417217,
      "grad_norm": 0.9792211152231995,
      "learning_rate": 0.00015873899996548678,
      "loss": 1.9297,
      "step": 5399
    },
    {
      "epoch": 3.576158940397351,
      "grad_norm": 0.9881906058986323,
      "learning_rate": 0.00015867236475796719,
      "loss": 1.7422,
      "step": 5400
    },
    {
      "epoch": 3.57682119205298,
      "grad_norm": 0.9391093415271614,
      "learning_rate": 0.00015860572783321989,
      "loss": 1.6641,
      "step": 5401
    },
    {
      "epoch": 3.5774834437086094,
      "grad_norm": 0.9246299320589654,
      "learning_rate": 0.00015853908920443984,
      "loss": 1.6562,
      "step": 5402
    },
    {
      "epoch": 3.5781456953642383,
      "grad_norm": 0.8889564698604083,
      "learning_rate": 0.0001584724488848222,
      "loss": 1.4766,
      "step": 5403
    },
    {
      "epoch": 3.5788079470198677,
      "grad_norm": 0.9876388835322506,
      "learning_rate": 0.00015840580688756262,
      "loss": 1.8516,
      "step": 5404
    },
    {
      "epoch": 3.5794701986754967,
      "grad_norm": 0.9859119781425864,
      "learning_rate": 0.0001583391632258568,
      "loss": 1.6719,
      "step": 5405
    },
    {
      "epoch": 3.580132450331126,
      "grad_norm": 1.0146871885059339,
      "learning_rate": 0.00015827251791290114,
      "loss": 1.8828,
      "step": 5406
    },
    {
      "epoch": 3.580794701986755,
      "grad_norm": 0.942691073926143,
      "learning_rate": 0.0001582058709618921,
      "loss": 1.6094,
      "step": 5407
    },
    {
      "epoch": 3.581456953642384,
      "grad_norm": 1.0201317386083246,
      "learning_rate": 0.00015813922238602652,
      "loss": 1.7812,
      "step": 5408
    },
    {
      "epoch": 3.5821192052980133,
      "grad_norm": 0.8919510088579388,
      "learning_rate": 0.00015807257219850157,
      "loss": 1.5703,
      "step": 5409
    },
    {
      "epoch": 3.5827814569536423,
      "grad_norm": 0.9469981899054225,
      "learning_rate": 0.00015800592041251478,
      "loss": 1.7109,
      "step": 5410
    },
    {
      "epoch": 3.5834437086092716,
      "grad_norm": 0.9967146262916864,
      "learning_rate": 0.00015793926704126397,
      "loss": 1.7656,
      "step": 5411
    },
    {
      "epoch": 3.5841059602649006,
      "grad_norm": 0.9841558606116027,
      "learning_rate": 0.00015787261209794723,
      "loss": 1.7578,
      "step": 5412
    },
    {
      "epoch": 3.58476821192053,
      "grad_norm": 1.042073372583229,
      "learning_rate": 0.00015780595559576312,
      "loss": 1.8828,
      "step": 5413
    },
    {
      "epoch": 3.585430463576159,
      "grad_norm": 1.0103138569457304,
      "learning_rate": 0.00015773929754791025,
      "loss": 1.7578,
      "step": 5414
    },
    {
      "epoch": 3.5860927152317883,
      "grad_norm": 1.0721599492092924,
      "learning_rate": 0.00015767263796758772,
      "loss": 1.9375,
      "step": 5415
    },
    {
      "epoch": 3.5867549668874172,
      "grad_norm": 0.8190898500881091,
      "learning_rate": 0.00015760597686799495,
      "loss": 1.4688,
      "step": 5416
    },
    {
      "epoch": 3.587417218543046,
      "grad_norm": 0.9644837922021989,
      "learning_rate": 0.00015753931426233154,
      "loss": 1.6484,
      "step": 5417
    },
    {
      "epoch": 3.5880794701986756,
      "grad_norm": 0.8153412020978149,
      "learning_rate": 0.00015747265016379745,
      "loss": 1.4766,
      "step": 5418
    },
    {
      "epoch": 3.5887417218543045,
      "grad_norm": 0.9515821289026316,
      "learning_rate": 0.00015740598458559302,
      "loss": 1.8516,
      "step": 5419
    },
    {
      "epoch": 3.589403973509934,
      "grad_norm": 0.8946083230115717,
      "learning_rate": 0.00015733931754091865,
      "loss": 1.5781,
      "step": 5420
    },
    {
      "epoch": 3.590066225165563,
      "grad_norm": 0.9707668224994146,
      "learning_rate": 0.00015727264904297534,
      "loss": 1.8125,
      "step": 5421
    },
    {
      "epoch": 3.590728476821192,
      "grad_norm": 0.8760502090432105,
      "learning_rate": 0.00015720597910496412,
      "loss": 1.7109,
      "step": 5422
    },
    {
      "epoch": 3.591390728476821,
      "grad_norm": 0.9384935661429146,
      "learning_rate": 0.0001571393077400864,
      "loss": 1.6406,
      "step": 5423
    },
    {
      "epoch": 3.5920529801324506,
      "grad_norm": 0.8972093627877783,
      "learning_rate": 0.0001570726349615439,
      "loss": 1.4766,
      "step": 5424
    },
    {
      "epoch": 3.5927152317880795,
      "grad_norm": 1.003480833521167,
      "learning_rate": 0.00015700596078253864,
      "loss": 1.7891,
      "step": 5425
    },
    {
      "epoch": 3.5933774834437084,
      "grad_norm": 1.0478319064574422,
      "learning_rate": 0.00015693928521627278,
      "loss": 1.7656,
      "step": 5426
    },
    {
      "epoch": 3.594039735099338,
      "grad_norm": 1.1018321155042545,
      "learning_rate": 0.0001568726082759489,
      "loss": 1.9219,
      "step": 5427
    },
    {
      "epoch": 3.5947019867549668,
      "grad_norm": 0.9772783293616389,
      "learning_rate": 0.0001568059299747698,
      "loss": 1.8203,
      "step": 5428
    },
    {
      "epoch": 3.595364238410596,
      "grad_norm": 0.9506712134957765,
      "learning_rate": 0.00015673925032593856,
      "loss": 1.7266,
      "step": 5429
    },
    {
      "epoch": 3.596026490066225,
      "grad_norm": 0.9121235943930099,
      "learning_rate": 0.00015667256934265842,
      "loss": 1.6328,
      "step": 5430
    },
    {
      "epoch": 3.596688741721854,
      "grad_norm": 1.0679961223352659,
      "learning_rate": 0.00015660588703813314,
      "loss": 1.8828,
      "step": 5431
    },
    {
      "epoch": 3.5973509933774834,
      "grad_norm": 0.9892472136282454,
      "learning_rate": 0.00015653920342556638,
      "loss": 1.7422,
      "step": 5432
    },
    {
      "epoch": 3.598013245033113,
      "grad_norm": 0.9590418855478537,
      "learning_rate": 0.0001564725185181625,
      "loss": 1.7812,
      "step": 5433
    },
    {
      "epoch": 3.5986754966887418,
      "grad_norm": 0.9473797454728158,
      "learning_rate": 0.00015640583232912566,
      "loss": 1.5938,
      "step": 5434
    },
    {
      "epoch": 3.5993377483443707,
      "grad_norm": 0.9859570800238129,
      "learning_rate": 0.00015633914487166062,
      "loss": 1.7734,
      "step": 5435
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0247291426891405,
      "learning_rate": 0.0001562724561589722,
      "loss": 1.8359,
      "step": 5436
    },
    {
      "epoch": 3.600662251655629,
      "grad_norm": 0.9919425626737989,
      "learning_rate": 0.00015620576620426556,
      "loss": 1.7109,
      "step": 5437
    },
    {
      "epoch": 3.6013245033112584,
      "grad_norm": 0.8605264856257943,
      "learning_rate": 0.00015613907502074606,
      "loss": 1.5,
      "step": 5438
    },
    {
      "epoch": 3.6019867549668874,
      "grad_norm": 0.9779958200009906,
      "learning_rate": 0.00015607238262161933,
      "loss": 1.7109,
      "step": 5439
    },
    {
      "epoch": 3.6026490066225163,
      "grad_norm": 1.0233186906882359,
      "learning_rate": 0.0001560056890200912,
      "loss": 1.7969,
      "step": 5440
    },
    {
      "epoch": 3.6033112582781457,
      "grad_norm": 0.9201585045169206,
      "learning_rate": 0.00015593899422936778,
      "loss": 1.5703,
      "step": 5441
    },
    {
      "epoch": 3.603973509933775,
      "grad_norm": 0.959662461867005,
      "learning_rate": 0.0001558722982626555,
      "loss": 1.6172,
      "step": 5442
    },
    {
      "epoch": 3.604635761589404,
      "grad_norm": 1.0150906420016441,
      "learning_rate": 0.00015580560113316076,
      "loss": 1.9688,
      "step": 5443
    },
    {
      "epoch": 3.605298013245033,
      "grad_norm": 0.8891405051120231,
      "learning_rate": 0.00015573890285409047,
      "loss": 1.5938,
      "step": 5444
    },
    {
      "epoch": 3.6059602649006623,
      "grad_norm": 0.9009181762928509,
      "learning_rate": 0.0001556722034386516,
      "loss": 1.5781,
      "step": 5445
    },
    {
      "epoch": 3.6066225165562913,
      "grad_norm": 1.1155995215838945,
      "learning_rate": 0.00015560550290005145,
      "loss": 1.9141,
      "step": 5446
    },
    {
      "epoch": 3.6072847682119207,
      "grad_norm": 1.0620802518976755,
      "learning_rate": 0.00015553880125149738,
      "loss": 1.9531,
      "step": 5447
    },
    {
      "epoch": 3.6079470198675496,
      "grad_norm": 0.8224461386855975,
      "learning_rate": 0.00015547209850619726,
      "loss": 1.3203,
      "step": 5448
    },
    {
      "epoch": 3.6086092715231786,
      "grad_norm": 0.9598294075484233,
      "learning_rate": 0.00015540539467735886,
      "loss": 1.7344,
      "step": 5449
    },
    {
      "epoch": 3.609271523178808,
      "grad_norm": 0.9778144971954751,
      "learning_rate": 0.0001553386897781903,
      "loss": 1.6484,
      "step": 5450
    },
    {
      "epoch": 3.6099337748344373,
      "grad_norm": 1.0048426328830289,
      "learning_rate": 0.00015527198382190002,
      "loss": 1.7969,
      "step": 5451
    },
    {
      "epoch": 3.6105960264900663,
      "grad_norm": 1.0005411540448157,
      "learning_rate": 0.0001552052768216965,
      "loss": 1.75,
      "step": 5452
    },
    {
      "epoch": 3.611258278145695,
      "grad_norm": 0.8560394317924507,
      "learning_rate": 0.00015513856879078845,
      "loss": 1.5234,
      "step": 5453
    },
    {
      "epoch": 3.6119205298013246,
      "grad_norm": 0.9774997545993043,
      "learning_rate": 0.00015507185974238493,
      "loss": 1.7734,
      "step": 5454
    },
    {
      "epoch": 3.6125827814569536,
      "grad_norm": 1.0343880114248794,
      "learning_rate": 0.00015500514968969496,
      "loss": 1.8906,
      "step": 5455
    },
    {
      "epoch": 3.613245033112583,
      "grad_norm": 0.9345352589876579,
      "learning_rate": 0.000154938438645928,
      "loss": 1.6016,
      "step": 5456
    },
    {
      "epoch": 3.613907284768212,
      "grad_norm": 0.93513857442134,
      "learning_rate": 0.0001548717266242936,
      "loss": 1.4766,
      "step": 5457
    },
    {
      "epoch": 3.614569536423841,
      "grad_norm": 1.0215768853711904,
      "learning_rate": 0.00015480501363800145,
      "loss": 1.6328,
      "step": 5458
    },
    {
      "epoch": 3.61523178807947,
      "grad_norm": 1.0812296879841494,
      "learning_rate": 0.00015473829970026154,
      "loss": 1.7734,
      "step": 5459
    },
    {
      "epoch": 3.6158940397350996,
      "grad_norm": 1.01091091373328,
      "learning_rate": 0.00015467158482428394,
      "loss": 1.6797,
      "step": 5460
    },
    {
      "epoch": 3.6165562913907285,
      "grad_norm": 1.0465564299690684,
      "learning_rate": 0.00015460486902327898,
      "loss": 1.8281,
      "step": 5461
    },
    {
      "epoch": 3.6172185430463575,
      "grad_norm": 0.982072133039844,
      "learning_rate": 0.00015453815231045718,
      "loss": 1.6094,
      "step": 5462
    },
    {
      "epoch": 3.617880794701987,
      "grad_norm": 0.9655544635893837,
      "learning_rate": 0.00015447143469902918,
      "loss": 1.5547,
      "step": 5463
    },
    {
      "epoch": 3.618543046357616,
      "grad_norm": 1.0164231452862569,
      "learning_rate": 0.00015440471620220583,
      "loss": 1.8047,
      "step": 5464
    },
    {
      "epoch": 3.619205298013245,
      "grad_norm": 0.9953279295614368,
      "learning_rate": 0.0001543379968331982,
      "loss": 1.6875,
      "step": 5465
    },
    {
      "epoch": 3.619867549668874,
      "grad_norm": 1.0639397585683934,
      "learning_rate": 0.00015427127660521748,
      "loss": 1.8828,
      "step": 5466
    },
    {
      "epoch": 3.620529801324503,
      "grad_norm": 0.9617729731907882,
      "learning_rate": 0.000154204555531475,
      "loss": 1.8047,
      "step": 5467
    },
    {
      "epoch": 3.6211920529801325,
      "grad_norm": 0.9457230297282906,
      "learning_rate": 0.00015413783362518233,
      "loss": 1.5703,
      "step": 5468
    },
    {
      "epoch": 3.621854304635762,
      "grad_norm": 0.8805514993806806,
      "learning_rate": 0.00015407111089955116,
      "loss": 1.4375,
      "step": 5469
    },
    {
      "epoch": 3.622516556291391,
      "grad_norm": 0.9964779430225779,
      "learning_rate": 0.00015400438736779332,
      "loss": 1.75,
      "step": 5470
    },
    {
      "epoch": 3.6231788079470197,
      "grad_norm": 0.8904535470639685,
      "learning_rate": 0.0001539376630431209,
      "loss": 1.4844,
      "step": 5471
    },
    {
      "epoch": 3.623841059602649,
      "grad_norm": 0.9349304044804878,
      "learning_rate": 0.00015387093793874606,
      "loss": 1.6797,
      "step": 5472
    },
    {
      "epoch": 3.624503311258278,
      "grad_norm": 1.0242792948477977,
      "learning_rate": 0.00015380421206788114,
      "loss": 1.7266,
      "step": 5473
    },
    {
      "epoch": 3.6251655629139075,
      "grad_norm": 0.9918116265299677,
      "learning_rate": 0.00015373748544373863,
      "loss": 1.7734,
      "step": 5474
    },
    {
      "epoch": 3.6258278145695364,
      "grad_norm": 0.9333770872602772,
      "learning_rate": 0.00015367075807953113,
      "loss": 1.7578,
      "step": 5475
    },
    {
      "epoch": 3.6264900662251653,
      "grad_norm": 0.8887581458336022,
      "learning_rate": 0.00015360402998847142,
      "loss": 1.5938,
      "step": 5476
    },
    {
      "epoch": 3.6271523178807947,
      "grad_norm": 0.9221880761289662,
      "learning_rate": 0.0001535373011837725,
      "loss": 1.6797,
      "step": 5477
    },
    {
      "epoch": 3.627814569536424,
      "grad_norm": 0.874205702348191,
      "learning_rate": 0.00015347057167864735,
      "loss": 1.4531,
      "step": 5478
    },
    {
      "epoch": 3.628476821192053,
      "grad_norm": 1.0475042398995513,
      "learning_rate": 0.00015340384148630924,
      "loss": 1.75,
      "step": 5479
    },
    {
      "epoch": 3.629139072847682,
      "grad_norm": 0.9773058547495003,
      "learning_rate": 0.00015333711061997152,
      "loss": 1.5156,
      "step": 5480
    },
    {
      "epoch": 3.6298013245033114,
      "grad_norm": 1.0526254354535474,
      "learning_rate": 0.0001532703790928476,
      "loss": 1.7344,
      "step": 5481
    },
    {
      "epoch": 3.6304635761589403,
      "grad_norm": 1.0073689996216701,
      "learning_rate": 0.0001532036469181511,
      "loss": 1.6875,
      "step": 5482
    },
    {
      "epoch": 3.6311258278145697,
      "grad_norm": 0.875935510454572,
      "learning_rate": 0.00015313691410909588,
      "loss": 1.3125,
      "step": 5483
    },
    {
      "epoch": 3.6317880794701987,
      "grad_norm": 0.9027009282184744,
      "learning_rate": 0.0001530701806788956,
      "loss": 1.5078,
      "step": 5484
    },
    {
      "epoch": 3.6324503311258276,
      "grad_norm": 0.8371768737462985,
      "learning_rate": 0.00015300344664076438,
      "loss": 1.3672,
      "step": 5485
    },
    {
      "epoch": 3.633112582781457,
      "grad_norm": 0.9766885151956994,
      "learning_rate": 0.0001529367120079163,
      "loss": 1.5938,
      "step": 5486
    },
    {
      "epoch": 3.633774834437086,
      "grad_norm": 1.0991650990996824,
      "learning_rate": 0.0001528699767935656,
      "loss": 2.0625,
      "step": 5487
    },
    {
      "epoch": 3.6344370860927153,
      "grad_norm": 1.115275419392536,
      "learning_rate": 0.00015280324101092654,
      "loss": 1.8438,
      "step": 5488
    },
    {
      "epoch": 3.6350993377483443,
      "grad_norm": 1.0121744638211336,
      "learning_rate": 0.00015273650467321367,
      "loss": 1.8984,
      "step": 5489
    },
    {
      "epoch": 3.6357615894039736,
      "grad_norm": 0.9421789207267149,
      "learning_rate": 0.00015266976779364146,
      "loss": 1.5547,
      "step": 5490
    },
    {
      "epoch": 3.6364238410596026,
      "grad_norm": 0.9668593254691284,
      "learning_rate": 0.00015260303038542466,
      "loss": 1.7188,
      "step": 5491
    },
    {
      "epoch": 3.637086092715232,
      "grad_norm": 0.9556438256318691,
      "learning_rate": 0.00015253629246177793,
      "loss": 1.7734,
      "step": 5492
    },
    {
      "epoch": 3.637748344370861,
      "grad_norm": 0.8742868386223109,
      "learning_rate": 0.00015246955403591629,
      "loss": 1.5234,
      "step": 5493
    },
    {
      "epoch": 3.63841059602649,
      "grad_norm": 0.9602324700577255,
      "learning_rate": 0.00015240281512105463,
      "loss": 1.7109,
      "step": 5494
    },
    {
      "epoch": 3.6390728476821192,
      "grad_norm": 0.9328884559263386,
      "learning_rate": 0.00015233607573040801,
      "loss": 1.6797,
      "step": 5495
    },
    {
      "epoch": 3.639735099337748,
      "grad_norm": 0.9405996314645788,
      "learning_rate": 0.00015226933587719158,
      "loss": 1.5938,
      "step": 5496
    },
    {
      "epoch": 3.6403973509933776,
      "grad_norm": 0.9576571673499773,
      "learning_rate": 0.00015220259557462067,
      "loss": 1.5547,
      "step": 5497
    },
    {
      "epoch": 3.6410596026490065,
      "grad_norm": 1.050272470898764,
      "learning_rate": 0.00015213585483591057,
      "loss": 1.9219,
      "step": 5498
    },
    {
      "epoch": 3.641721854304636,
      "grad_norm": 0.9987198495981005,
      "learning_rate": 0.00015206911367427674,
      "loss": 1.7578,
      "step": 5499
    },
    {
      "epoch": 3.642384105960265,
      "grad_norm": 0.996248277055577,
      "learning_rate": 0.00015200237210293469,
      "loss": 1.625,
      "step": 5500
    },
    {
      "epoch": 3.6430463576158942,
      "grad_norm": 0.9590741462116509,
      "learning_rate": 0.00015193563013509998,
      "loss": 1.6016,
      "step": 5501
    },
    {
      "epoch": 3.643708609271523,
      "grad_norm": 0.91603837073998,
      "learning_rate": 0.00015186888778398832,
      "loss": 1.5312,
      "step": 5502
    },
    {
      "epoch": 3.644370860927152,
      "grad_norm": 0.9534665359162454,
      "learning_rate": 0.00015180214506281547,
      "loss": 1.6094,
      "step": 5503
    },
    {
      "epoch": 3.6450331125827815,
      "grad_norm": 0.9672835885698239,
      "learning_rate": 0.0001517354019847972,
      "loss": 1.6875,
      "step": 5504
    },
    {
      "epoch": 3.6456953642384105,
      "grad_norm": 1.0360052841837173,
      "learning_rate": 0.00015166865856314947,
      "loss": 1.7656,
      "step": 5505
    },
    {
      "epoch": 3.64635761589404,
      "grad_norm": 0.9298236501823403,
      "learning_rate": 0.00015160191481108822,
      "loss": 1.5703,
      "step": 5506
    },
    {
      "epoch": 3.647019867549669,
      "grad_norm": 0.8476777562742751,
      "learning_rate": 0.00015153517074182945,
      "loss": 1.3984,
      "step": 5507
    },
    {
      "epoch": 3.647682119205298,
      "grad_norm": 0.8970675329374492,
      "learning_rate": 0.0001514684263685893,
      "loss": 1.4922,
      "step": 5508
    },
    {
      "epoch": 3.648344370860927,
      "grad_norm": 0.9466460120849547,
      "learning_rate": 0.00015140168170458385,
      "loss": 1.5781,
      "step": 5509
    },
    {
      "epoch": 3.6490066225165565,
      "grad_norm": 0.9440707442843377,
      "learning_rate": 0.0001513349367630294,
      "loss": 1.7188,
      "step": 5510
    },
    {
      "epoch": 3.6496688741721854,
      "grad_norm": 1.0377063444488313,
      "learning_rate": 0.0001512681915571421,
      "loss": 1.7422,
      "step": 5511
    },
    {
      "epoch": 3.6503311258278144,
      "grad_norm": 0.9822735634082073,
      "learning_rate": 0.00015120144610013841,
      "loss": 1.625,
      "step": 5512
    },
    {
      "epoch": 3.6509933774834438,
      "grad_norm": 0.8639155918115943,
      "learning_rate": 0.00015113470040523458,
      "loss": 1.4453,
      "step": 5513
    },
    {
      "epoch": 3.6516556291390727,
      "grad_norm": 0.9374240610196501,
      "learning_rate": 0.00015106795448564708,
      "loss": 1.5625,
      "step": 5514
    },
    {
      "epoch": 3.652317880794702,
      "grad_norm": 1.1014439805689693,
      "learning_rate": 0.00015100120835459234,
      "loss": 1.8281,
      "step": 5515
    },
    {
      "epoch": 3.652980132450331,
      "grad_norm": 0.9814054303046156,
      "learning_rate": 0.00015093446202528685,
      "loss": 1.7344,
      "step": 5516
    },
    {
      "epoch": 3.6536423841059604,
      "grad_norm": 0.8684138356787554,
      "learning_rate": 0.0001508677155109472,
      "loss": 1.3984,
      "step": 5517
    },
    {
      "epoch": 3.6543046357615894,
      "grad_norm": 1.0439118104088005,
      "learning_rate": 0.0001508009688247899,
      "loss": 1.7266,
      "step": 5518
    },
    {
      "epoch": 3.6549668874172188,
      "grad_norm": 0.923249160212107,
      "learning_rate": 0.00015073422198003155,
      "loss": 1.5703,
      "step": 5519
    },
    {
      "epoch": 3.6556291390728477,
      "grad_norm": 1.0011446626755878,
      "learning_rate": 0.00015066747498988895,
      "loss": 1.5781,
      "step": 5520
    },
    {
      "epoch": 3.6562913907284766,
      "grad_norm": 1.039299280898437,
      "learning_rate": 0.00015060072786757852,
      "loss": 1.8281,
      "step": 5521
    },
    {
      "epoch": 3.656953642384106,
      "grad_norm": 1.0096593010718178,
      "learning_rate": 0.00015053398062631716,
      "loss": 1.7734,
      "step": 5522
    },
    {
      "epoch": 3.657615894039735,
      "grad_norm": 0.9821805777952813,
      "learning_rate": 0.0001504672332793215,
      "loss": 1.6875,
      "step": 5523
    },
    {
      "epoch": 3.6582781456953644,
      "grad_norm": 0.9079183521622091,
      "learning_rate": 0.0001504004858398083,
      "loss": 1.5,
      "step": 5524
    },
    {
      "epoch": 3.6589403973509933,
      "grad_norm": 1.1327686814064506,
      "learning_rate": 0.00015033373832099427,
      "loss": 1.8203,
      "step": 5525
    },
    {
      "epoch": 3.6596026490066222,
      "grad_norm": 0.9042112388775178,
      "learning_rate": 0.0001502669907360963,
      "loss": 1.5391,
      "step": 5526
    },
    {
      "epoch": 3.6602649006622516,
      "grad_norm": 1.0148392219883262,
      "learning_rate": 0.00015020024309833103,
      "loss": 1.7344,
      "step": 5527
    },
    {
      "epoch": 3.660927152317881,
      "grad_norm": 0.9219452470146404,
      "learning_rate": 0.00015013349542091539,
      "loss": 1.5391,
      "step": 5528
    },
    {
      "epoch": 3.66158940397351,
      "grad_norm": 0.9425000423485886,
      "learning_rate": 0.00015006674771706607,
      "loss": 1.8281,
      "step": 5529
    },
    {
      "epoch": 3.662251655629139,
      "grad_norm": 0.7583464420091842,
      "learning_rate": 0.00015,
      "loss": 1.2891,
      "step": 5530
    },
    {
      "epoch": 3.6629139072847683,
      "grad_norm": 1.0030341484290775,
      "learning_rate": 0.0001499332522829339,
      "loss": 1.7656,
      "step": 5531
    },
    {
      "epoch": 3.6635761589403972,
      "grad_norm": 1.0134292386614951,
      "learning_rate": 0.00014986650457908456,
      "loss": 1.7656,
      "step": 5532
    },
    {
      "epoch": 3.6642384105960266,
      "grad_norm": 1.0119688181796418,
      "learning_rate": 0.00014979975690166897,
      "loss": 1.8281,
      "step": 5533
    },
    {
      "epoch": 3.6649006622516556,
      "grad_norm": 0.9146532036052958,
      "learning_rate": 0.0001497330092639037,
      "loss": 1.5391,
      "step": 5534
    },
    {
      "epoch": 3.6655629139072845,
      "grad_norm": 0.8710071585930678,
      "learning_rate": 0.00014966626167900568,
      "loss": 1.4219,
      "step": 5535
    },
    {
      "epoch": 3.666225165562914,
      "grad_norm": 0.9279411330783591,
      "learning_rate": 0.0001495995141601917,
      "loss": 1.5625,
      "step": 5536
    },
    {
      "epoch": 3.6668874172185433,
      "grad_norm": 1.0350663930933217,
      "learning_rate": 0.00014953276672067847,
      "loss": 1.7031,
      "step": 5537
    },
    {
      "epoch": 3.667549668874172,
      "grad_norm": 1.0128820889897836,
      "learning_rate": 0.0001494660193736828,
      "loss": 1.8594,
      "step": 5538
    },
    {
      "epoch": 3.668211920529801,
      "grad_norm": 1.0317139319114441,
      "learning_rate": 0.00014939927213242145,
      "loss": 1.8203,
      "step": 5539
    },
    {
      "epoch": 3.6688741721854305,
      "grad_norm": 0.9491985540004312,
      "learning_rate": 0.00014933252501011108,
      "loss": 1.75,
      "step": 5540
    },
    {
      "epoch": 3.6695364238410595,
      "grad_norm": 1.0361653749675888,
      "learning_rate": 0.0001492657780199684,
      "loss": 1.75,
      "step": 5541
    },
    {
      "epoch": 3.670198675496689,
      "grad_norm": 1.0256050113708572,
      "learning_rate": 0.0001491990311752101,
      "loss": 1.8203,
      "step": 5542
    },
    {
      "epoch": 3.670860927152318,
      "grad_norm": 0.9881460390296565,
      "learning_rate": 0.00014913228448905285,
      "loss": 1.6953,
      "step": 5543
    },
    {
      "epoch": 3.6715231788079468,
      "grad_norm": 1.0443177739119482,
      "learning_rate": 0.0001490655379747131,
      "loss": 1.7969,
      "step": 5544
    },
    {
      "epoch": 3.672185430463576,
      "grad_norm": 0.9222403091905449,
      "learning_rate": 0.00014899879164540766,
      "loss": 1.6484,
      "step": 5545
    },
    {
      "epoch": 3.6728476821192055,
      "grad_norm": 0.945148596621292,
      "learning_rate": 0.00014893204551435292,
      "loss": 1.5312,
      "step": 5546
    },
    {
      "epoch": 3.6735099337748345,
      "grad_norm": 0.9288247341791913,
      "learning_rate": 0.00014886529959476537,
      "loss": 1.5547,
      "step": 5547
    },
    {
      "epoch": 3.6741721854304634,
      "grad_norm": 0.9952181484905053,
      "learning_rate": 0.00014879855389986156,
      "loss": 1.75,
      "step": 5548
    },
    {
      "epoch": 3.674834437086093,
      "grad_norm": 0.8988603331179746,
      "learning_rate": 0.0001487318084428579,
      "loss": 1.5625,
      "step": 5549
    },
    {
      "epoch": 3.6754966887417218,
      "grad_norm": 0.8599023511872526,
      "learning_rate": 0.0001486650632369706,
      "loss": 1.3281,
      "step": 5550
    },
    {
      "epoch": 3.676158940397351,
      "grad_norm": 0.9919892401932802,
      "learning_rate": 0.00014859831829541612,
      "loss": 1.5703,
      "step": 5551
    },
    {
      "epoch": 3.67682119205298,
      "grad_norm": 0.897263953060312,
      "learning_rate": 0.00014853157363141071,
      "loss": 1.4375,
      "step": 5552
    },
    {
      "epoch": 3.677483443708609,
      "grad_norm": 1.032144369501003,
      "learning_rate": 0.00014846482925817052,
      "loss": 1.7734,
      "step": 5553
    },
    {
      "epoch": 3.6781456953642384,
      "grad_norm": 0.9562754487394266,
      "learning_rate": 0.00014839808518891175,
      "loss": 1.6406,
      "step": 5554
    },
    {
      "epoch": 3.678807947019868,
      "grad_norm": 0.985348141173966,
      "learning_rate": 0.00014833134143685053,
      "loss": 1.5469,
      "step": 5555
    },
    {
      "epoch": 3.6794701986754967,
      "grad_norm": 1.0444345008922429,
      "learning_rate": 0.00014826459801520276,
      "loss": 1.8047,
      "step": 5556
    },
    {
      "epoch": 3.6801324503311257,
      "grad_norm": 1.002745888366712,
      "learning_rate": 0.00014819785493718453,
      "loss": 1.6875,
      "step": 5557
    },
    {
      "epoch": 3.680794701986755,
      "grad_norm": 0.974481678075266,
      "learning_rate": 0.00014813111221601168,
      "loss": 1.6172,
      "step": 5558
    },
    {
      "epoch": 3.681456953642384,
      "grad_norm": 1.1126230037973754,
      "learning_rate": 0.00014806436986490004,
      "loss": 1.8203,
      "step": 5559
    },
    {
      "epoch": 3.6821192052980134,
      "grad_norm": 1.0757588345701155,
      "learning_rate": 0.00014799762789706531,
      "loss": 1.7422,
      "step": 5560
    },
    {
      "epoch": 3.6827814569536423,
      "grad_norm": 0.9551034195794778,
      "learning_rate": 0.00014793088632572324,
      "loss": 1.4609,
      "step": 5561
    },
    {
      "epoch": 3.6834437086092713,
      "grad_norm": 1.0102837414680566,
      "learning_rate": 0.00014786414516408943,
      "loss": 1.7266,
      "step": 5562
    },
    {
      "epoch": 3.6841059602649007,
      "grad_norm": 0.9832592519926433,
      "learning_rate": 0.0001477974044253793,
      "loss": 1.7031,
      "step": 5563
    },
    {
      "epoch": 3.68476821192053,
      "grad_norm": 0.9345885251456978,
      "learning_rate": 0.0001477306641228084,
      "loss": 1.5234,
      "step": 5564
    },
    {
      "epoch": 3.685430463576159,
      "grad_norm": 0.9824194908956344,
      "learning_rate": 0.000147663924269592,
      "loss": 1.8203,
      "step": 5565
    },
    {
      "epoch": 3.686092715231788,
      "grad_norm": 0.9264013293335893,
      "learning_rate": 0.0001475971848789454,
      "loss": 1.625,
      "step": 5566
    },
    {
      "epoch": 3.6867549668874173,
      "grad_norm": 0.9382544908818071,
      "learning_rate": 0.00014753044596408371,
      "loss": 1.5547,
      "step": 5567
    },
    {
      "epoch": 3.6874172185430463,
      "grad_norm": 0.9088663435231644,
      "learning_rate": 0.00014746370753822204,
      "loss": 1.6172,
      "step": 5568
    },
    {
      "epoch": 3.6880794701986757,
      "grad_norm": 0.9252757998979266,
      "learning_rate": 0.00014739696961457534,
      "loss": 1.6562,
      "step": 5569
    },
    {
      "epoch": 3.6887417218543046,
      "grad_norm": 1.0094862449849653,
      "learning_rate": 0.00014733023220635857,
      "loss": 1.8438,
      "step": 5570
    },
    {
      "epoch": 3.6894039735099335,
      "grad_norm": 0.970846480457611,
      "learning_rate": 0.00014726349532678636,
      "loss": 1.6875,
      "step": 5571
    },
    {
      "epoch": 3.690066225165563,
      "grad_norm": 0.9444455615418398,
      "learning_rate": 0.00014719675898907343,
      "loss": 1.7031,
      "step": 5572
    },
    {
      "epoch": 3.6907284768211923,
      "grad_norm": 0.9980956811982235,
      "learning_rate": 0.00014713002320643442,
      "loss": 1.625,
      "step": 5573
    },
    {
      "epoch": 3.6913907284768213,
      "grad_norm": 1.0025500010848614,
      "learning_rate": 0.00014706328799208372,
      "loss": 1.8984,
      "step": 5574
    },
    {
      "epoch": 3.69205298013245,
      "grad_norm": 0.9323150777124989,
      "learning_rate": 0.00014699655335923556,
      "loss": 1.6328,
      "step": 5575
    },
    {
      "epoch": 3.6927152317880796,
      "grad_norm": 0.8944490473037413,
      "learning_rate": 0.00014692981932110439,
      "loss": 1.4531,
      "step": 5576
    },
    {
      "epoch": 3.6933774834437085,
      "grad_norm": 0.869931796322871,
      "learning_rate": 0.00014686308589090418,
      "loss": 1.4531,
      "step": 5577
    },
    {
      "epoch": 3.694039735099338,
      "grad_norm": 0.9703009668262551,
      "learning_rate": 0.00014679635308184884,
      "loss": 1.7109,
      "step": 5578
    },
    {
      "epoch": 3.694701986754967,
      "grad_norm": 0.9644091753060987,
      "learning_rate": 0.0001467296209071524,
      "loss": 1.6641,
      "step": 5579
    },
    {
      "epoch": 3.695364238410596,
      "grad_norm": 0.9035572053381756,
      "learning_rate": 0.0001466628893800285,
      "loss": 1.5469,
      "step": 5580
    },
    {
      "epoch": 3.696026490066225,
      "grad_norm": 1.017841226749287,
      "learning_rate": 0.0001465961585136907,
      "loss": 1.8594,
      "step": 5581
    },
    {
      "epoch": 3.6966887417218546,
      "grad_norm": 0.9989350431442365,
      "learning_rate": 0.00014652942832135265,
      "loss": 1.7266,
      "step": 5582
    },
    {
      "epoch": 3.6973509933774835,
      "grad_norm": 0.9403895232462992,
      "learning_rate": 0.00014646269881622752,
      "loss": 1.7188,
      "step": 5583
    },
    {
      "epoch": 3.6980132450331125,
      "grad_norm": 0.9739936708752042,
      "learning_rate": 0.00014639597001152855,
      "loss": 1.6406,
      "step": 5584
    },
    {
      "epoch": 3.698675496688742,
      "grad_norm": 0.9491168587978074,
      "learning_rate": 0.00014632924192046885,
      "loss": 1.7578,
      "step": 5585
    },
    {
      "epoch": 3.699337748344371,
      "grad_norm": 0.9897125475624153,
      "learning_rate": 0.0001462625145562614,
      "loss": 1.6016,
      "step": 5586
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.8939728112319097,
      "learning_rate": 0.00014619578793211883,
      "loss": 1.5859,
      "step": 5587
    },
    {
      "epoch": 3.700662251655629,
      "grad_norm": 0.9705381869647484,
      "learning_rate": 0.00014612906206125392,
      "loss": 1.6484,
      "step": 5588
    },
    {
      "epoch": 3.701324503311258,
      "grad_norm": 0.9405711655891572,
      "learning_rate": 0.0001460623369568791,
      "loss": 1.5859,
      "step": 5589
    },
    {
      "epoch": 3.7019867549668874,
      "grad_norm": 1.0313009082623266,
      "learning_rate": 0.00014599561263220663,
      "loss": 1.7266,
      "step": 5590
    },
    {
      "epoch": 3.7026490066225164,
      "grad_norm": 0.9005687496376479,
      "learning_rate": 0.00014592888910044884,
      "loss": 1.4922,
      "step": 5591
    },
    {
      "epoch": 3.703311258278146,
      "grad_norm": 0.97008795498272,
      "learning_rate": 0.0001458621663748177,
      "loss": 1.6328,
      "step": 5592
    },
    {
      "epoch": 3.7039735099337747,
      "grad_norm": 0.9953498354116098,
      "learning_rate": 0.000145795444468525,
      "loss": 1.7422,
      "step": 5593
    },
    {
      "epoch": 3.704635761589404,
      "grad_norm": 0.8908695882478231,
      "learning_rate": 0.00014572872339478252,
      "loss": 1.4297,
      "step": 5594
    },
    {
      "epoch": 3.705298013245033,
      "grad_norm": 0.9761189124975352,
      "learning_rate": 0.00014566200316680177,
      "loss": 1.625,
      "step": 5595
    },
    {
      "epoch": 3.7059602649006624,
      "grad_norm": 0.9398596380609003,
      "learning_rate": 0.00014559528379779412,
      "loss": 1.6406,
      "step": 5596
    },
    {
      "epoch": 3.7066225165562914,
      "grad_norm": 0.9761935981793552,
      "learning_rate": 0.00014552856530097082,
      "loss": 1.6328,
      "step": 5597
    },
    {
      "epoch": 3.7072847682119203,
      "grad_norm": 1.0888091177479453,
      "learning_rate": 0.00014546184768954285,
      "loss": 1.875,
      "step": 5598
    },
    {
      "epoch": 3.7079470198675497,
      "grad_norm": 0.9579716210560884,
      "learning_rate": 0.000145395130976721,
      "loss": 1.7266,
      "step": 5599
    },
    {
      "epoch": 3.7086092715231787,
      "grad_norm": 1.128682628999998,
      "learning_rate": 0.00014532841517571603,
      "loss": 1.9922,
      "step": 5600
    },
    {
      "epoch": 3.709271523178808,
      "grad_norm": 0.8242983068854691,
      "learning_rate": 0.0001452617002997385,
      "loss": 1.3672,
      "step": 5601
    },
    {
      "epoch": 3.709933774834437,
      "grad_norm": 0.9482687352475686,
      "learning_rate": 0.0001451949863619985,
      "loss": 1.7578,
      "step": 5602
    },
    {
      "epoch": 3.7105960264900664,
      "grad_norm": 0.9925017791024802,
      "learning_rate": 0.0001451282733757064,
      "loss": 1.7109,
      "step": 5603
    },
    {
      "epoch": 3.7112582781456953,
      "grad_norm": 0.9561807048584128,
      "learning_rate": 0.00014506156135407196,
      "loss": 1.6562,
      "step": 5604
    },
    {
      "epoch": 3.7119205298013247,
      "grad_norm": 0.9526472922476253,
      "learning_rate": 0.00014499485031030498,
      "loss": 1.5547,
      "step": 5605
    },
    {
      "epoch": 3.7125827814569536,
      "grad_norm": 0.9827693551193025,
      "learning_rate": 0.00014492814025761504,
      "loss": 1.6953,
      "step": 5606
    },
    {
      "epoch": 3.7132450331125826,
      "grad_norm": 1.012678017791769,
      "learning_rate": 0.00014486143120921155,
      "loss": 1.8672,
      "step": 5607
    },
    {
      "epoch": 3.713907284768212,
      "grad_norm": 0.9254644204380189,
      "learning_rate": 0.00014479472317830347,
      "loss": 1.6016,
      "step": 5608
    },
    {
      "epoch": 3.714569536423841,
      "grad_norm": 0.8970856822627125,
      "learning_rate": 0.00014472801617809996,
      "loss": 1.5469,
      "step": 5609
    },
    {
      "epoch": 3.7152317880794703,
      "grad_norm": 0.9475262876547241,
      "learning_rate": 0.00014466131022180967,
      "loss": 1.6797,
      "step": 5610
    },
    {
      "epoch": 3.7158940397350992,
      "grad_norm": 0.8499995844607321,
      "learning_rate": 0.0001445946053226411,
      "loss": 1.4922,
      "step": 5611
    },
    {
      "epoch": 3.7165562913907286,
      "grad_norm": 0.9982125249649828,
      "learning_rate": 0.00014452790149380274,
      "loss": 1.7266,
      "step": 5612
    },
    {
      "epoch": 3.7172185430463576,
      "grad_norm": 0.9367850805374772,
      "learning_rate": 0.0001444611987485026,
      "loss": 1.5625,
      "step": 5613
    },
    {
      "epoch": 3.717880794701987,
      "grad_norm": 1.0543349519410439,
      "learning_rate": 0.00014439449709994858,
      "loss": 1.8828,
      "step": 5614
    },
    {
      "epoch": 3.718543046357616,
      "grad_norm": 1.0254683118674408,
      "learning_rate": 0.00014432779656134838,
      "loss": 1.8984,
      "step": 5615
    },
    {
      "epoch": 3.719205298013245,
      "grad_norm": 0.9233628097734294,
      "learning_rate": 0.00014426109714590956,
      "loss": 1.5938,
      "step": 5616
    },
    {
      "epoch": 3.7198675496688742,
      "grad_norm": 1.0153939976167983,
      "learning_rate": 0.00014419439886683927,
      "loss": 1.7344,
      "step": 5617
    },
    {
      "epoch": 3.720529801324503,
      "grad_norm": 1.0405714573072997,
      "learning_rate": 0.0001441277017373445,
      "loss": 1.7656,
      "step": 5618
    },
    {
      "epoch": 3.7211920529801326,
      "grad_norm": 0.885350648078471,
      "learning_rate": 0.0001440610057706322,
      "loss": 1.5,
      "step": 5619
    },
    {
      "epoch": 3.7218543046357615,
      "grad_norm": 1.016074204471749,
      "learning_rate": 0.0001439943109799088,
      "loss": 1.7891,
      "step": 5620
    },
    {
      "epoch": 3.722516556291391,
      "grad_norm": 1.0248513332024847,
      "learning_rate": 0.00014392761737838067,
      "loss": 1.9219,
      "step": 5621
    },
    {
      "epoch": 3.72317880794702,
      "grad_norm": 0.9449960291192131,
      "learning_rate": 0.00014386092497925394,
      "loss": 1.6328,
      "step": 5622
    },
    {
      "epoch": 3.723841059602649,
      "grad_norm": 0.9471670889436311,
      "learning_rate": 0.00014379423379573447,
      "loss": 1.6406,
      "step": 5623
    },
    {
      "epoch": 3.724503311258278,
      "grad_norm": 0.9386835968082387,
      "learning_rate": 0.0001437275438410278,
      "loss": 1.6641,
      "step": 5624
    },
    {
      "epoch": 3.725165562913907,
      "grad_norm": 0.941368694439891,
      "learning_rate": 0.00014366085512833939,
      "loss": 1.6094,
      "step": 5625
    },
    {
      "epoch": 3.7258278145695365,
      "grad_norm": 0.9960179578596038,
      "learning_rate": 0.00014359416767087434,
      "loss": 1.7578,
      "step": 5626
    },
    {
      "epoch": 3.7264900662251654,
      "grad_norm": 0.9685617809806073,
      "learning_rate": 0.0001435274814818375,
      "loss": 1.5469,
      "step": 5627
    },
    {
      "epoch": 3.727152317880795,
      "grad_norm": 0.9677780454351947,
      "learning_rate": 0.0001434607965744336,
      "loss": 1.6406,
      "step": 5628
    },
    {
      "epoch": 3.7278145695364238,
      "grad_norm": 0.8760744613168246,
      "learning_rate": 0.00014339411296186691,
      "loss": 1.4609,
      "step": 5629
    },
    {
      "epoch": 3.7284768211920527,
      "grad_norm": 1.017822928026569,
      "learning_rate": 0.00014332743065734155,
      "loss": 1.7734,
      "step": 5630
    },
    {
      "epoch": 3.729139072847682,
      "grad_norm": 0.9272345174728477,
      "learning_rate": 0.00014326074967406147,
      "loss": 1.5703,
      "step": 5631
    },
    {
      "epoch": 3.7298013245033115,
      "grad_norm": 0.9894624685316453,
      "learning_rate": 0.0001431940700252302,
      "loss": 1.8438,
      "step": 5632
    },
    {
      "epoch": 3.7304635761589404,
      "grad_norm": 0.9783515355177508,
      "learning_rate": 0.00014312739172405106,
      "loss": 1.5625,
      "step": 5633
    },
    {
      "epoch": 3.7311258278145694,
      "grad_norm": 1.11083942076743,
      "learning_rate": 0.00014306071478372722,
      "loss": 2.0156,
      "step": 5634
    },
    {
      "epoch": 3.7317880794701987,
      "grad_norm": 0.9797835951686907,
      "learning_rate": 0.00014299403921746136,
      "loss": 1.6797,
      "step": 5635
    },
    {
      "epoch": 3.7324503311258277,
      "grad_norm": 0.9416095701021461,
      "learning_rate": 0.00014292736503845606,
      "loss": 1.6797,
      "step": 5636
    },
    {
      "epoch": 3.733112582781457,
      "grad_norm": 0.9940043599716795,
      "learning_rate": 0.00014286069225991356,
      "loss": 1.6641,
      "step": 5637
    },
    {
      "epoch": 3.733774834437086,
      "grad_norm": 0.9826100116067852,
      "learning_rate": 0.0001427940208950359,
      "loss": 1.7266,
      "step": 5638
    },
    {
      "epoch": 3.734437086092715,
      "grad_norm": 0.9451758549611984,
      "learning_rate": 0.0001427273509570246,
      "loss": 1.6719,
      "step": 5639
    },
    {
      "epoch": 3.7350993377483444,
      "grad_norm": 0.9703228818629791,
      "learning_rate": 0.00014266068245908132,
      "loss": 1.6875,
      "step": 5640
    },
    {
      "epoch": 3.7357615894039737,
      "grad_norm": 1.0106032018736029,
      "learning_rate": 0.00014259401541440698,
      "loss": 1.8438,
      "step": 5641
    },
    {
      "epoch": 3.7364238410596027,
      "grad_norm": 1.0404719289854627,
      "learning_rate": 0.0001425273498362025,
      "loss": 1.7578,
      "step": 5642
    },
    {
      "epoch": 3.7370860927152316,
      "grad_norm": 0.998132834745957,
      "learning_rate": 0.00014246068573766846,
      "loss": 1.7656,
      "step": 5643
    },
    {
      "epoch": 3.737748344370861,
      "grad_norm": 1.0336219051160884,
      "learning_rate": 0.00014239402313200508,
      "loss": 1.7344,
      "step": 5644
    },
    {
      "epoch": 3.73841059602649,
      "grad_norm": 0.9353352245417232,
      "learning_rate": 0.00014232736203241225,
      "loss": 1.7266,
      "step": 5645
    },
    {
      "epoch": 3.7390728476821193,
      "grad_norm": 0.9300243616148803,
      "learning_rate": 0.00014226070245208978,
      "loss": 1.5156,
      "step": 5646
    },
    {
      "epoch": 3.7397350993377483,
      "grad_norm": 0.9244521891266527,
      "learning_rate": 0.0001421940444042369,
      "loss": 1.6562,
      "step": 5647
    },
    {
      "epoch": 3.7403973509933772,
      "grad_norm": 0.7956324220321047,
      "learning_rate": 0.00014212738790205274,
      "loss": 1.3516,
      "step": 5648
    },
    {
      "epoch": 3.7410596026490066,
      "grad_norm": 0.9126502973640434,
      "learning_rate": 0.00014206073295873603,
      "loss": 1.5312,
      "step": 5649
    },
    {
      "epoch": 3.741721854304636,
      "grad_norm": 0.9586999478373044,
      "learning_rate": 0.00014199407958748525,
      "loss": 1.6016,
      "step": 5650
    },
    {
      "epoch": 3.742384105960265,
      "grad_norm": 0.9777055959622017,
      "learning_rate": 0.00014192742780149843,
      "loss": 1.6797,
      "step": 5651
    },
    {
      "epoch": 3.743046357615894,
      "grad_norm": 0.9853199011330255,
      "learning_rate": 0.00014186077761397348,
      "loss": 1.6719,
      "step": 5652
    },
    {
      "epoch": 3.7437086092715233,
      "grad_norm": 0.9662368713866214,
      "learning_rate": 0.0001417941290381079,
      "loss": 1.6484,
      "step": 5653
    },
    {
      "epoch": 3.744370860927152,
      "grad_norm": 1.0773504417427564,
      "learning_rate": 0.0001417274820870988,
      "loss": 1.9141,
      "step": 5654
    },
    {
      "epoch": 3.7450331125827816,
      "grad_norm": 0.9815264205569884,
      "learning_rate": 0.00014166083677414316,
      "loss": 1.6484,
      "step": 5655
    },
    {
      "epoch": 3.7456953642384105,
      "grad_norm": 0.9972954942664096,
      "learning_rate": 0.0001415941931124374,
      "loss": 1.7266,
      "step": 5656
    },
    {
      "epoch": 3.7463576158940395,
      "grad_norm": 0.9799990803632999,
      "learning_rate": 0.00014152755111517776,
      "loss": 1.6797,
      "step": 5657
    },
    {
      "epoch": 3.747019867549669,
      "grad_norm": 1.0469392724177857,
      "learning_rate": 0.00014146091079556014,
      "loss": 1.8672,
      "step": 5658
    },
    {
      "epoch": 3.7476821192052983,
      "grad_norm": 1.0194287665751198,
      "learning_rate": 0.00014139427216678012,
      "loss": 1.8203,
      "step": 5659
    },
    {
      "epoch": 3.748344370860927,
      "grad_norm": 0.9181467112141901,
      "learning_rate": 0.0001413276352420328,
      "loss": 1.6719,
      "step": 5660
    },
    {
      "epoch": 3.749006622516556,
      "grad_norm": 0.949606732470327,
      "learning_rate": 0.0001412610000345132,
      "loss": 1.6875,
      "step": 5661
    },
    {
      "epoch": 3.7496688741721855,
      "grad_norm": 0.9410252988020519,
      "learning_rate": 0.00014119436655741575,
      "loss": 1.6094,
      "step": 5662
    },
    {
      "epoch": 3.7503311258278145,
      "grad_norm": 0.9825457443446158,
      "learning_rate": 0.00014112773482393467,
      "loss": 1.75,
      "step": 5663
    },
    {
      "epoch": 3.750993377483444,
      "grad_norm": 0.9440412432588978,
      "learning_rate": 0.00014106110484726382,
      "loss": 1.8125,
      "step": 5664
    },
    {
      "epoch": 3.751655629139073,
      "grad_norm": 0.863725226358742,
      "learning_rate": 0.00014099447664059673,
      "loss": 1.4688,
      "step": 5665
    },
    {
      "epoch": 3.7523178807947017,
      "grad_norm": 1.0028840702324409,
      "learning_rate": 0.0001409278502171264,
      "loss": 1.8047,
      "step": 5666
    },
    {
      "epoch": 3.752980132450331,
      "grad_norm": 1.054399273564284,
      "learning_rate": 0.00014086122559004585,
      "loss": 1.7188,
      "step": 5667
    },
    {
      "epoch": 3.7536423841059605,
      "grad_norm": 0.8163596490350989,
      "learning_rate": 0.00014079460277254736,
      "loss": 1.3672,
      "step": 5668
    },
    {
      "epoch": 3.7543046357615895,
      "grad_norm": 1.0618980530904467,
      "learning_rate": 0.00014072798177782302,
      "loss": 1.9688,
      "step": 5669
    },
    {
      "epoch": 3.7549668874172184,
      "grad_norm": 0.8952504387012096,
      "learning_rate": 0.00014066136261906465,
      "loss": 1.6094,
      "step": 5670
    },
    {
      "epoch": 3.755629139072848,
      "grad_norm": 0.9223267227442431,
      "learning_rate": 0.00014059474530946355,
      "loss": 1.5703,
      "step": 5671
    },
    {
      "epoch": 3.7562913907284767,
      "grad_norm": 1.0095120967015605,
      "learning_rate": 0.00014052812986221065,
      "loss": 1.7734,
      "step": 5672
    },
    {
      "epoch": 3.756953642384106,
      "grad_norm": 1.0827830327076555,
      "learning_rate": 0.00014046151629049667,
      "loss": 2.0625,
      "step": 5673
    },
    {
      "epoch": 3.757615894039735,
      "grad_norm": 0.941133340300096,
      "learning_rate": 0.0001403949046075118,
      "loss": 1.5781,
      "step": 5674
    },
    {
      "epoch": 3.758278145695364,
      "grad_norm": 0.9630145732901069,
      "learning_rate": 0.00014032829482644597,
      "loss": 1.6562,
      "step": 5675
    },
    {
      "epoch": 3.7589403973509934,
      "grad_norm": 1.0327799604157473,
      "learning_rate": 0.00014026168696048858,
      "loss": 1.8047,
      "step": 5676
    },
    {
      "epoch": 3.7596026490066228,
      "grad_norm": 0.9055650407301284,
      "learning_rate": 0.0001401950810228289,
      "loss": 1.5156,
      "step": 5677
    },
    {
      "epoch": 3.7602649006622517,
      "grad_norm": 1.1233065021637858,
      "learning_rate": 0.00014012847702665553,
      "loss": 1.9531,
      "step": 5678
    },
    {
      "epoch": 3.7609271523178807,
      "grad_norm": 1.0491289542925442,
      "learning_rate": 0.0001400618749851569,
      "loss": 1.7891,
      "step": 5679
    },
    {
      "epoch": 3.76158940397351,
      "grad_norm": 0.961104917441985,
      "learning_rate": 0.00013999527491152095,
      "loss": 1.6875,
      "step": 5680
    },
    {
      "epoch": 3.762251655629139,
      "grad_norm": 1.0077368971253076,
      "learning_rate": 0.0001399286768189353,
      "loss": 1.7891,
      "step": 5681
    },
    {
      "epoch": 3.7629139072847684,
      "grad_norm": 1.0042633887930412,
      "learning_rate": 0.00013986208072058707,
      "loss": 1.7734,
      "step": 5682
    },
    {
      "epoch": 3.7635761589403973,
      "grad_norm": 0.8589847078877201,
      "learning_rate": 0.0001397954866296631,
      "loss": 1.4141,
      "step": 5683
    },
    {
      "epoch": 3.7642384105960263,
      "grad_norm": 1.0018422344688633,
      "learning_rate": 0.00013972889455934976,
      "loss": 1.5625,
      "step": 5684
    },
    {
      "epoch": 3.7649006622516556,
      "grad_norm": 0.939286224245937,
      "learning_rate": 0.00013966230452283304,
      "loss": 1.6094,
      "step": 5685
    },
    {
      "epoch": 3.765562913907285,
      "grad_norm": 0.892072243592512,
      "learning_rate": 0.00013959571653329855,
      "loss": 1.6328,
      "step": 5686
    },
    {
      "epoch": 3.766225165562914,
      "grad_norm": 0.9856769682634343,
      "learning_rate": 0.0001395291306039315,
      "loss": 1.6641,
      "step": 5687
    },
    {
      "epoch": 3.766887417218543,
      "grad_norm": 0.9568813037262116,
      "learning_rate": 0.00013946254674791658,
      "loss": 1.6094,
      "step": 5688
    },
    {
      "epoch": 3.7675496688741723,
      "grad_norm": 0.9896710300184778,
      "learning_rate": 0.00013939596497843825,
      "loss": 1.6406,
      "step": 5689
    },
    {
      "epoch": 3.7682119205298013,
      "grad_norm": 0.9445385746993286,
      "learning_rate": 0.0001393293853086804,
      "loss": 1.625,
      "step": 5690
    },
    {
      "epoch": 3.7688741721854306,
      "grad_norm": 1.0133760765212727,
      "learning_rate": 0.0001392628077518266,
      "loss": 1.8359,
      "step": 5691
    },
    {
      "epoch": 3.7695364238410596,
      "grad_norm": 1.0071765208390107,
      "learning_rate": 0.00013919623232106,
      "loss": 1.7344,
      "step": 5692
    },
    {
      "epoch": 3.7701986754966885,
      "grad_norm": 0.9703738590723116,
      "learning_rate": 0.00013912965902956325,
      "loss": 1.625,
      "step": 5693
    },
    {
      "epoch": 3.770860927152318,
      "grad_norm": 0.9923639914237589,
      "learning_rate": 0.0001390630878905186,
      "loss": 1.7344,
      "step": 5694
    },
    {
      "epoch": 3.7715231788079473,
      "grad_norm": 1.0083452971202145,
      "learning_rate": 0.00013899651891710797,
      "loss": 1.7578,
      "step": 5695
    },
    {
      "epoch": 3.7721854304635762,
      "grad_norm": 1.0150064655603623,
      "learning_rate": 0.00013892995212251276,
      "loss": 1.7266,
      "step": 5696
    },
    {
      "epoch": 3.772847682119205,
      "grad_norm": 0.9279819379467651,
      "learning_rate": 0.00013886338751991387,
      "loss": 1.6719,
      "step": 5697
    },
    {
      "epoch": 3.7735099337748346,
      "grad_norm": 1.056865820097976,
      "learning_rate": 0.00013879682512249198,
      "loss": 1.6797,
      "step": 5698
    },
    {
      "epoch": 3.7741721854304635,
      "grad_norm": 0.9444921512427924,
      "learning_rate": 0.00013873026494342716,
      "loss": 1.625,
      "step": 5699
    },
    {
      "epoch": 3.774834437086093,
      "grad_norm": 1.0709212671201553,
      "learning_rate": 0.00013866370699589905,
      "loss": 1.7656,
      "step": 5700
    },
    {
      "epoch": 3.775496688741722,
      "grad_norm": 0.9082300104270435,
      "learning_rate": 0.00013859715129308696,
      "loss": 1.4922,
      "step": 5701
    },
    {
      "epoch": 3.776158940397351,
      "grad_norm": 0.9542834236978023,
      "learning_rate": 0.00013853059784816967,
      "loss": 1.5859,
      "step": 5702
    },
    {
      "epoch": 3.77682119205298,
      "grad_norm": 1.0728101835737411,
      "learning_rate": 0.00013846404667432538,
      "loss": 1.8359,
      "step": 5703
    },
    {
      "epoch": 3.777483443708609,
      "grad_norm": 1.0591772128029,
      "learning_rate": 0.00013839749778473223,
      "loss": 1.6641,
      "step": 5704
    },
    {
      "epoch": 3.7781456953642385,
      "grad_norm": 1.0358660860675137,
      "learning_rate": 0.00013833095119256749,
      "loss": 1.75,
      "step": 5705
    },
    {
      "epoch": 3.7788079470198674,
      "grad_norm": 1.0325599126047047,
      "learning_rate": 0.00013826440691100817,
      "loss": 1.6797,
      "step": 5706
    },
    {
      "epoch": 3.779470198675497,
      "grad_norm": 1.0399512326645468,
      "learning_rate": 0.00013819786495323084,
      "loss": 1.6172,
      "step": 5707
    },
    {
      "epoch": 3.7801324503311258,
      "grad_norm": 0.9781114398453498,
      "learning_rate": 0.00013813132533241161,
      "loss": 1.6328,
      "step": 5708
    },
    {
      "epoch": 3.780794701986755,
      "grad_norm": 0.9855326889046522,
      "learning_rate": 0.00013806478806172595,
      "loss": 1.5938,
      "step": 5709
    },
    {
      "epoch": 3.781456953642384,
      "grad_norm": 1.0201773486335914,
      "learning_rate": 0.00013799825315434914,
      "loss": 1.7344,
      "step": 5710
    },
    {
      "epoch": 3.782119205298013,
      "grad_norm": 0.9371631054935312,
      "learning_rate": 0.00013793172062345578,
      "loss": 1.6406,
      "step": 5711
    },
    {
      "epoch": 3.7827814569536424,
      "grad_norm": 0.9428023034748756,
      "learning_rate": 0.0001378651904822201,
      "loss": 1.6484,
      "step": 5712
    },
    {
      "epoch": 3.7834437086092714,
      "grad_norm": 0.9562675729634152,
      "learning_rate": 0.0001377986627438158,
      "loss": 1.5547,
      "step": 5713
    },
    {
      "epoch": 3.7841059602649008,
      "grad_norm": 0.8808872507289454,
      "learning_rate": 0.00013773213742141624,
      "loss": 1.4297,
      "step": 5714
    },
    {
      "epoch": 3.7847682119205297,
      "grad_norm": 1.0909838555229883,
      "learning_rate": 0.00013766561452819404,
      "loss": 1.9531,
      "step": 5715
    },
    {
      "epoch": 3.785430463576159,
      "grad_norm": 1.1484771048173754,
      "learning_rate": 0.00013759909407732158,
      "loss": 2.2969,
      "step": 5716
    },
    {
      "epoch": 3.786092715231788,
      "grad_norm": 1.0203841321488,
      "learning_rate": 0.0001375325760819707,
      "loss": 1.6953,
      "step": 5717
    },
    {
      "epoch": 3.7867549668874174,
      "grad_norm": 0.9628254838927495,
      "learning_rate": 0.00013746606055531265,
      "loss": 1.6484,
      "step": 5718
    },
    {
      "epoch": 3.7874172185430464,
      "grad_norm": 1.0106842071481297,
      "learning_rate": 0.00013739954751051836,
      "loss": 1.7422,
      "step": 5719
    },
    {
      "epoch": 3.7880794701986753,
      "grad_norm": 0.9536315792364277,
      "learning_rate": 0.00013733303696075808,
      "loss": 1.6406,
      "step": 5720
    },
    {
      "epoch": 3.7887417218543047,
      "grad_norm": 0.9162744150712108,
      "learning_rate": 0.00013726652891920171,
      "loss": 1.6094,
      "step": 5721
    },
    {
      "epoch": 3.7894039735099336,
      "grad_norm": 0.9748402605860955,
      "learning_rate": 0.00013720002339901862,
      "loss": 1.8281,
      "step": 5722
    },
    {
      "epoch": 3.790066225165563,
      "grad_norm": 0.7916566815676708,
      "learning_rate": 0.00013713352041337765,
      "loss": 1.2656,
      "step": 5723
    },
    {
      "epoch": 3.790728476821192,
      "grad_norm": 0.9344195977762946,
      "learning_rate": 0.00013706701997544704,
      "loss": 1.7109,
      "step": 5724
    },
    {
      "epoch": 3.7913907284768213,
      "grad_norm": 0.8938308065855307,
      "learning_rate": 0.0001370005220983949,
      "loss": 1.6016,
      "step": 5725
    },
    {
      "epoch": 3.7920529801324503,
      "grad_norm": 0.881142606188734,
      "learning_rate": 0.00013693402679538835,
      "loss": 1.6172,
      "step": 5726
    },
    {
      "epoch": 3.7927152317880797,
      "grad_norm": 1.017744141979622,
      "learning_rate": 0.00013686753407959426,
      "loss": 1.7812,
      "step": 5727
    },
    {
      "epoch": 3.7933774834437086,
      "grad_norm": 0.9836010370691388,
      "learning_rate": 0.000136801043964179,
      "loss": 1.4844,
      "step": 5728
    },
    {
      "epoch": 3.7940397350993376,
      "grad_norm": 1.123892077746329,
      "learning_rate": 0.00013673455646230842,
      "loss": 2.0156,
      "step": 5729
    },
    {
      "epoch": 3.794701986754967,
      "grad_norm": 0.9526271989357165,
      "learning_rate": 0.0001366680715871477,
      "loss": 1.6094,
      "step": 5730
    },
    {
      "epoch": 3.795364238410596,
      "grad_norm": 0.8911064896091004,
      "learning_rate": 0.0001366015893518616,
      "loss": 1.4766,
      "step": 5731
    },
    {
      "epoch": 3.7960264900662253,
      "grad_norm": 1.023839038526304,
      "learning_rate": 0.00013653510976961448,
      "loss": 1.7109,
      "step": 5732
    },
    {
      "epoch": 3.796688741721854,
      "grad_norm": 1.0803501312572128,
      "learning_rate": 0.00013646863285357003,
      "loss": 1.7578,
      "step": 5733
    },
    {
      "epoch": 3.7973509933774836,
      "grad_norm": 0.9897947708568728,
      "learning_rate": 0.00013640215861689128,
      "loss": 1.8828,
      "step": 5734
    },
    {
      "epoch": 3.7980132450331126,
      "grad_norm": 1.0981155807569185,
      "learning_rate": 0.00013633568707274114,
      "loss": 1.9453,
      "step": 5735
    },
    {
      "epoch": 3.798675496688742,
      "grad_norm": 0.9383758702936661,
      "learning_rate": 0.00013626921823428156,
      "loss": 1.5625,
      "step": 5736
    },
    {
      "epoch": 3.799337748344371,
      "grad_norm": 0.95693712344548,
      "learning_rate": 0.00013620275211467415,
      "loss": 1.6875,
      "step": 5737
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.971831467738029,
      "learning_rate": 0.00013613628872708006,
      "loss": 1.7656,
      "step": 5738
    },
    {
      "epoch": 3.800662251655629,
      "grad_norm": 0.9231508566210885,
      "learning_rate": 0.00013606982808465976,
      "loss": 1.6328,
      "step": 5739
    },
    {
      "epoch": 3.801324503311258,
      "grad_norm": 0.9649222469752964,
      "learning_rate": 0.00013600337020057312,
      "loss": 1.6094,
      "step": 5740
    },
    {
      "epoch": 3.8019867549668875,
      "grad_norm": 0.8832598447323994,
      "learning_rate": 0.00013593691508797972,
      "loss": 1.5,
      "step": 5741
    },
    {
      "epoch": 3.8026490066225165,
      "grad_norm": 1.0198826404788734,
      "learning_rate": 0.00013587046276003836,
      "loss": 1.8359,
      "step": 5742
    },
    {
      "epoch": 3.8033112582781454,
      "grad_norm": 0.9476413823835215,
      "learning_rate": 0.00013580401322990733,
      "loss": 1.5859,
      "step": 5743
    },
    {
      "epoch": 3.803973509933775,
      "grad_norm": 0.9483183820247099,
      "learning_rate": 0.00013573756651074443,
      "loss": 1.6484,
      "step": 5744
    },
    {
      "epoch": 3.804635761589404,
      "grad_norm": 0.8537848353306242,
      "learning_rate": 0.00013567112261570692,
      "loss": 1.4062,
      "step": 5745
    },
    {
      "epoch": 3.805298013245033,
      "grad_norm": 0.9334984580580629,
      "learning_rate": 0.00013560468155795135,
      "loss": 1.6328,
      "step": 5746
    },
    {
      "epoch": 3.805960264900662,
      "grad_norm": 0.9610968453009481,
      "learning_rate": 0.0001355382433506339,
      "loss": 1.5938,
      "step": 5747
    },
    {
      "epoch": 3.8066225165562915,
      "grad_norm": 0.91203956856607,
      "learning_rate": 0.00013547180800691007,
      "loss": 1.5312,
      "step": 5748
    },
    {
      "epoch": 3.8072847682119204,
      "grad_norm": 0.9830084941334887,
      "learning_rate": 0.00013540537553993477,
      "loss": 1.6484,
      "step": 5749
    },
    {
      "epoch": 3.80794701986755,
      "grad_norm": 1.0499276764915741,
      "learning_rate": 0.00013533894596286252,
      "loss": 1.8438,
      "step": 5750
    },
    {
      "epoch": 3.8086092715231787,
      "grad_norm": 0.8836277757580524,
      "learning_rate": 0.000135272519288847,
      "loss": 1.4531,
      "step": 5751
    },
    {
      "epoch": 3.8092715231788077,
      "grad_norm": 1.0752627978238134,
      "learning_rate": 0.0001352060955310415,
      "loss": 1.7969,
      "step": 5752
    },
    {
      "epoch": 3.809933774834437,
      "grad_norm": 1.0158898445699451,
      "learning_rate": 0.0001351396747025987,
      "loss": 1.9531,
      "step": 5753
    },
    {
      "epoch": 3.8105960264900665,
      "grad_norm": 1.0127802233900909,
      "learning_rate": 0.00013507325681667067,
      "loss": 1.7734,
      "step": 5754
    },
    {
      "epoch": 3.8112582781456954,
      "grad_norm": 1.0480319364823467,
      "learning_rate": 0.0001350068418864089,
      "loss": 1.8516,
      "step": 5755
    },
    {
      "epoch": 3.8119205298013243,
      "grad_norm": 0.9138218201489762,
      "learning_rate": 0.00013494042992496442,
      "loss": 1.5312,
      "step": 5756
    },
    {
      "epoch": 3.8125827814569537,
      "grad_norm": 0.9301062370901482,
      "learning_rate": 0.0001348740209454874,
      "loss": 1.5078,
      "step": 5757
    },
    {
      "epoch": 3.8132450331125827,
      "grad_norm": 1.0198606091774272,
      "learning_rate": 0.00013480761496112762,
      "loss": 1.7656,
      "step": 5758
    },
    {
      "epoch": 3.813907284768212,
      "grad_norm": 0.8557286299139155,
      "learning_rate": 0.00013474121198503428,
      "loss": 1.4219,
      "step": 5759
    },
    {
      "epoch": 3.814569536423841,
      "grad_norm": 0.9573858866747225,
      "learning_rate": 0.00013467481203035595,
      "loss": 1.6406,
      "step": 5760
    },
    {
      "epoch": 3.81523178807947,
      "grad_norm": 0.9718778552723135,
      "learning_rate": 0.0001346084151102404,
      "loss": 1.5781,
      "step": 5761
    },
    {
      "epoch": 3.8158940397350993,
      "grad_norm": 0.9862821023605535,
      "learning_rate": 0.00013454202123783528,
      "loss": 1.6875,
      "step": 5762
    },
    {
      "epoch": 3.8165562913907287,
      "grad_norm": 0.9334446754601656,
      "learning_rate": 0.00013447563042628708,
      "loss": 1.6172,
      "step": 5763
    },
    {
      "epoch": 3.8172185430463577,
      "grad_norm": 0.974817216879401,
      "learning_rate": 0.00013440924268874206,
      "loss": 1.6562,
      "step": 5764
    },
    {
      "epoch": 3.8178807947019866,
      "grad_norm": 1.0578977318239806,
      "learning_rate": 0.00013434285803834572,
      "loss": 1.7188,
      "step": 5765
    },
    {
      "epoch": 3.818543046357616,
      "grad_norm": 1.0485375514928774,
      "learning_rate": 0.00013427647648824303,
      "loss": 1.7812,
      "step": 5766
    },
    {
      "epoch": 3.819205298013245,
      "grad_norm": 0.9300662014167557,
      "learning_rate": 0.00013421009805157817,
      "loss": 1.5312,
      "step": 5767
    },
    {
      "epoch": 3.8198675496688743,
      "grad_norm": 1.0821623397165827,
      "learning_rate": 0.00013414372274149502,
      "loss": 1.8906,
      "step": 5768
    },
    {
      "epoch": 3.8205298013245033,
      "grad_norm": 1.0399652774971395,
      "learning_rate": 0.00013407735057113652,
      "loss": 1.8672,
      "step": 5769
    },
    {
      "epoch": 3.821192052980132,
      "grad_norm": 0.8122532344545084,
      "learning_rate": 0.00013401098155364514,
      "loss": 1.3672,
      "step": 5770
    },
    {
      "epoch": 3.8218543046357616,
      "grad_norm": 1.0526021554578184,
      "learning_rate": 0.00013394461570216272,
      "loss": 1.8516,
      "step": 5771
    },
    {
      "epoch": 3.822516556291391,
      "grad_norm": 0.9447389305926531,
      "learning_rate": 0.0001338782530298305,
      "loss": 1.7266,
      "step": 5772
    },
    {
      "epoch": 3.82317880794702,
      "grad_norm": 1.0594615717732863,
      "learning_rate": 0.00013381189354978897,
      "loss": 1.7969,
      "step": 5773
    },
    {
      "epoch": 3.823841059602649,
      "grad_norm": 1.00006804971187,
      "learning_rate": 0.0001337455372751781,
      "loss": 1.5859,
      "step": 5774
    },
    {
      "epoch": 3.8245033112582782,
      "grad_norm": 0.7803282827523832,
      "learning_rate": 0.00013367918421913722,
      "loss": 1.3359,
      "step": 5775
    },
    {
      "epoch": 3.825165562913907,
      "grad_norm": 0.9063099270267305,
      "learning_rate": 0.00013361283439480495,
      "loss": 1.6484,
      "step": 5776
    },
    {
      "epoch": 3.8258278145695366,
      "grad_norm": 0.8350609777442217,
      "learning_rate": 0.0001335464878153194,
      "loss": 1.4766,
      "step": 5777
    },
    {
      "epoch": 3.8264900662251655,
      "grad_norm": 1.0030996858016865,
      "learning_rate": 0.00013348014449381787,
      "loss": 1.8125,
      "step": 5778
    },
    {
      "epoch": 3.8271523178807945,
      "grad_norm": 0.9710877033752598,
      "learning_rate": 0.0001334138044434371,
      "loss": 1.7656,
      "step": 5779
    },
    {
      "epoch": 3.827814569536424,
      "grad_norm": 0.9509175288003212,
      "learning_rate": 0.00013334746767731324,
      "loss": 1.7266,
      "step": 5780
    },
    {
      "epoch": 3.8284768211920532,
      "grad_norm": 0.9559609777899778,
      "learning_rate": 0.00013328113420858168,
      "loss": 1.6172,
      "step": 5781
    },
    {
      "epoch": 3.829139072847682,
      "grad_norm": 0.9405955291328864,
      "learning_rate": 0.00013321480405037718,
      "loss": 1.5781,
      "step": 5782
    },
    {
      "epoch": 3.829801324503311,
      "grad_norm": 0.8743275146132014,
      "learning_rate": 0.000133148477215834,
      "loss": 1.6016,
      "step": 5783
    },
    {
      "epoch": 3.8304635761589405,
      "grad_norm": 0.8799514484085708,
      "learning_rate": 0.00013308215371808549,
      "loss": 1.6328,
      "step": 5784
    },
    {
      "epoch": 3.8311258278145695,
      "grad_norm": 1.016599624345667,
      "learning_rate": 0.00013301583357026444,
      "loss": 1.8828,
      "step": 5785
    },
    {
      "epoch": 3.831788079470199,
      "grad_norm": 1.048670063429407,
      "learning_rate": 0.0001329495167855031,
      "loss": 1.8047,
      "step": 5786
    },
    {
      "epoch": 3.832450331125828,
      "grad_norm": 0.9489994891076399,
      "learning_rate": 0.00013288320337693298,
      "loss": 1.5547,
      "step": 5787
    },
    {
      "epoch": 3.8331125827814567,
      "grad_norm": 1.100992065568837,
      "learning_rate": 0.0001328168933576848,
      "loss": 1.8672,
      "step": 5788
    },
    {
      "epoch": 3.833774834437086,
      "grad_norm": 0.9956615078108998,
      "learning_rate": 0.00013275058674088868,
      "loss": 1.75,
      "step": 5789
    },
    {
      "epoch": 3.8344370860927155,
      "grad_norm": 1.0008777241587552,
      "learning_rate": 0.00013268428353967415,
      "loss": 1.5625,
      "step": 5790
    },
    {
      "epoch": 3.8350993377483444,
      "grad_norm": 0.9444321188884052,
      "learning_rate": 0.00013261798376717007,
      "loss": 1.6641,
      "step": 5791
    },
    {
      "epoch": 3.8357615894039734,
      "grad_norm": 1.0468922444149253,
      "learning_rate": 0.00013255168743650438,
      "loss": 1.6953,
      "step": 5792
    },
    {
      "epoch": 3.8364238410596028,
      "grad_norm": 0.889149022846279,
      "learning_rate": 0.00013248539456080469,
      "loss": 1.4453,
      "step": 5793
    },
    {
      "epoch": 3.8370860927152317,
      "grad_norm": 1.0077299520477745,
      "learning_rate": 0.00013241910515319767,
      "loss": 1.7109,
      "step": 5794
    },
    {
      "epoch": 3.837748344370861,
      "grad_norm": 1.0191500892496874,
      "learning_rate": 0.00013235281922680932,
      "loss": 1.7344,
      "step": 5795
    },
    {
      "epoch": 3.83841059602649,
      "grad_norm": 1.0595608007313757,
      "learning_rate": 0.00013228653679476515,
      "loss": 1.8828,
      "step": 5796
    },
    {
      "epoch": 3.839072847682119,
      "grad_norm": 1.0129762218529146,
      "learning_rate": 0.00013222025787018978,
      "loss": 1.7031,
      "step": 5797
    },
    {
      "epoch": 3.8397350993377484,
      "grad_norm": 0.8719262290535541,
      "learning_rate": 0.0001321539824662071,
      "loss": 1.5156,
      "step": 5798
    },
    {
      "epoch": 3.8403973509933778,
      "grad_norm": 0.8307836141053977,
      "learning_rate": 0.0001320877105959406,
      "loss": 1.3203,
      "step": 5799
    },
    {
      "epoch": 3.8410596026490067,
      "grad_norm": 0.9715188005003972,
      "learning_rate": 0.00013202144227251274,
      "loss": 1.6797,
      "step": 5800
    },
    {
      "epoch": 3.8417218543046356,
      "grad_norm": 1.0725347373714311,
      "learning_rate": 0.00013195517750904536,
      "loss": 1.7656,
      "step": 5801
    },
    {
      "epoch": 3.842384105960265,
      "grad_norm": 1.0191276107829323,
      "learning_rate": 0.00013188891631865977,
      "loss": 1.7969,
      "step": 5802
    },
    {
      "epoch": 3.843046357615894,
      "grad_norm": 0.8977150611532076,
      "learning_rate": 0.00013182265871447638,
      "loss": 1.3672,
      "step": 5803
    },
    {
      "epoch": 3.8437086092715234,
      "grad_norm": 0.9420445304668542,
      "learning_rate": 0.00013175640470961493,
      "loss": 1.5625,
      "step": 5804
    },
    {
      "epoch": 3.8443708609271523,
      "grad_norm": 0.9906768205028295,
      "learning_rate": 0.00013169015431719453,
      "loss": 1.6641,
      "step": 5805
    },
    {
      "epoch": 3.8450331125827812,
      "grad_norm": 0.9318403938997516,
      "learning_rate": 0.00013162390755033348,
      "loss": 1.5312,
      "step": 5806
    },
    {
      "epoch": 3.8456953642384106,
      "grad_norm": 0.8727805657192615,
      "learning_rate": 0.00013155766442214938,
      "loss": 1.5156,
      "step": 5807
    },
    {
      "epoch": 3.8463576158940396,
      "grad_norm": 1.0982128282010468,
      "learning_rate": 0.0001314914249457592,
      "loss": 1.9297,
      "step": 5808
    },
    {
      "epoch": 3.847019867549669,
      "grad_norm": 1.0576031700647566,
      "learning_rate": 0.0001314251891342791,
      "loss": 1.8516,
      "step": 5809
    },
    {
      "epoch": 3.847682119205298,
      "grad_norm": 0.9628591444325123,
      "learning_rate": 0.00013135895700082444,
      "loss": 1.6641,
      "step": 5810
    },
    {
      "epoch": 3.8483443708609273,
      "grad_norm": 0.9244436574679447,
      "learning_rate": 0.00013129272855851005,
      "loss": 1.5703,
      "step": 5811
    },
    {
      "epoch": 3.8490066225165562,
      "grad_norm": 0.976728360030882,
      "learning_rate": 0.00013122650382044984,
      "loss": 1.5625,
      "step": 5812
    },
    {
      "epoch": 3.8496688741721856,
      "grad_norm": 0.9952981797860961,
      "learning_rate": 0.00013116028279975712,
      "loss": 1.7188,
      "step": 5813
    },
    {
      "epoch": 3.8503311258278146,
      "grad_norm": 0.9721000097141832,
      "learning_rate": 0.00013109406550954446,
      "loss": 1.4453,
      "step": 5814
    },
    {
      "epoch": 3.8509933774834435,
      "grad_norm": 1.014108730709295,
      "learning_rate": 0.00013102785196292353,
      "loss": 1.7031,
      "step": 5815
    },
    {
      "epoch": 3.851655629139073,
      "grad_norm": 0.9593041523752063,
      "learning_rate": 0.00013096164217300536,
      "loss": 1.5781,
      "step": 5816
    },
    {
      "epoch": 3.852317880794702,
      "grad_norm": 1.0542042708130113,
      "learning_rate": 0.00013089543615290038,
      "loss": 1.9141,
      "step": 5817
    },
    {
      "epoch": 3.852980132450331,
      "grad_norm": 0.9557246524231798,
      "learning_rate": 0.0001308292339157181,
      "loss": 1.5469,
      "step": 5818
    },
    {
      "epoch": 3.85364238410596,
      "grad_norm": 0.9282879936423754,
      "learning_rate": 0.00013076303547456715,
      "loss": 1.5391,
      "step": 5819
    },
    {
      "epoch": 3.8543046357615895,
      "grad_norm": 1.0935065347961586,
      "learning_rate": 0.00013069684084255584,
      "loss": 1.7578,
      "step": 5820
    },
    {
      "epoch": 3.8549668874172185,
      "grad_norm": 0.9754163101470362,
      "learning_rate": 0.00013063065003279136,
      "loss": 1.7031,
      "step": 5821
    },
    {
      "epoch": 3.855629139072848,
      "grad_norm": 0.9676640778149372,
      "learning_rate": 0.00013056446305838014,
      "loss": 1.5625,
      "step": 5822
    },
    {
      "epoch": 3.856291390728477,
      "grad_norm": 1.0008248604068417,
      "learning_rate": 0.00013049827993242812,
      "loss": 1.7578,
      "step": 5823
    },
    {
      "epoch": 3.8569536423841058,
      "grad_norm": 0.9306254484276619,
      "learning_rate": 0.00013043210066804027,
      "loss": 1.4688,
      "step": 5824
    },
    {
      "epoch": 3.857615894039735,
      "grad_norm": 0.8605743067730766,
      "learning_rate": 0.00013036592527832075,
      "loss": 1.4453,
      "step": 5825
    },
    {
      "epoch": 3.858278145695364,
      "grad_norm": 1.0383352838103712,
      "learning_rate": 0.0001302997537763732,
      "loss": 1.6172,
      "step": 5826
    },
    {
      "epoch": 3.8589403973509935,
      "grad_norm": 1.0587724770920803,
      "learning_rate": 0.00013023358617530025,
      "loss": 1.8359,
      "step": 5827
    },
    {
      "epoch": 3.8596026490066224,
      "grad_norm": 0.9754182362862439,
      "learning_rate": 0.00013016742248820378,
      "loss": 1.5625,
      "step": 5828
    },
    {
      "epoch": 3.860264900662252,
      "grad_norm": 0.9727320257296663,
      "learning_rate": 0.0001301012627281851,
      "loss": 1.6719,
      "step": 5829
    },
    {
      "epoch": 3.8609271523178808,
      "grad_norm": 1.0423407276771637,
      "learning_rate": 0.00013003510690834453,
      "loss": 1.7734,
      "step": 5830
    },
    {
      "epoch": 3.86158940397351,
      "grad_norm": 1.0002689746998945,
      "learning_rate": 0.00012996895504178163,
      "loss": 1.7266,
      "step": 5831
    },
    {
      "epoch": 3.862251655629139,
      "grad_norm": 0.947248356988845,
      "learning_rate": 0.00012990280714159533,
      "loss": 1.7109,
      "step": 5832
    },
    {
      "epoch": 3.862913907284768,
      "grad_norm": 0.9193020070269123,
      "learning_rate": 0.0001298366632208836,
      "loss": 1.6484,
      "step": 5833
    },
    {
      "epoch": 3.8635761589403974,
      "grad_norm": 0.929468572661472,
      "learning_rate": 0.00012977052329274372,
      "loss": 1.5469,
      "step": 5834
    },
    {
      "epoch": 3.8642384105960264,
      "grad_norm": 1.0230020514441112,
      "learning_rate": 0.00012970438737027216,
      "loss": 1.7656,
      "step": 5835
    },
    {
      "epoch": 3.8649006622516557,
      "grad_norm": 0.9662214349565599,
      "learning_rate": 0.00012963825546656463,
      "loss": 1.7656,
      "step": 5836
    },
    {
      "epoch": 3.8655629139072847,
      "grad_norm": 1.0294159562360312,
      "learning_rate": 0.00012957212759471585,
      "loss": 1.8203,
      "step": 5837
    },
    {
      "epoch": 3.866225165562914,
      "grad_norm": 0.9728227647816327,
      "learning_rate": 0.00012950600376782008,
      "loss": 1.6562,
      "step": 5838
    },
    {
      "epoch": 3.866887417218543,
      "grad_norm": 0.8496198782771706,
      "learning_rate": 0.0001294398839989705,
      "loss": 1.4922,
      "step": 5839
    },
    {
      "epoch": 3.8675496688741724,
      "grad_norm": 0.9990526014455295,
      "learning_rate": 0.00012937376830125956,
      "loss": 1.8594,
      "step": 5840
    },
    {
      "epoch": 3.8682119205298013,
      "grad_norm": 1.0005052144183477,
      "learning_rate": 0.00012930765668777905,
      "loss": 1.7656,
      "step": 5841
    },
    {
      "epoch": 3.8688741721854303,
      "grad_norm": 1.0134860465495785,
      "learning_rate": 0.00012924154917161974,
      "loss": 1.6328,
      "step": 5842
    },
    {
      "epoch": 3.8695364238410597,
      "grad_norm": 0.9501853581674249,
      "learning_rate": 0.0001291754457658716,
      "loss": 1.6562,
      "step": 5843
    },
    {
      "epoch": 3.8701986754966886,
      "grad_norm": 0.9392465949302188,
      "learning_rate": 0.000129109346483624,
      "loss": 1.5391,
      "step": 5844
    },
    {
      "epoch": 3.870860927152318,
      "grad_norm": 0.9766935264149513,
      "learning_rate": 0.00012904325133796532,
      "loss": 1.625,
      "step": 5845
    },
    {
      "epoch": 3.871523178807947,
      "grad_norm": 0.9997392811043313,
      "learning_rate": 0.0001289771603419831,
      "loss": 1.6875,
      "step": 5846
    },
    {
      "epoch": 3.872185430463576,
      "grad_norm": 0.9211031771490494,
      "learning_rate": 0.00012891107350876415,
      "loss": 1.5781,
      "step": 5847
    },
    {
      "epoch": 3.8728476821192053,
      "grad_norm": 1.0506552094682335,
      "learning_rate": 0.00012884499085139445,
      "loss": 1.7969,
      "step": 5848
    },
    {
      "epoch": 3.8735099337748347,
      "grad_norm": 1.0350422627669538,
      "learning_rate": 0.00012877891238295905,
      "loss": 1.8438,
      "step": 5849
    },
    {
      "epoch": 3.8741721854304636,
      "grad_norm": 1.0429563806091422,
      "learning_rate": 0.00012871283811654232,
      "loss": 1.7109,
      "step": 5850
    },
    {
      "epoch": 3.8748344370860925,
      "grad_norm": 1.1129951296119285,
      "learning_rate": 0.0001286467680652277,
      "loss": 1.9766,
      "step": 5851
    },
    {
      "epoch": 3.875496688741722,
      "grad_norm": 1.0690717537303676,
      "learning_rate": 0.00012858070224209782,
      "loss": 1.8516,
      "step": 5852
    },
    {
      "epoch": 3.876158940397351,
      "grad_norm": 0.9170043640436769,
      "learning_rate": 0.0001285146406602344,
      "loss": 1.6797,
      "step": 5853
    },
    {
      "epoch": 3.8768211920529803,
      "grad_norm": 1.09998301954349,
      "learning_rate": 0.0001284485833327185,
      "loss": 1.9922,
      "step": 5854
    },
    {
      "epoch": 3.877483443708609,
      "grad_norm": 0.9953467382613767,
      "learning_rate": 0.00012838253027263021,
      "loss": 1.9062,
      "step": 5855
    },
    {
      "epoch": 3.878145695364238,
      "grad_norm": 0.957293347289259,
      "learning_rate": 0.00012831648149304867,
      "loss": 1.7656,
      "step": 5856
    },
    {
      "epoch": 3.8788079470198675,
      "grad_norm": 0.9369865366557445,
      "learning_rate": 0.00012825043700705245,
      "loss": 1.8516,
      "step": 5857
    },
    {
      "epoch": 3.879470198675497,
      "grad_norm": 0.8740387815601298,
      "learning_rate": 0.00012818439682771906,
      "loss": 1.4375,
      "step": 5858
    },
    {
      "epoch": 3.880132450331126,
      "grad_norm": 0.9748721026645389,
      "learning_rate": 0.0001281183609681252,
      "loss": 1.6172,
      "step": 5859
    },
    {
      "epoch": 3.880794701986755,
      "grad_norm": 0.9648312107805943,
      "learning_rate": 0.00012805232944134668,
      "loss": 1.5781,
      "step": 5860
    },
    {
      "epoch": 3.881456953642384,
      "grad_norm": 1.0169843003840577,
      "learning_rate": 0.00012798630226045865,
      "loss": 1.8359,
      "step": 5861
    },
    {
      "epoch": 3.882119205298013,
      "grad_norm": 0.950052281545286,
      "learning_rate": 0.00012792027943853504,
      "loss": 1.7109,
      "step": 5862
    },
    {
      "epoch": 3.8827814569536425,
      "grad_norm": 0.9227243137423408,
      "learning_rate": 0.0001278542609886493,
      "loss": 1.5469,
      "step": 5863
    },
    {
      "epoch": 3.8834437086092715,
      "grad_norm": 1.036666229188495,
      "learning_rate": 0.0001277882469238738,
      "loss": 1.875,
      "step": 5864
    },
    {
      "epoch": 3.8841059602649004,
      "grad_norm": 0.96127561131324,
      "learning_rate": 0.00012772223725728,
      "loss": 1.6562,
      "step": 5865
    },
    {
      "epoch": 3.88476821192053,
      "grad_norm": 0.8542816196165587,
      "learning_rate": 0.00012765623200193866,
      "loss": 1.3906,
      "step": 5866
    },
    {
      "epoch": 3.885430463576159,
      "grad_norm": 0.920788879607266,
      "learning_rate": 0.00012759023117091958,
      "loss": 1.4922,
      "step": 5867
    },
    {
      "epoch": 3.886092715231788,
      "grad_norm": 1.0282958755440192,
      "learning_rate": 0.00012752423477729158,
      "loss": 1.7344,
      "step": 5868
    },
    {
      "epoch": 3.886754966887417,
      "grad_norm": 0.9867654313763583,
      "learning_rate": 0.0001274582428341228,
      "loss": 1.6953,
      "step": 5869
    },
    {
      "epoch": 3.8874172185430464,
      "grad_norm": 1.0467102657211078,
      "learning_rate": 0.0001273922553544804,
      "loss": 1.9844,
      "step": 5870
    },
    {
      "epoch": 3.8880794701986754,
      "grad_norm": 0.869763609556392,
      "learning_rate": 0.00012732627235143056,
      "loss": 1.4297,
      "step": 5871
    },
    {
      "epoch": 3.888741721854305,
      "grad_norm": 1.0545068625668788,
      "learning_rate": 0.00012726029383803883,
      "loss": 1.7578,
      "step": 5872
    },
    {
      "epoch": 3.8894039735099337,
      "grad_norm": 0.9842961714746176,
      "learning_rate": 0.00012719431982736957,
      "loss": 1.6406,
      "step": 5873
    },
    {
      "epoch": 3.8900662251655627,
      "grad_norm": 1.004152845974112,
      "learning_rate": 0.00012712835033248642,
      "loss": 1.6641,
      "step": 5874
    },
    {
      "epoch": 3.890728476821192,
      "grad_norm": 0.964149595314734,
      "learning_rate": 0.00012706238536645215,
      "loss": 1.7188,
      "step": 5875
    },
    {
      "epoch": 3.8913907284768214,
      "grad_norm": 0.9131544370962704,
      "learning_rate": 0.00012699642494232855,
      "loss": 1.5859,
      "step": 5876
    },
    {
      "epoch": 3.8920529801324504,
      "grad_norm": 1.0021870372446064,
      "learning_rate": 0.0001269304690731765,
      "loss": 1.8203,
      "step": 5877
    },
    {
      "epoch": 3.8927152317880793,
      "grad_norm": 0.9783535160857425,
      "learning_rate": 0.00012686451777205613,
      "loss": 1.6094,
      "step": 5878
    },
    {
      "epoch": 3.8933774834437087,
      "grad_norm": 0.8709895770505574,
      "learning_rate": 0.00012679857105202645,
      "loss": 1.3906,
      "step": 5879
    },
    {
      "epoch": 3.8940397350993377,
      "grad_norm": 1.0268328740158517,
      "learning_rate": 0.00012673262892614565,
      "loss": 1.9141,
      "step": 5880
    },
    {
      "epoch": 3.894701986754967,
      "grad_norm": 0.9372645488307642,
      "learning_rate": 0.00012666669140747115,
      "loss": 1.4688,
      "step": 5881
    },
    {
      "epoch": 3.895364238410596,
      "grad_norm": 0.8629146653294747,
      "learning_rate": 0.00012660075850905926,
      "loss": 1.3828,
      "step": 5882
    },
    {
      "epoch": 3.896026490066225,
      "grad_norm": 0.9777150611349158,
      "learning_rate": 0.00012653483024396533,
      "loss": 1.7344,
      "step": 5883
    },
    {
      "epoch": 3.8966887417218543,
      "grad_norm": 0.9929216255709661,
      "learning_rate": 0.00012646890662524414,
      "loss": 1.5938,
      "step": 5884
    },
    {
      "epoch": 3.8973509933774837,
      "grad_norm": 1.1211016784698822,
      "learning_rate": 0.0001264029876659492,
      "loss": 1.8359,
      "step": 5885
    },
    {
      "epoch": 3.8980132450331126,
      "grad_norm": 1.072560670779608,
      "learning_rate": 0.0001263370733791332,
      "loss": 1.8047,
      "step": 5886
    },
    {
      "epoch": 3.8986754966887416,
      "grad_norm": 1.0068384054783504,
      "learning_rate": 0.00012627116377784798,
      "loss": 1.6562,
      "step": 5887
    },
    {
      "epoch": 3.899337748344371,
      "grad_norm": 1.0653065836067976,
      "learning_rate": 0.00012620525887514439,
      "loss": 1.8125,
      "step": 5888
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0283730253106607,
      "learning_rate": 0.00012613935868407226,
      "loss": 1.6953,
      "step": 5889
    },
    {
      "epoch": 3.9006622516556293,
      "grad_norm": 1.0458955382640824,
      "learning_rate": 0.00012607346321768078,
      "loss": 1.8125,
      "step": 5890
    },
    {
      "epoch": 3.9013245033112582,
      "grad_norm": 0.9267144549066699,
      "learning_rate": 0.00012600757248901784,
      "loss": 1.6094,
      "step": 5891
    },
    {
      "epoch": 3.901986754966887,
      "grad_norm": 1.0351049928685543,
      "learning_rate": 0.0001259416865111306,
      "loss": 1.8281,
      "step": 5892
    },
    {
      "epoch": 3.9026490066225166,
      "grad_norm": 0.9135677889404842,
      "learning_rate": 0.00012587580529706523,
      "loss": 1.4297,
      "step": 5893
    },
    {
      "epoch": 3.903311258278146,
      "grad_norm": 0.9449771234070313,
      "learning_rate": 0.00012580992885986706,
      "loss": 1.6953,
      "step": 5894
    },
    {
      "epoch": 3.903973509933775,
      "grad_norm": 0.8346852287918528,
      "learning_rate": 0.00012574405721258025,
      "loss": 1.4062,
      "step": 5895
    },
    {
      "epoch": 3.904635761589404,
      "grad_norm": 0.9715417089160522,
      "learning_rate": 0.0001256781903682482,
      "loss": 1.7188,
      "step": 5896
    },
    {
      "epoch": 3.9052980132450332,
      "grad_norm": 0.9947939215971671,
      "learning_rate": 0.00012561232833991333,
      "loss": 1.5859,
      "step": 5897
    },
    {
      "epoch": 3.905960264900662,
      "grad_norm": 0.9793867250789362,
      "learning_rate": 0.00012554647114061698,
      "loss": 1.5859,
      "step": 5898
    },
    {
      "epoch": 3.9066225165562916,
      "grad_norm": 1.048347069842256,
      "learning_rate": 0.0001254806187833998,
      "loss": 1.7969,
      "step": 5899
    },
    {
      "epoch": 3.9072847682119205,
      "grad_norm": 1.0826426274022765,
      "learning_rate": 0.00012541477128130113,
      "loss": 1.9219,
      "step": 5900
    },
    {
      "epoch": 3.9079470198675494,
      "grad_norm": 0.9690639843832329,
      "learning_rate": 0.0001253489286473596,
      "loss": 1.5312,
      "step": 5901
    },
    {
      "epoch": 3.908609271523179,
      "grad_norm": 1.048630131969388,
      "learning_rate": 0.0001252830908946128,
      "loss": 1.7656,
      "step": 5902
    },
    {
      "epoch": 3.909271523178808,
      "grad_norm": 1.0125384441146907,
      "learning_rate": 0.00012521725803609743,
      "loss": 1.6484,
      "step": 5903
    },
    {
      "epoch": 3.909933774834437,
      "grad_norm": 1.0592439975757326,
      "learning_rate": 0.00012515143008484908,
      "loss": 1.8984,
      "step": 5904
    },
    {
      "epoch": 3.910596026490066,
      "grad_norm": 1.0347052524874842,
      "learning_rate": 0.0001250856070539024,
      "loss": 1.8594,
      "step": 5905
    },
    {
      "epoch": 3.9112582781456955,
      "grad_norm": 0.9763155137248515,
      "learning_rate": 0.00012501978895629122,
      "loss": 1.6953,
      "step": 5906
    },
    {
      "epoch": 3.9119205298013244,
      "grad_norm": 0.9430461396406326,
      "learning_rate": 0.0001249539758050482,
      "loss": 1.5859,
      "step": 5907
    },
    {
      "epoch": 3.912582781456954,
      "grad_norm": 0.9297064875882691,
      "learning_rate": 0.00012488816761320508,
      "loss": 1.4531,
      "step": 5908
    },
    {
      "epoch": 3.9132450331125828,
      "grad_norm": 1.0624510206137043,
      "learning_rate": 0.00012482236439379275,
      "loss": 1.8125,
      "step": 5909
    },
    {
      "epoch": 3.9139072847682117,
      "grad_norm": 0.9874905536719288,
      "learning_rate": 0.00012475656615984086,
      "loss": 1.6094,
      "step": 5910
    },
    {
      "epoch": 3.914569536423841,
      "grad_norm": 0.8719244310316486,
      "learning_rate": 0.00012469077292437826,
      "loss": 1.5625,
      "step": 5911
    },
    {
      "epoch": 3.91523178807947,
      "grad_norm": 0.9836125894637188,
      "learning_rate": 0.0001246249847004328,
      "loss": 1.5703,
      "step": 5912
    },
    {
      "epoch": 3.9158940397350994,
      "grad_norm": 0.8982377513872685,
      "learning_rate": 0.00012455920150103135,
      "loss": 1.4219,
      "step": 5913
    },
    {
      "epoch": 3.9165562913907284,
      "grad_norm": 0.8948373727141572,
      "learning_rate": 0.00012449342333919954,
      "loss": 1.4609,
      "step": 5914
    },
    {
      "epoch": 3.9172185430463577,
      "grad_norm": 1.0581692211663396,
      "learning_rate": 0.00012442765022796246,
      "loss": 1.8594,
      "step": 5915
    },
    {
      "epoch": 3.9178807947019867,
      "grad_norm": 0.9364056756042234,
      "learning_rate": 0.00012436188218034377,
      "loss": 1.5312,
      "step": 5916
    },
    {
      "epoch": 3.918543046357616,
      "grad_norm": 0.9663274615176431,
      "learning_rate": 0.00012429611920936629,
      "loss": 1.6172,
      "step": 5917
    },
    {
      "epoch": 3.919205298013245,
      "grad_norm": 1.0397212101636628,
      "learning_rate": 0.00012423036132805192,
      "loss": 1.6484,
      "step": 5918
    },
    {
      "epoch": 3.919867549668874,
      "grad_norm": 1.06424461569124,
      "learning_rate": 0.0001241646085494215,
      "loss": 1.8594,
      "step": 5919
    },
    {
      "epoch": 3.9205298013245033,
      "grad_norm": 0.9810239936307137,
      "learning_rate": 0.0001240988608864947,
      "loss": 1.5156,
      "step": 5920
    },
    {
      "epoch": 3.9211920529801323,
      "grad_norm": 1.1091338189020195,
      "learning_rate": 0.00012403311835229043,
      "loss": 1.8438,
      "step": 5921
    },
    {
      "epoch": 3.9218543046357617,
      "grad_norm": 0.9969446338102435,
      "learning_rate": 0.00012396738095982646,
      "loss": 1.6406,
      "step": 5922
    },
    {
      "epoch": 3.9225165562913906,
      "grad_norm": 0.9563424963387369,
      "learning_rate": 0.00012390164872211944,
      "loss": 1.6875,
      "step": 5923
    },
    {
      "epoch": 3.92317880794702,
      "grad_norm": 0.9985475950276158,
      "learning_rate": 0.00012383592165218522,
      "loss": 1.6719,
      "step": 5924
    },
    {
      "epoch": 3.923841059602649,
      "grad_norm": 1.0030571849737733,
      "learning_rate": 0.00012377019976303854,
      "loss": 1.8203,
      "step": 5925
    },
    {
      "epoch": 3.9245033112582783,
      "grad_norm": 1.0023968425257104,
      "learning_rate": 0.00012370448306769292,
      "loss": 1.8438,
      "step": 5926
    },
    {
      "epoch": 3.9251655629139073,
      "grad_norm": 0.9814640043562964,
      "learning_rate": 0.00012363877157916116,
      "loss": 1.5703,
      "step": 5927
    },
    {
      "epoch": 3.9258278145695362,
      "grad_norm": 0.9219190252015685,
      "learning_rate": 0.00012357306531045483,
      "loss": 1.4219,
      "step": 5928
    },
    {
      "epoch": 3.9264900662251656,
      "grad_norm": 1.0046704231576273,
      "learning_rate": 0.00012350736427458453,
      "loss": 1.6328,
      "step": 5929
    },
    {
      "epoch": 3.9271523178807946,
      "grad_norm": 1.1287223928500771,
      "learning_rate": 0.00012344166848455987,
      "loss": 1.9609,
      "step": 5930
    },
    {
      "epoch": 3.927814569536424,
      "grad_norm": 1.0293598361819116,
      "learning_rate": 0.00012337597795338936,
      "loss": 1.7578,
      "step": 5931
    },
    {
      "epoch": 3.928476821192053,
      "grad_norm": 0.9281624602392251,
      "learning_rate": 0.0001233102926940804,
      "loss": 1.625,
      "step": 5932
    },
    {
      "epoch": 3.9291390728476823,
      "grad_norm": 1.0896770117777286,
      "learning_rate": 0.00012324461271963949,
      "loss": 1.8828,
      "step": 5933
    },
    {
      "epoch": 3.929801324503311,
      "grad_norm": 0.9079346473691484,
      "learning_rate": 0.000123178938043072,
      "loss": 1.5234,
      "step": 5934
    },
    {
      "epoch": 3.9304635761589406,
      "grad_norm": 0.950803156088209,
      "learning_rate": 0.00012311326867738226,
      "loss": 1.5234,
      "step": 5935
    },
    {
      "epoch": 3.9311258278145695,
      "grad_norm": 0.9729264769664621,
      "learning_rate": 0.00012304760463557359,
      "loss": 1.7031,
      "step": 5936
    },
    {
      "epoch": 3.9317880794701985,
      "grad_norm": 1.044359922820051,
      "learning_rate": 0.0001229819459306482,
      "loss": 1.7656,
      "step": 5937
    },
    {
      "epoch": 3.932450331125828,
      "grad_norm": 1.0102225825056053,
      "learning_rate": 0.00012291629257560723,
      "loss": 1.6875,
      "step": 5938
    },
    {
      "epoch": 3.933112582781457,
      "grad_norm": 1.0713630773780578,
      "learning_rate": 0.00012285064458345086,
      "loss": 1.8516,
      "step": 5939
    },
    {
      "epoch": 3.933774834437086,
      "grad_norm": 1.0214244574864582,
      "learning_rate": 0.00012278500196717817,
      "loss": 1.7188,
      "step": 5940
    },
    {
      "epoch": 3.934437086092715,
      "grad_norm": 1.0500488854471932,
      "learning_rate": 0.00012271936473978699,
      "loss": 1.7891,
      "step": 5941
    },
    {
      "epoch": 3.9350993377483445,
      "grad_norm": 0.9670262089761591,
      "learning_rate": 0.00012265373291427446,
      "loss": 1.625,
      "step": 5942
    },
    {
      "epoch": 3.9357615894039735,
      "grad_norm": 0.8897064329153318,
      "learning_rate": 0.00012258810650363629,
      "loss": 1.5,
      "step": 5943
    },
    {
      "epoch": 3.936423841059603,
      "grad_norm": 0.8195610557444866,
      "learning_rate": 0.00012252248552086723,
      "loss": 1.3203,
      "step": 5944
    },
    {
      "epoch": 3.937086092715232,
      "grad_norm": 1.0036867643688288,
      "learning_rate": 0.00012245686997896114,
      "loss": 1.6719,
      "step": 5945
    },
    {
      "epoch": 3.9377483443708607,
      "grad_norm": 0.9095383813866591,
      "learning_rate": 0.0001223912598909106,
      "loss": 1.4297,
      "step": 5946
    },
    {
      "epoch": 3.93841059602649,
      "grad_norm": 0.9776995773326178,
      "learning_rate": 0.000122325655269707,
      "loss": 1.5156,
      "step": 5947
    },
    {
      "epoch": 3.939072847682119,
      "grad_norm": 0.9488376480693868,
      "learning_rate": 0.00012226005612834103,
      "loss": 1.6875,
      "step": 5948
    },
    {
      "epoch": 3.9397350993377485,
      "grad_norm": 1.0010495663157601,
      "learning_rate": 0.00012219446247980196,
      "loss": 1.625,
      "step": 5949
    },
    {
      "epoch": 3.9403973509933774,
      "grad_norm": 0.9643902103222894,
      "learning_rate": 0.00012212887433707805,
      "loss": 1.5078,
      "step": 5950
    },
    {
      "epoch": 3.941059602649007,
      "grad_norm": 1.0643656736341296,
      "learning_rate": 0.0001220632917131566,
      "loss": 1.9609,
      "step": 5951
    },
    {
      "epoch": 3.9417218543046357,
      "grad_norm": 0.9874863803986343,
      "learning_rate": 0.0001219977146210237,
      "loss": 1.5625,
      "step": 5952
    },
    {
      "epoch": 3.942384105960265,
      "grad_norm": 1.0440571283964017,
      "learning_rate": 0.00012193214307366428,
      "loss": 1.6016,
      "step": 5953
    },
    {
      "epoch": 3.943046357615894,
      "grad_norm": 1.0033139474758266,
      "learning_rate": 0.00012186657708406236,
      "loss": 1.6562,
      "step": 5954
    },
    {
      "epoch": 3.943708609271523,
      "grad_norm": 0.8221849043276196,
      "learning_rate": 0.0001218010166652007,
      "loss": 1.2969,
      "step": 5955
    },
    {
      "epoch": 3.9443708609271524,
      "grad_norm": 0.9716002401389392,
      "learning_rate": 0.000121735461830061,
      "loss": 1.5234,
      "step": 5956
    },
    {
      "epoch": 3.9450331125827813,
      "grad_norm": 0.9974501994455075,
      "learning_rate": 0.00012166991259162392,
      "loss": 1.6094,
      "step": 5957
    },
    {
      "epoch": 3.9456953642384107,
      "grad_norm": 1.0216206300416164,
      "learning_rate": 0.000121604368962869,
      "loss": 1.7266,
      "step": 5958
    },
    {
      "epoch": 3.9463576158940397,
      "grad_norm": 0.9967694006653924,
      "learning_rate": 0.00012153883095677454,
      "loss": 1.6719,
      "step": 5959
    },
    {
      "epoch": 3.9470198675496686,
      "grad_norm": 1.0359284825958364,
      "learning_rate": 0.00012147329858631779,
      "loss": 1.8359,
      "step": 5960
    },
    {
      "epoch": 3.947682119205298,
      "grad_norm": 0.9633156652934463,
      "learning_rate": 0.00012140777186447502,
      "loss": 1.6172,
      "step": 5961
    },
    {
      "epoch": 3.9483443708609274,
      "grad_norm": 0.9452832282136642,
      "learning_rate": 0.00012134225080422124,
      "loss": 1.6719,
      "step": 5962
    },
    {
      "epoch": 3.9490066225165563,
      "grad_norm": 1.0042114035486978,
      "learning_rate": 0.00012127673541853028,
      "loss": 1.7422,
      "step": 5963
    },
    {
      "epoch": 3.9496688741721853,
      "grad_norm": 0.9415386021728477,
      "learning_rate": 0.00012121122572037506,
      "loss": 1.5469,
      "step": 5964
    },
    {
      "epoch": 3.9503311258278146,
      "grad_norm": 1.01445552286751,
      "learning_rate": 0.00012114572172272719,
      "loss": 1.6719,
      "step": 5965
    },
    {
      "epoch": 3.9509933774834436,
      "grad_norm": 0.7838841489783515,
      "learning_rate": 0.0001210802234385572,
      "loss": 1.3125,
      "step": 5966
    },
    {
      "epoch": 3.951655629139073,
      "grad_norm": 0.880522595695106,
      "learning_rate": 0.00012101473088083458,
      "loss": 1.5156,
      "step": 5967
    },
    {
      "epoch": 3.952317880794702,
      "grad_norm": 0.9835173692418814,
      "learning_rate": 0.00012094924406252752,
      "loss": 1.6016,
      "step": 5968
    },
    {
      "epoch": 3.952980132450331,
      "grad_norm": 1.0265704125492976,
      "learning_rate": 0.00012088376299660316,
      "loss": 1.7812,
      "step": 5969
    },
    {
      "epoch": 3.9536423841059603,
      "grad_norm": 1.0227762749913194,
      "learning_rate": 0.00012081828769602756,
      "loss": 1.5703,
      "step": 5970
    },
    {
      "epoch": 3.9543046357615896,
      "grad_norm": 0.9037523186626133,
      "learning_rate": 0.00012075281817376557,
      "loss": 1.4375,
      "step": 5971
    },
    {
      "epoch": 3.9549668874172186,
      "grad_norm": 0.9868944250775389,
      "learning_rate": 0.00012068735444278082,
      "loss": 1.6562,
      "step": 5972
    },
    {
      "epoch": 3.9556291390728475,
      "grad_norm": 0.9209851956532732,
      "learning_rate": 0.00012062189651603605,
      "loss": 1.5078,
      "step": 5973
    },
    {
      "epoch": 3.956291390728477,
      "grad_norm": 1.0208474925142401,
      "learning_rate": 0.00012055644440649255,
      "loss": 1.8438,
      "step": 5974
    },
    {
      "epoch": 3.956953642384106,
      "grad_norm": 0.8344083178918509,
      "learning_rate": 0.00012049099812711055,
      "loss": 1.3203,
      "step": 5975
    },
    {
      "epoch": 3.9576158940397352,
      "grad_norm": 1.0165818378408986,
      "learning_rate": 0.00012042555769084928,
      "loss": 1.7188,
      "step": 5976
    },
    {
      "epoch": 3.958278145695364,
      "grad_norm": 0.8743450452572842,
      "learning_rate": 0.00012036012311066669,
      "loss": 1.5078,
      "step": 5977
    },
    {
      "epoch": 3.958940397350993,
      "grad_norm": 0.9494047538185969,
      "learning_rate": 0.00012029469439951942,
      "loss": 1.625,
      "step": 5978
    },
    {
      "epoch": 3.9596026490066225,
      "grad_norm": 1.0440633321122885,
      "learning_rate": 0.00012022927157036333,
      "loss": 2.0,
      "step": 5979
    },
    {
      "epoch": 3.960264900662252,
      "grad_norm": 0.9011934978391383,
      "learning_rate": 0.00012016385463615275,
      "loss": 1.3984,
      "step": 5980
    },
    {
      "epoch": 3.960927152317881,
      "grad_norm": 1.045971981148548,
      "learning_rate": 0.00012009844360984096,
      "loss": 1.8594,
      "step": 5981
    },
    {
      "epoch": 3.96158940397351,
      "grad_norm": 0.9554913497313078,
      "learning_rate": 0.00012003303850438019,
      "loss": 1.6484,
      "step": 5982
    },
    {
      "epoch": 3.962251655629139,
      "grad_norm": 0.9611305299763574,
      "learning_rate": 0.0001199676393327214,
      "loss": 1.6016,
      "step": 5983
    },
    {
      "epoch": 3.962913907284768,
      "grad_norm": 0.9438569027705873,
      "learning_rate": 0.00011990224610781421,
      "loss": 1.6016,
      "step": 5984
    },
    {
      "epoch": 3.9635761589403975,
      "grad_norm": 0.957260968360181,
      "learning_rate": 0.00011983685884260747,
      "loss": 1.6484,
      "step": 5985
    },
    {
      "epoch": 3.9642384105960264,
      "grad_norm": 0.8249719930959608,
      "learning_rate": 0.00011977147755004844,
      "loss": 1.3438,
      "step": 5986
    },
    {
      "epoch": 3.9649006622516554,
      "grad_norm": 1.030002225196568,
      "learning_rate": 0.00011970610224308341,
      "loss": 1.8281,
      "step": 5987
    },
    {
      "epoch": 3.9655629139072848,
      "grad_norm": 0.9614518389287605,
      "learning_rate": 0.00011964073293465744,
      "loss": 1.5312,
      "step": 5988
    },
    {
      "epoch": 3.966225165562914,
      "grad_norm": 1.0197959521456355,
      "learning_rate": 0.00011957536963771445,
      "loss": 1.8125,
      "step": 5989
    },
    {
      "epoch": 3.966887417218543,
      "grad_norm": 1.0295459023892108,
      "learning_rate": 0.00011951001236519701,
      "loss": 1.9375,
      "step": 5990
    },
    {
      "epoch": 3.967549668874172,
      "grad_norm": 1.0102801695364851,
      "learning_rate": 0.00011944466113004672,
      "loss": 1.75,
      "step": 5991
    },
    {
      "epoch": 3.9682119205298014,
      "grad_norm": 0.8783749337395543,
      "learning_rate": 0.00011937931594520385,
      "loss": 1.4609,
      "step": 5992
    },
    {
      "epoch": 3.9688741721854304,
      "grad_norm": 1.0270371887594265,
      "learning_rate": 0.00011931397682360741,
      "loss": 1.6797,
      "step": 5993
    },
    {
      "epoch": 3.9695364238410598,
      "grad_norm": 1.044666297418703,
      "learning_rate": 0.00011924864377819545,
      "loss": 1.8906,
      "step": 5994
    },
    {
      "epoch": 3.9701986754966887,
      "grad_norm": 0.9429889919620572,
      "learning_rate": 0.00011918331682190454,
      "loss": 1.625,
      "step": 5995
    },
    {
      "epoch": 3.9708609271523176,
      "grad_norm": 1.047992362691198,
      "learning_rate": 0.00011911799596767016,
      "loss": 1.7656,
      "step": 5996
    },
    {
      "epoch": 3.971523178807947,
      "grad_norm": 1.0892263762817587,
      "learning_rate": 0.00011905268122842665,
      "loss": 1.8281,
      "step": 5997
    },
    {
      "epoch": 3.9721854304635764,
      "grad_norm": 0.9938213694353012,
      "learning_rate": 0.00011898737261710707,
      "loss": 1.7266,
      "step": 5998
    },
    {
      "epoch": 3.9728476821192054,
      "grad_norm": 1.0173092095974985,
      "learning_rate": 0.00011892207014664324,
      "loss": 1.6562,
      "step": 5999
    },
    {
      "epoch": 3.9735099337748343,
      "grad_norm": 1.1267495337940885,
      "learning_rate": 0.00011885677382996587,
      "loss": 2.0156,
      "step": 6000
    },
    {
      "epoch": 3.9741721854304637,
      "grad_norm": 0.9297865360219322,
      "learning_rate": 0.00011879148368000429,
      "loss": 1.6328,
      "step": 6001
    },
    {
      "epoch": 3.9748344370860926,
      "grad_norm": 8.587726232289832,
      "learning_rate": 0.00011872619970968671,
      "loss": 1.6016,
      "step": 6002
    },
    {
      "epoch": 3.975496688741722,
      "grad_norm": 0.9185089325463119,
      "learning_rate": 0.00011866092193194017,
      "loss": 1.5391,
      "step": 6003
    },
    {
      "epoch": 3.976158940397351,
      "grad_norm": 0.9202026167686796,
      "learning_rate": 0.00011859565035969039,
      "loss": 1.4766,
      "step": 6004
    },
    {
      "epoch": 3.97682119205298,
      "grad_norm": 0.7841295988463723,
      "learning_rate": 0.0001185303850058618,
      "loss": 1.2266,
      "step": 6005
    },
    {
      "epoch": 3.9774834437086093,
      "grad_norm": 0.9823307399984005,
      "learning_rate": 0.00011846512588337786,
      "loss": 1.6094,
      "step": 6006
    },
    {
      "epoch": 3.9781456953642387,
      "grad_norm": 0.9848627286114742,
      "learning_rate": 0.0001183998730051605,
      "loss": 1.7266,
      "step": 6007
    },
    {
      "epoch": 3.9788079470198676,
      "grad_norm": 0.9895148946902289,
      "learning_rate": 0.00011833462638413053,
      "loss": 1.5391,
      "step": 6008
    },
    {
      "epoch": 3.9794701986754966,
      "grad_norm": 0.9650377811018416,
      "learning_rate": 0.00011826938603320759,
      "loss": 1.6328,
      "step": 6009
    },
    {
      "epoch": 3.980132450331126,
      "grad_norm": 1.0260845778965442,
      "learning_rate": 0.00011820415196531004,
      "loss": 1.6172,
      "step": 6010
    },
    {
      "epoch": 3.980794701986755,
      "grad_norm": 0.9528885037589999,
      "learning_rate": 0.00011813892419335483,
      "loss": 1.5391,
      "step": 6011
    },
    {
      "epoch": 3.9814569536423843,
      "grad_norm": 0.9856470385874971,
      "learning_rate": 0.000118073702730258,
      "loss": 1.6094,
      "step": 6012
    },
    {
      "epoch": 3.982119205298013,
      "grad_norm": 1.016281948867364,
      "learning_rate": 0.00011800848758893402,
      "loss": 1.7188,
      "step": 6013
    },
    {
      "epoch": 3.982781456953642,
      "grad_norm": 1.0071027348979533,
      "learning_rate": 0.00011794327878229629,
      "loss": 1.6328,
      "step": 6014
    },
    {
      "epoch": 3.9834437086092715,
      "grad_norm": 1.0196810093717645,
      "learning_rate": 0.0001178780763232568,
      "loss": 1.6719,
      "step": 6015
    },
    {
      "epoch": 3.984105960264901,
      "grad_norm": 0.9583299437773576,
      "learning_rate": 0.00011781288022472655,
      "loss": 1.5078,
      "step": 6016
    },
    {
      "epoch": 3.98476821192053,
      "grad_norm": 1.0359187393386555,
      "learning_rate": 0.00011774769049961501,
      "loss": 1.6562,
      "step": 6017
    },
    {
      "epoch": 3.985430463576159,
      "grad_norm": 1.1924712882228408,
      "learning_rate": 0.00011768250716083046,
      "loss": 2.0156,
      "step": 6018
    },
    {
      "epoch": 3.986092715231788,
      "grad_norm": 1.089328310253908,
      "learning_rate": 0.00011761733022128006,
      "loss": 1.7344,
      "step": 6019
    },
    {
      "epoch": 3.986754966887417,
      "grad_norm": 1.0246044973369064,
      "learning_rate": 0.00011755215969386954,
      "loss": 1.6797,
      "step": 6020
    },
    {
      "epoch": 3.9874172185430465,
      "grad_norm": 1.0244430954779407,
      "learning_rate": 0.00011748699559150336,
      "loss": 1.8828,
      "step": 6021
    },
    {
      "epoch": 3.9880794701986755,
      "grad_norm": 0.8980850522942763,
      "learning_rate": 0.0001174218379270848,
      "loss": 1.4922,
      "step": 6022
    },
    {
      "epoch": 3.9887417218543044,
      "grad_norm": 1.0316601983913096,
      "learning_rate": 0.00011735668671351585,
      "loss": 1.7656,
      "step": 6023
    },
    {
      "epoch": 3.989403973509934,
      "grad_norm": 1.0674478580151399,
      "learning_rate": 0.00011729154196369715,
      "loss": 1.8047,
      "step": 6024
    },
    {
      "epoch": 3.9900662251655628,
      "grad_norm": 0.979390877514744,
      "learning_rate": 0.00011722640369052814,
      "loss": 1.7422,
      "step": 6025
    },
    {
      "epoch": 3.990728476821192,
      "grad_norm": 1.049215990898217,
      "learning_rate": 0.00011716127190690697,
      "loss": 1.7578,
      "step": 6026
    },
    {
      "epoch": 3.991390728476821,
      "grad_norm": 1.0380192101443995,
      "learning_rate": 0.0001170961466257304,
      "loss": 1.8828,
      "step": 6027
    },
    {
      "epoch": 3.9920529801324505,
      "grad_norm": 0.9705093209116584,
      "learning_rate": 0.00011703102785989405,
      "loss": 1.6406,
      "step": 6028
    },
    {
      "epoch": 3.9927152317880794,
      "grad_norm": 1.0071879018076397,
      "learning_rate": 0.00011696591562229217,
      "loss": 1.6484,
      "step": 6029
    },
    {
      "epoch": 3.993377483443709,
      "grad_norm": 0.9147838453901272,
      "learning_rate": 0.00011690080992581768,
      "loss": 1.5781,
      "step": 6030
    },
    {
      "epoch": 3.9940397350993377,
      "grad_norm": 0.9532397247780353,
      "learning_rate": 0.00011683571078336236,
      "loss": 1.7734,
      "step": 6031
    },
    {
      "epoch": 3.9947019867549667,
      "grad_norm": 0.9470989205599568,
      "learning_rate": 0.00011677061820781649,
      "loss": 1.6172,
      "step": 6032
    },
    {
      "epoch": 3.995364238410596,
      "grad_norm": 0.9660650446767381,
      "learning_rate": 0.00011670553221206914,
      "loss": 1.7188,
      "step": 6033
    },
    {
      "epoch": 3.996026490066225,
      "grad_norm": 0.9931590095839206,
      "learning_rate": 0.00011664045280900815,
      "loss": 1.7656,
      "step": 6034
    },
    {
      "epoch": 3.9966887417218544,
      "grad_norm": 0.9464276998651157,
      "learning_rate": 0.00011657538001151997,
      "loss": 1.5469,
      "step": 6035
    },
    {
      "epoch": 3.9973509933774833,
      "grad_norm": 0.9267903575032951,
      "learning_rate": 0.00011651031383248971,
      "loss": 1.3906,
      "step": 6036
    },
    {
      "epoch": 3.9980132450331127,
      "grad_norm": 1.0232663393473593,
      "learning_rate": 0.00011644525428480133,
      "loss": 1.7031,
      "step": 6037
    },
    {
      "epoch": 3.9986754966887417,
      "grad_norm": 0.9589901748908097,
      "learning_rate": 0.00011638020138133727,
      "loss": 1.5469,
      "step": 6038
    },
    {
      "epoch": 3.999337748344371,
      "grad_norm": 1.000400829760944,
      "learning_rate": 0.00011631515513497872,
      "loss": 1.5312,
      "step": 6039
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.9968704337632017,
      "learning_rate": 0.00011625011555860567,
      "loss": 1.7266,
      "step": 6040
    },
    {
      "epoch": 4.0,
      "eval_loss": 2.2305381298065186,
      "eval_runtime": 33.9662,
      "eval_samples_per_second": 9.951,
      "eval_steps_per_second": 9.951,
      "step": 6040
    },
    {
      "epoch": 4.000662251655629,
      "grad_norm": 0.8295225117868664,
      "learning_rate": 0.00011618508266509672,
      "loss": 1.1797,
      "step": 6041
    },
    {
      "epoch": 4.001324503311258,
      "grad_norm": 0.7266332386137614,
      "learning_rate": 0.00011612005646732897,
      "loss": 0.8516,
      "step": 6042
    },
    {
      "epoch": 4.001986754966888,
      "grad_norm": 0.7727855688630928,
      "learning_rate": 0.00011605503697817855,
      "loss": 1.0625,
      "step": 6043
    },
    {
      "epoch": 4.002649006622517,
      "grad_norm": 0.9028321973589735,
      "learning_rate": 0.00011599002421051996,
      "loss": 1.2344,
      "step": 6044
    },
    {
      "epoch": 4.003311258278146,
      "grad_norm": 0.765166278852301,
      "learning_rate": 0.00011592501817722645,
      "loss": 0.9688,
      "step": 6045
    },
    {
      "epoch": 4.0039735099337745,
      "grad_norm": 0.6569216129782834,
      "learning_rate": 0.00011586001889117002,
      "loss": 0.6719,
      "step": 6046
    },
    {
      "epoch": 4.004635761589404,
      "grad_norm": 0.7571735255158986,
      "learning_rate": 0.0001157950263652213,
      "loss": 0.8789,
      "step": 6047
    },
    {
      "epoch": 4.005298013245033,
      "grad_norm": 0.7659321357900977,
      "learning_rate": 0.00011573004061224944,
      "loss": 0.8477,
      "step": 6048
    },
    {
      "epoch": 4.005960264900662,
      "grad_norm": 0.9017468661723219,
      "learning_rate": 0.00011566506164512245,
      "loss": 0.9922,
      "step": 6049
    },
    {
      "epoch": 4.006622516556291,
      "grad_norm": 1.1376938621436214,
      "learning_rate": 0.00011560008947670689,
      "loss": 1.2266,
      "step": 6050
    },
    {
      "epoch": 4.00728476821192,
      "grad_norm": 1.0500217707359532,
      "learning_rate": 0.00011553512411986797,
      "loss": 1.0938,
      "step": 6051
    },
    {
      "epoch": 4.00794701986755,
      "grad_norm": 0.9951311941412714,
      "learning_rate": 0.00011547016558746963,
      "loss": 1.0312,
      "step": 6052
    },
    {
      "epoch": 4.008609271523179,
      "grad_norm": 0.9128312414997436,
      "learning_rate": 0.00011540521389237442,
      "loss": 0.9453,
      "step": 6053
    },
    {
      "epoch": 4.009271523178808,
      "grad_norm": 0.8697596319334847,
      "learning_rate": 0.0001153402690474434,
      "loss": 0.8789,
      "step": 6054
    },
    {
      "epoch": 4.009933774834437,
      "grad_norm": 0.9499494979222866,
      "learning_rate": 0.0001152753310655365,
      "loss": 0.8672,
      "step": 6055
    },
    {
      "epoch": 4.010596026490067,
      "grad_norm": 0.6655464246200212,
      "learning_rate": 0.00011521039995951217,
      "loss": 0.6172,
      "step": 6056
    },
    {
      "epoch": 4.011258278145696,
      "grad_norm": 1.0863903920417386,
      "learning_rate": 0.00011514547574222744,
      "loss": 0.9453,
      "step": 6057
    },
    {
      "epoch": 4.0119205298013245,
      "grad_norm": 0.9628375304753726,
      "learning_rate": 0.0001150805584265382,
      "loss": 0.8906,
      "step": 6058
    },
    {
      "epoch": 4.0125827814569535,
      "grad_norm": 1.147439567381915,
      "learning_rate": 0.00011501564802529869,
      "loss": 1.1016,
      "step": 6059
    },
    {
      "epoch": 4.013245033112582,
      "grad_norm": 0.7799343678360273,
      "learning_rate": 0.0001149507445513619,
      "loss": 0.6719,
      "step": 6060
    },
    {
      "epoch": 4.013907284768212,
      "grad_norm": 1.0997259834379338,
      "learning_rate": 0.00011488584801757957,
      "loss": 1.0547,
      "step": 6061
    },
    {
      "epoch": 4.014569536423841,
      "grad_norm": 0.867350769500672,
      "learning_rate": 0.00011482095843680189,
      "loss": 0.7578,
      "step": 6062
    },
    {
      "epoch": 4.01523178807947,
      "grad_norm": 0.7771363448981655,
      "learning_rate": 0.00011475607582187772,
      "loss": 0.6758,
      "step": 6063
    },
    {
      "epoch": 4.015894039735099,
      "grad_norm": 1.097822320698339,
      "learning_rate": 0.00011469120018565468,
      "loss": 1.0469,
      "step": 6064
    },
    {
      "epoch": 4.016556291390729,
      "grad_norm": 1.111989715189192,
      "learning_rate": 0.00011462633154097876,
      "loss": 1.1094,
      "step": 6065
    },
    {
      "epoch": 4.017218543046358,
      "grad_norm": 0.9528213136522604,
      "learning_rate": 0.0001145614699006947,
      "loss": 0.8906,
      "step": 6066
    },
    {
      "epoch": 4.017880794701987,
      "grad_norm": 0.933669287032352,
      "learning_rate": 0.00011449661527764594,
      "loss": 0.8281,
      "step": 6067
    },
    {
      "epoch": 4.018543046357616,
      "grad_norm": 0.9514160534723897,
      "learning_rate": 0.00011443176768467442,
      "loss": 0.8516,
      "step": 6068
    },
    {
      "epoch": 4.019205298013245,
      "grad_norm": 1.153412539569718,
      "learning_rate": 0.00011436692713462056,
      "loss": 1.1172,
      "step": 6069
    },
    {
      "epoch": 4.0198675496688745,
      "grad_norm": 1.1092156739458603,
      "learning_rate": 0.00011430209364032377,
      "loss": 1.0312,
      "step": 6070
    },
    {
      "epoch": 4.020529801324503,
      "grad_norm": 0.9980405443252298,
      "learning_rate": 0.00011423726721462164,
      "loss": 0.9453,
      "step": 6071
    },
    {
      "epoch": 4.021192052980132,
      "grad_norm": 1.050254890740168,
      "learning_rate": 0.00011417244787035066,
      "loss": 0.9688,
      "step": 6072
    },
    {
      "epoch": 4.021854304635761,
      "grad_norm": 0.9388678320829094,
      "learning_rate": 0.00011410763562034565,
      "loss": 0.9102,
      "step": 6073
    },
    {
      "epoch": 4.022516556291391,
      "grad_norm": 1.1209802759420444,
      "learning_rate": 0.00011404283047744036,
      "loss": 0.9297,
      "step": 6074
    },
    {
      "epoch": 4.02317880794702,
      "grad_norm": 0.9147887448046268,
      "learning_rate": 0.00011397803245446686,
      "loss": 0.8164,
      "step": 6075
    },
    {
      "epoch": 4.023841059602649,
      "grad_norm": 1.1873637000071138,
      "learning_rate": 0.00011391324156425588,
      "loss": 1.2266,
      "step": 6076
    },
    {
      "epoch": 4.024503311258278,
      "grad_norm": 1.1242035847406417,
      "learning_rate": 0.00011384845781963681,
      "loss": 1.0625,
      "step": 6077
    },
    {
      "epoch": 4.025165562913907,
      "grad_norm": 0.8701647628634304,
      "learning_rate": 0.0001137836812334376,
      "loss": 0.8086,
      "step": 6078
    },
    {
      "epoch": 4.025827814569537,
      "grad_norm": 1.0469717812158525,
      "learning_rate": 0.0001137189118184846,
      "loss": 0.9297,
      "step": 6079
    },
    {
      "epoch": 4.026490066225166,
      "grad_norm": 1.0041101064210378,
      "learning_rate": 0.00011365414958760312,
      "loss": 0.9258,
      "step": 6080
    },
    {
      "epoch": 4.027152317880795,
      "grad_norm": 0.8938651505809888,
      "learning_rate": 0.00011358939455361668,
      "loss": 0.7188,
      "step": 6081
    },
    {
      "epoch": 4.027814569536424,
      "grad_norm": 1.0322947369784174,
      "learning_rate": 0.00011352464672934752,
      "loss": 0.8984,
      "step": 6082
    },
    {
      "epoch": 4.028476821192053,
      "grad_norm": 1.1054143590771612,
      "learning_rate": 0.00011345990612761652,
      "loss": 0.9766,
      "step": 6083
    },
    {
      "epoch": 4.029139072847682,
      "grad_norm": 1.1125494691704148,
      "learning_rate": 0.00011339517276124305,
      "loss": 1.0391,
      "step": 6084
    },
    {
      "epoch": 4.029801324503311,
      "grad_norm": 1.0355392695858143,
      "learning_rate": 0.00011333044664304496,
      "loss": 0.9766,
      "step": 6085
    },
    {
      "epoch": 4.03046357615894,
      "grad_norm": 0.894740905730587,
      "learning_rate": 0.00011326572778583889,
      "loss": 0.8281,
      "step": 6086
    },
    {
      "epoch": 4.031125827814569,
      "grad_norm": 0.7829140849328088,
      "learning_rate": 0.00011320101620243984,
      "loss": 0.7031,
      "step": 6087
    },
    {
      "epoch": 4.031788079470199,
      "grad_norm": 1.0405278653112255,
      "learning_rate": 0.00011313631190566146,
      "loss": 1.0391,
      "step": 6088
    },
    {
      "epoch": 4.032450331125828,
      "grad_norm": 1.2253934130017077,
      "learning_rate": 0.000113071614908316,
      "loss": 1.1953,
      "step": 6089
    },
    {
      "epoch": 4.033112582781457,
      "grad_norm": 0.8015260583976082,
      "learning_rate": 0.00011300692522321413,
      "loss": 0.707,
      "step": 6090
    },
    {
      "epoch": 4.033774834437086,
      "grad_norm": 0.8143916875866967,
      "learning_rate": 0.00011294224286316516,
      "loss": 0.7031,
      "step": 6091
    },
    {
      "epoch": 4.034437086092716,
      "grad_norm": 1.273912609960614,
      "learning_rate": 0.00011287756784097697,
      "loss": 1.2266,
      "step": 6092
    },
    {
      "epoch": 4.035099337748345,
      "grad_norm": 0.6643608268915918,
      "learning_rate": 0.00011281290016945595,
      "loss": 0.5547,
      "step": 6093
    },
    {
      "epoch": 4.035761589403974,
      "grad_norm": 1.024292407413232,
      "learning_rate": 0.00011274823986140698,
      "loss": 0.875,
      "step": 6094
    },
    {
      "epoch": 4.0364238410596025,
      "grad_norm": 1.1113498589763122,
      "learning_rate": 0.00011268358692963368,
      "loss": 0.9609,
      "step": 6095
    },
    {
      "epoch": 4.0370860927152314,
      "grad_norm": 0.9856726075807472,
      "learning_rate": 0.00011261894138693796,
      "loss": 0.8789,
      "step": 6096
    },
    {
      "epoch": 4.037748344370861,
      "grad_norm": 0.9228205284624255,
      "learning_rate": 0.00011255430324612037,
      "loss": 0.7773,
      "step": 6097
    },
    {
      "epoch": 4.03841059602649,
      "grad_norm": 1.1404135386531435,
      "learning_rate": 0.00011248967251998005,
      "loss": 1.0469,
      "step": 6098
    },
    {
      "epoch": 4.039072847682119,
      "grad_norm": 0.9045183758269307,
      "learning_rate": 0.00011242504922131467,
      "loss": 0.6797,
      "step": 6099
    },
    {
      "epoch": 4.039735099337748,
      "grad_norm": 0.863250279444504,
      "learning_rate": 0.00011236043336292022,
      "loss": 0.6719,
      "step": 6100
    },
    {
      "epoch": 4.040397350993377,
      "grad_norm": 1.1707184880112114,
      "learning_rate": 0.00011229582495759159,
      "loss": 1.0625,
      "step": 6101
    },
    {
      "epoch": 4.041059602649007,
      "grad_norm": 0.8792400476696068,
      "learning_rate": 0.00011223122401812185,
      "loss": 0.7656,
      "step": 6102
    },
    {
      "epoch": 4.041721854304636,
      "grad_norm": 0.9834222264926916,
      "learning_rate": 0.00011216663055730273,
      "loss": 0.7812,
      "step": 6103
    },
    {
      "epoch": 4.042384105960265,
      "grad_norm": 1.0520949822226686,
      "learning_rate": 0.00011210204458792452,
      "loss": 0.8242,
      "step": 6104
    },
    {
      "epoch": 4.043046357615894,
      "grad_norm": 0.9018274966222466,
      "learning_rate": 0.000112037466122776,
      "loss": 0.8203,
      "step": 6105
    },
    {
      "epoch": 4.0437086092715235,
      "grad_norm": 1.1083785234738406,
      "learning_rate": 0.00011197289517464434,
      "loss": 0.9414,
      "step": 6106
    },
    {
      "epoch": 4.0443708609271525,
      "grad_norm": 1.1360670442136473,
      "learning_rate": 0.00011190833175631548,
      "loss": 1.0391,
      "step": 6107
    },
    {
      "epoch": 4.045033112582781,
      "grad_norm": 1.0727601198901249,
      "learning_rate": 0.00011184377588057361,
      "loss": 0.8867,
      "step": 6108
    },
    {
      "epoch": 4.04569536423841,
      "grad_norm": 1.0317899958147387,
      "learning_rate": 0.00011177922756020155,
      "loss": 0.8164,
      "step": 6109
    },
    {
      "epoch": 4.046357615894039,
      "grad_norm": 0.9532874345891553,
      "learning_rate": 0.00011171468680798063,
      "loss": 0.832,
      "step": 6110
    },
    {
      "epoch": 4.047019867549669,
      "grad_norm": 1.2175977993133529,
      "learning_rate": 0.00011165015363669068,
      "loss": 1.125,
      "step": 6111
    },
    {
      "epoch": 4.047682119205298,
      "grad_norm": 1.003430968101119,
      "learning_rate": 0.00011158562805910991,
      "loss": 0.8633,
      "step": 6112
    },
    {
      "epoch": 4.048344370860927,
      "grad_norm": 0.8657745475806908,
      "learning_rate": 0.00011152111008801525,
      "loss": 0.6719,
      "step": 6113
    },
    {
      "epoch": 4.049006622516556,
      "grad_norm": 1.098832170323678,
      "learning_rate": 0.0001114565997361819,
      "loss": 1.0,
      "step": 6114
    },
    {
      "epoch": 4.049668874172186,
      "grad_norm": 0.7862532444176566,
      "learning_rate": 0.00011139209701638369,
      "loss": 0.6094,
      "step": 6115
    },
    {
      "epoch": 4.050331125827815,
      "grad_norm": 1.002899318607169,
      "learning_rate": 0.00011132760194139293,
      "loss": 0.8281,
      "step": 6116
    },
    {
      "epoch": 4.050993377483444,
      "grad_norm": 1.2031492280036507,
      "learning_rate": 0.00011126311452398034,
      "loss": 0.9766,
      "step": 6117
    },
    {
      "epoch": 4.051655629139073,
      "grad_norm": 0.819001014965865,
      "learning_rate": 0.00011119863477691513,
      "loss": 0.6172,
      "step": 6118
    },
    {
      "epoch": 4.052317880794702,
      "grad_norm": 0.9490399510267773,
      "learning_rate": 0.00011113416271296514,
      "loss": 0.7461,
      "step": 6119
    },
    {
      "epoch": 4.052980132450331,
      "grad_norm": 1.1142188873033128,
      "learning_rate": 0.00011106969834489648,
      "loss": 0.918,
      "step": 6120
    },
    {
      "epoch": 4.05364238410596,
      "grad_norm": 1.0055561494429706,
      "learning_rate": 0.00011100524168547385,
      "loss": 0.7617,
      "step": 6121
    },
    {
      "epoch": 4.054304635761589,
      "grad_norm": 0.9863092902918323,
      "learning_rate": 0.00011094079274746048,
      "loss": 0.9023,
      "step": 6122
    },
    {
      "epoch": 4.054966887417218,
      "grad_norm": 0.8008173578446519,
      "learning_rate": 0.00011087635154361792,
      "loss": 0.625,
      "step": 6123
    },
    {
      "epoch": 4.055629139072848,
      "grad_norm": 1.1288406592370104,
      "learning_rate": 0.00011081191808670627,
      "loss": 0.9375,
      "step": 6124
    },
    {
      "epoch": 4.056291390728477,
      "grad_norm": 1.181138573846123,
      "learning_rate": 0.00011074749238948415,
      "loss": 0.9453,
      "step": 6125
    },
    {
      "epoch": 4.056953642384106,
      "grad_norm": 1.1128121527033252,
      "learning_rate": 0.00011068307446470858,
      "loss": 0.9492,
      "step": 6126
    },
    {
      "epoch": 4.057615894039735,
      "grad_norm": 1.2333781639211527,
      "learning_rate": 0.00011061866432513492,
      "loss": 1.0781,
      "step": 6127
    },
    {
      "epoch": 4.058278145695364,
      "grad_norm": 1.140504128823657,
      "learning_rate": 0.00011055426198351732,
      "loss": 0.9609,
      "step": 6128
    },
    {
      "epoch": 4.058940397350994,
      "grad_norm": 1.0578637746407857,
      "learning_rate": 0.00011048986745260803,
      "loss": 0.918,
      "step": 6129
    },
    {
      "epoch": 4.059602649006623,
      "grad_norm": 0.9837589691809756,
      "learning_rate": 0.00011042548074515797,
      "loss": 0.8359,
      "step": 6130
    },
    {
      "epoch": 4.0602649006622515,
      "grad_norm": 1.1316393457165368,
      "learning_rate": 0.00011036110187391639,
      "loss": 1.0469,
      "step": 6131
    },
    {
      "epoch": 4.0609271523178805,
      "grad_norm": 1.0999729518262902,
      "learning_rate": 0.00011029673085163114,
      "loss": 1.0,
      "step": 6132
    },
    {
      "epoch": 4.06158940397351,
      "grad_norm": 1.003516059397668,
      "learning_rate": 0.00011023236769104833,
      "loss": 0.8789,
      "step": 6133
    },
    {
      "epoch": 4.062251655629139,
      "grad_norm": 1.1241101033866776,
      "learning_rate": 0.00011016801240491263,
      "loss": 0.9297,
      "step": 6134
    },
    {
      "epoch": 4.062913907284768,
      "grad_norm": 0.862811325364301,
      "learning_rate": 0.00011010366500596715,
      "loss": 0.7578,
      "step": 6135
    },
    {
      "epoch": 4.063576158940397,
      "grad_norm": 1.0526687590971255,
      "learning_rate": 0.00011003932550695343,
      "loss": 0.918,
      "step": 6136
    },
    {
      "epoch": 4.064238410596026,
      "grad_norm": 1.040565041492152,
      "learning_rate": 0.00010997499392061129,
      "loss": 0.9453,
      "step": 6137
    },
    {
      "epoch": 4.064900662251656,
      "grad_norm": 1.1558093656867783,
      "learning_rate": 0.00010991067025967934,
      "loss": 0.9648,
      "step": 6138
    },
    {
      "epoch": 4.065562913907285,
      "grad_norm": 1.0914902805465116,
      "learning_rate": 0.00010984635453689423,
      "loss": 0.9102,
      "step": 6139
    },
    {
      "epoch": 4.066225165562914,
      "grad_norm": 0.9802797646791277,
      "learning_rate": 0.00010978204676499125,
      "loss": 0.8359,
      "step": 6140
    },
    {
      "epoch": 4.066887417218543,
      "grad_norm": 1.0352682665193418,
      "learning_rate": 0.00010971774695670414,
      "loss": 0.8398,
      "step": 6141
    },
    {
      "epoch": 4.067549668874173,
      "grad_norm": 1.0816479971775392,
      "learning_rate": 0.00010965345512476497,
      "loss": 0.8516,
      "step": 6142
    },
    {
      "epoch": 4.0682119205298015,
      "grad_norm": 1.1738892439614412,
      "learning_rate": 0.00010958917128190417,
      "loss": 0.9727,
      "step": 6143
    },
    {
      "epoch": 4.0688741721854305,
      "grad_norm": 1.190389227837336,
      "learning_rate": 0.0001095248954408508,
      "loss": 0.9492,
      "step": 6144
    },
    {
      "epoch": 4.069536423841059,
      "grad_norm": 1.164894174868382,
      "learning_rate": 0.00010946062761433217,
      "loss": 1.0703,
      "step": 6145
    },
    {
      "epoch": 4.070198675496688,
      "grad_norm": 0.8320490788695099,
      "learning_rate": 0.00010939636781507397,
      "loss": 0.6797,
      "step": 6146
    },
    {
      "epoch": 4.070860927152318,
      "grad_norm": 1.1075848503818417,
      "learning_rate": 0.00010933211605580052,
      "loss": 0.9219,
      "step": 6147
    },
    {
      "epoch": 4.071523178807947,
      "grad_norm": 1.0844108160278423,
      "learning_rate": 0.00010926787234923435,
      "loss": 1.0781,
      "step": 6148
    },
    {
      "epoch": 4.072185430463576,
      "grad_norm": 0.9548600743564585,
      "learning_rate": 0.00010920363670809637,
      "loss": 0.7695,
      "step": 6149
    },
    {
      "epoch": 4.072847682119205,
      "grad_norm": 1.0922439035925138,
      "learning_rate": 0.00010913940914510609,
      "loss": 0.9023,
      "step": 6150
    },
    {
      "epoch": 4.073509933774835,
      "grad_norm": 1.1207046484323397,
      "learning_rate": 0.00010907518967298123,
      "loss": 1.0,
      "step": 6151
    },
    {
      "epoch": 4.074172185430464,
      "grad_norm": 1.1158815742620745,
      "learning_rate": 0.00010901097830443796,
      "loss": 0.9297,
      "step": 6152
    },
    {
      "epoch": 4.074834437086093,
      "grad_norm": 0.9304070896365855,
      "learning_rate": 0.00010894677505219099,
      "loss": 0.7305,
      "step": 6153
    },
    {
      "epoch": 4.075496688741722,
      "grad_norm": 0.9282693787096283,
      "learning_rate": 0.00010888257992895317,
      "loss": 0.7734,
      "step": 6154
    },
    {
      "epoch": 4.076158940397351,
      "grad_norm": 1.1720616606525114,
      "learning_rate": 0.0001088183929474359,
      "loss": 1.0391,
      "step": 6155
    },
    {
      "epoch": 4.07682119205298,
      "grad_norm": 1.1520419968611935,
      "learning_rate": 0.000108754214120349,
      "loss": 0.9414,
      "step": 6156
    },
    {
      "epoch": 4.077483443708609,
      "grad_norm": 0.9666933805090103,
      "learning_rate": 0.00010869004346040055,
      "loss": 0.8008,
      "step": 6157
    },
    {
      "epoch": 4.078145695364238,
      "grad_norm": 1.0378605322634982,
      "learning_rate": 0.00010862588098029709,
      "loss": 0.8242,
      "step": 6158
    },
    {
      "epoch": 4.078807947019867,
      "grad_norm": 0.9499468543678622,
      "learning_rate": 0.00010856172669274358,
      "loss": 0.8438,
      "step": 6159
    },
    {
      "epoch": 4.079470198675497,
      "grad_norm": 1.2171483290736038,
      "learning_rate": 0.00010849758061044323,
      "loss": 1.0703,
      "step": 6160
    },
    {
      "epoch": 4.080132450331126,
      "grad_norm": 1.0604980876986472,
      "learning_rate": 0.00010843344274609772,
      "loss": 0.8555,
      "step": 6161
    },
    {
      "epoch": 4.080794701986755,
      "grad_norm": 0.9375595359931704,
      "learning_rate": 0.0001083693131124071,
      "loss": 0.7266,
      "step": 6162
    },
    {
      "epoch": 4.081456953642384,
      "grad_norm": 1.1731259180404756,
      "learning_rate": 0.00010830519172206981,
      "loss": 1.0234,
      "step": 6163
    },
    {
      "epoch": 4.082119205298013,
      "grad_norm": 1.052047635000336,
      "learning_rate": 0.00010824107858778245,
      "loss": 0.8438,
      "step": 6164
    },
    {
      "epoch": 4.082781456953643,
      "grad_norm": 1.0203960908284417,
      "learning_rate": 0.00010817697372224041,
      "loss": 0.7852,
      "step": 6165
    },
    {
      "epoch": 4.083443708609272,
      "grad_norm": 0.9726764675584749,
      "learning_rate": 0.00010811287713813701,
      "loss": 0.7266,
      "step": 6166
    },
    {
      "epoch": 4.084105960264901,
      "grad_norm": 0.9202087953719736,
      "learning_rate": 0.00010804878884816412,
      "loss": 0.7773,
      "step": 6167
    },
    {
      "epoch": 4.0847682119205295,
      "grad_norm": 1.003795981323917,
      "learning_rate": 0.00010798470886501203,
      "loss": 0.7461,
      "step": 6168
    },
    {
      "epoch": 4.085430463576159,
      "grad_norm": 0.994724144978067,
      "learning_rate": 0.0001079206372013693,
      "loss": 0.8125,
      "step": 6169
    },
    {
      "epoch": 4.086092715231788,
      "grad_norm": 1.1484884308538337,
      "learning_rate": 0.00010785657386992279,
      "loss": 0.9414,
      "step": 6170
    },
    {
      "epoch": 4.086754966887417,
      "grad_norm": 0.9676172352162731,
      "learning_rate": 0.00010779251888335782,
      "loss": 0.7852,
      "step": 6171
    },
    {
      "epoch": 4.087417218543046,
      "grad_norm": 0.9960351981894656,
      "learning_rate": 0.00010772847225435801,
      "loss": 0.7852,
      "step": 6172
    },
    {
      "epoch": 4.088079470198675,
      "grad_norm": 1.2089386000553126,
      "learning_rate": 0.00010766443399560529,
      "loss": 0.9961,
      "step": 6173
    },
    {
      "epoch": 4.088741721854305,
      "grad_norm": 0.9875402351823904,
      "learning_rate": 0.00010760040411978004,
      "loss": 0.8398,
      "step": 6174
    },
    {
      "epoch": 4.089403973509934,
      "grad_norm": 1.1503639680417312,
      "learning_rate": 0.00010753638263956089,
      "loss": 0.957,
      "step": 6175
    },
    {
      "epoch": 4.090066225165563,
      "grad_norm": 1.1444604033106045,
      "learning_rate": 0.00010747236956762473,
      "loss": 0.957,
      "step": 6176
    },
    {
      "epoch": 4.090728476821192,
      "grad_norm": 0.9620191937312393,
      "learning_rate": 0.000107408364916647,
      "loss": 0.8125,
      "step": 6177
    },
    {
      "epoch": 4.091390728476822,
      "grad_norm": 0.9050149486915279,
      "learning_rate": 0.00010734436869930134,
      "loss": 0.7188,
      "step": 6178
    },
    {
      "epoch": 4.092052980132451,
      "grad_norm": 1.013704251816913,
      "learning_rate": 0.00010728038092825967,
      "loss": 0.8398,
      "step": 6179
    },
    {
      "epoch": 4.0927152317880795,
      "grad_norm": 0.9317626812105108,
      "learning_rate": 0.00010721640161619238,
      "loss": 0.8398,
      "step": 6180
    },
    {
      "epoch": 4.093377483443708,
      "grad_norm": 1.1083403838074983,
      "learning_rate": 0.00010715243077576805,
      "loss": 0.9023,
      "step": 6181
    },
    {
      "epoch": 4.094039735099337,
      "grad_norm": 0.8991871627697058,
      "learning_rate": 0.00010708846841965363,
      "loss": 0.7422,
      "step": 6182
    },
    {
      "epoch": 4.094701986754967,
      "grad_norm": 1.0569232401961193,
      "learning_rate": 0.00010702451456051445,
      "loss": 0.8086,
      "step": 6183
    },
    {
      "epoch": 4.095364238410596,
      "grad_norm": 0.9293558481989543,
      "learning_rate": 0.00010696056921101409,
      "loss": 0.668,
      "step": 6184
    },
    {
      "epoch": 4.096026490066225,
      "grad_norm": 1.0967084173775539,
      "learning_rate": 0.00010689663238381448,
      "loss": 0.8945,
      "step": 6185
    },
    {
      "epoch": 4.096688741721854,
      "grad_norm": 1.2070857534071646,
      "learning_rate": 0.00010683270409157577,
      "loss": 0.8945,
      "step": 6186
    },
    {
      "epoch": 4.097350993377484,
      "grad_norm": 0.9638353882112738,
      "learning_rate": 0.00010676878434695656,
      "loss": 0.7266,
      "step": 6187
    },
    {
      "epoch": 4.098013245033113,
      "grad_norm": 1.13338973242338,
      "learning_rate": 0.0001067048731626137,
      "loss": 0.8242,
      "step": 6188
    },
    {
      "epoch": 4.098675496688742,
      "grad_norm": 1.0361474108741482,
      "learning_rate": 0.00010664097055120228,
      "loss": 0.7461,
      "step": 6189
    },
    {
      "epoch": 4.099337748344371,
      "grad_norm": 1.19580275207176,
      "learning_rate": 0.00010657707652537585,
      "loss": 0.9961,
      "step": 6190
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.1300805453008844,
      "learning_rate": 0.00010651319109778606,
      "loss": 0.9258,
      "step": 6191
    },
    {
      "epoch": 4.1006622516556295,
      "grad_norm": 1.1577862467213746,
      "learning_rate": 0.00010644931428108295,
      "loss": 1.0234,
      "step": 6192
    },
    {
      "epoch": 4.101324503311258,
      "grad_norm": 1.121482014852132,
      "learning_rate": 0.00010638544608791496,
      "loss": 0.9102,
      "step": 6193
    },
    {
      "epoch": 4.101986754966887,
      "grad_norm": 1.11831185126957,
      "learning_rate": 0.00010632158653092868,
      "loss": 1.0,
      "step": 6194
    },
    {
      "epoch": 4.102649006622516,
      "grad_norm": 1.2356005011263262,
      "learning_rate": 0.00010625773562276894,
      "loss": 1.0547,
      "step": 6195
    },
    {
      "epoch": 4.103311258278145,
      "grad_norm": 0.9628279953061157,
      "learning_rate": 0.00010619389337607912,
      "loss": 0.7969,
      "step": 6196
    },
    {
      "epoch": 4.103973509933775,
      "grad_norm": 1.032529703126546,
      "learning_rate": 0.00010613005980350061,
      "loss": 0.7852,
      "step": 6197
    },
    {
      "epoch": 4.104635761589404,
      "grad_norm": 1.2501592861588524,
      "learning_rate": 0.00010606623491767319,
      "loss": 1.0469,
      "step": 6198
    },
    {
      "epoch": 4.105298013245033,
      "grad_norm": 0.8417599479593245,
      "learning_rate": 0.00010600241873123495,
      "loss": 0.6055,
      "step": 6199
    },
    {
      "epoch": 4.105960264900662,
      "grad_norm": 1.0220351338588083,
      "learning_rate": 0.00010593861125682226,
      "loss": 0.8008,
      "step": 6200
    },
    {
      "epoch": 4.106622516556292,
      "grad_norm": 1.2194440204982333,
      "learning_rate": 0.00010587481250706961,
      "loss": 1.1484,
      "step": 6201
    },
    {
      "epoch": 4.107284768211921,
      "grad_norm": 1.102084334508904,
      "learning_rate": 0.00010581102249461002,
      "loss": 0.9688,
      "step": 6202
    },
    {
      "epoch": 4.10794701986755,
      "grad_norm": 1.2041159758137538,
      "learning_rate": 0.00010574724123207457,
      "loss": 1.0234,
      "step": 6203
    },
    {
      "epoch": 4.108609271523179,
      "grad_norm": 0.9921530871313249,
      "learning_rate": 0.00010568346873209269,
      "loss": 0.8438,
      "step": 6204
    },
    {
      "epoch": 4.109271523178808,
      "grad_norm": 1.0154330240935314,
      "learning_rate": 0.0001056197050072921,
      "loss": 0.7852,
      "step": 6205
    },
    {
      "epoch": 4.109933774834437,
      "grad_norm": 1.1583271974980496,
      "learning_rate": 0.00010555595007029872,
      "loss": 0.9688,
      "step": 6206
    },
    {
      "epoch": 4.110596026490066,
      "grad_norm": 0.9789733912654255,
      "learning_rate": 0.0001054922039337367,
      "loss": 0.7578,
      "step": 6207
    },
    {
      "epoch": 4.111258278145695,
      "grad_norm": 1.1754808439068591,
      "learning_rate": 0.00010542846661022863,
      "loss": 1.0234,
      "step": 6208
    },
    {
      "epoch": 4.111920529801324,
      "grad_norm": 0.9889619008802039,
      "learning_rate": 0.00010536473811239514,
      "loss": 0.7812,
      "step": 6209
    },
    {
      "epoch": 4.112582781456954,
      "grad_norm": 0.9091287209670218,
      "learning_rate": 0.00010530101845285522,
      "loss": 0.7227,
      "step": 6210
    },
    {
      "epoch": 4.113245033112583,
      "grad_norm": 1.3688291205661167,
      "learning_rate": 0.00010523730764422613,
      "loss": 1.1719,
      "step": 6211
    },
    {
      "epoch": 4.113907284768212,
      "grad_norm": 0.9980378099575175,
      "learning_rate": 0.00010517360569912335,
      "loss": 0.7617,
      "step": 6212
    },
    {
      "epoch": 4.114569536423841,
      "grad_norm": 0.9175836566535053,
      "learning_rate": 0.00010510991263016048,
      "loss": 0.8125,
      "step": 6213
    },
    {
      "epoch": 4.11523178807947,
      "grad_norm": 1.0516569556209148,
      "learning_rate": 0.00010504622844994957,
      "loss": 0.8086,
      "step": 6214
    },
    {
      "epoch": 4.1158940397351,
      "grad_norm": 1.0365959501155582,
      "learning_rate": 0.00010498255317110086,
      "loss": 0.7891,
      "step": 6215
    },
    {
      "epoch": 4.1165562913907285,
      "grad_norm": 1.15720381907385,
      "learning_rate": 0.00010491888680622267,
      "loss": 1.0156,
      "step": 6216
    },
    {
      "epoch": 4.1172185430463575,
      "grad_norm": 1.199408812094867,
      "learning_rate": 0.00010485522936792179,
      "loss": 0.9609,
      "step": 6217
    },
    {
      "epoch": 4.117880794701986,
      "grad_norm": 1.1599399951107454,
      "learning_rate": 0.00010479158086880304,
      "loss": 0.9766,
      "step": 6218
    },
    {
      "epoch": 4.118543046357616,
      "grad_norm": 1.1622324165112725,
      "learning_rate": 0.00010472794132146953,
      "loss": 1.0,
      "step": 6219
    },
    {
      "epoch": 4.119205298013245,
      "grad_norm": 0.8778188781542734,
      "learning_rate": 0.0001046643107385227,
      "loss": 0.6719,
      "step": 6220
    },
    {
      "epoch": 4.119867549668874,
      "grad_norm": 1.216221303151822,
      "learning_rate": 0.00010460068913256214,
      "loss": 1.0,
      "step": 6221
    },
    {
      "epoch": 4.120529801324503,
      "grad_norm": 1.172145430794656,
      "learning_rate": 0.00010453707651618552,
      "loss": 0.9648,
      "step": 6222
    },
    {
      "epoch": 4.121192052980132,
      "grad_norm": 0.9187557537192379,
      "learning_rate": 0.00010447347290198907,
      "loss": 0.6875,
      "step": 6223
    },
    {
      "epoch": 4.121854304635762,
      "grad_norm": 0.9386624881605963,
      "learning_rate": 0.00010440987830256689,
      "loss": 0.7617,
      "step": 6224
    },
    {
      "epoch": 4.122516556291391,
      "grad_norm": 1.1798476886717546,
      "learning_rate": 0.00010434629273051143,
      "loss": 0.9766,
      "step": 6225
    },
    {
      "epoch": 4.12317880794702,
      "grad_norm": 0.8959804762812561,
      "learning_rate": 0.00010428271619841347,
      "loss": 0.6523,
      "step": 6226
    },
    {
      "epoch": 4.123841059602649,
      "grad_norm": 0.9649395788065253,
      "learning_rate": 0.00010421914871886185,
      "loss": 0.7969,
      "step": 6227
    },
    {
      "epoch": 4.1245033112582785,
      "grad_norm": 1.02846148805379,
      "learning_rate": 0.00010415559030444357,
      "loss": 0.7422,
      "step": 6228
    },
    {
      "epoch": 4.1251655629139075,
      "grad_norm": 0.9720999574885072,
      "learning_rate": 0.00010409204096774407,
      "loss": 0.7227,
      "step": 6229
    },
    {
      "epoch": 4.125827814569536,
      "grad_norm": 1.1522091483808263,
      "learning_rate": 0.00010402850072134676,
      "loss": 0.9219,
      "step": 6230
    },
    {
      "epoch": 4.126490066225165,
      "grad_norm": 1.1068860264224933,
      "learning_rate": 0.00010396496957783331,
      "loss": 0.8438,
      "step": 6231
    },
    {
      "epoch": 4.127152317880794,
      "grad_norm": 1.1027702806537238,
      "learning_rate": 0.0001039014475497837,
      "loss": 0.8984,
      "step": 6232
    },
    {
      "epoch": 4.127814569536424,
      "grad_norm": 1.149804371153029,
      "learning_rate": 0.000103837934649776,
      "loss": 0.8633,
      "step": 6233
    },
    {
      "epoch": 4.128476821192053,
      "grad_norm": 1.2400737285732375,
      "learning_rate": 0.00010377443089038641,
      "loss": 0.9648,
      "step": 6234
    },
    {
      "epoch": 4.129139072847682,
      "grad_norm": 1.2723883363486652,
      "learning_rate": 0.00010371093628418951,
      "loss": 1.1016,
      "step": 6235
    },
    {
      "epoch": 4.129801324503311,
      "grad_norm": 1.1050848093090886,
      "learning_rate": 0.0001036474508437579,
      "loss": 0.8086,
      "step": 6236
    },
    {
      "epoch": 4.130463576158941,
      "grad_norm": 1.1280700179925658,
      "learning_rate": 0.0001035839745816624,
      "loss": 0.9023,
      "step": 6237
    },
    {
      "epoch": 4.13112582781457,
      "grad_norm": 1.0943141206115414,
      "learning_rate": 0.0001035205075104721,
      "loss": 0.7891,
      "step": 6238
    },
    {
      "epoch": 4.131788079470199,
      "grad_norm": 0.9567971820617802,
      "learning_rate": 0.00010345704964275423,
      "loss": 0.7148,
      "step": 6239
    },
    {
      "epoch": 4.132450331125828,
      "grad_norm": 1.1528538727358535,
      "learning_rate": 0.00010339360099107405,
      "loss": 1.0078,
      "step": 6240
    },
    {
      "epoch": 4.1331125827814565,
      "grad_norm": 1.1189192638773982,
      "learning_rate": 0.00010333016156799524,
      "loss": 0.9375,
      "step": 6241
    },
    {
      "epoch": 4.133774834437086,
      "grad_norm": 1.1216563973328242,
      "learning_rate": 0.00010326673138607952,
      "loss": 0.8555,
      "step": 6242
    },
    {
      "epoch": 4.134437086092715,
      "grad_norm": 1.1434124835532453,
      "learning_rate": 0.00010320331045788676,
      "loss": 0.9727,
      "step": 6243
    },
    {
      "epoch": 4.135099337748344,
      "grad_norm": 1.0984156454885794,
      "learning_rate": 0.00010313989879597497,
      "loss": 0.9336,
      "step": 6244
    },
    {
      "epoch": 4.135761589403973,
      "grad_norm": 0.9002061013936019,
      "learning_rate": 0.00010307649641290049,
      "loss": 0.6992,
      "step": 6245
    },
    {
      "epoch": 4.136423841059603,
      "grad_norm": 0.9842071533064782,
      "learning_rate": 0.00010301310332121771,
      "loss": 0.7812,
      "step": 6246
    },
    {
      "epoch": 4.137086092715232,
      "grad_norm": 1.2086862486439607,
      "learning_rate": 0.0001029497195334791,
      "loss": 0.9648,
      "step": 6247
    },
    {
      "epoch": 4.137748344370861,
      "grad_norm": 0.8534806547668771,
      "learning_rate": 0.00010288634506223549,
      "loss": 0.6406,
      "step": 6248
    },
    {
      "epoch": 4.13841059602649,
      "grad_norm": 0.9911694586844878,
      "learning_rate": 0.00010282297992003569,
      "loss": 0.7734,
      "step": 6249
    },
    {
      "epoch": 4.139072847682119,
      "grad_norm": 1.1491288452083723,
      "learning_rate": 0.0001027596241194267,
      "loss": 0.9609,
      "step": 6250
    },
    {
      "epoch": 4.139735099337749,
      "grad_norm": 0.983134839466708,
      "learning_rate": 0.00010269627767295374,
      "loss": 0.7305,
      "step": 6251
    },
    {
      "epoch": 4.140397350993378,
      "grad_norm": 1.0738283286572465,
      "learning_rate": 0.00010263294059316012,
      "loss": 0.8398,
      "step": 6252
    },
    {
      "epoch": 4.1410596026490065,
      "grad_norm": 1.0088453490717972,
      "learning_rate": 0.0001025696128925873,
      "loss": 0.7969,
      "step": 6253
    },
    {
      "epoch": 4.1417218543046355,
      "grad_norm": 0.9287928479677533,
      "learning_rate": 0.00010250629458377493,
      "loss": 0.7383,
      "step": 6254
    },
    {
      "epoch": 4.142384105960265,
      "grad_norm": 1.1301020691322565,
      "learning_rate": 0.00010244298567926073,
      "loss": 0.8594,
      "step": 6255
    },
    {
      "epoch": 4.143046357615894,
      "grad_norm": 1.1209471438878709,
      "learning_rate": 0.00010237968619158056,
      "loss": 0.8594,
      "step": 6256
    },
    {
      "epoch": 4.143708609271523,
      "grad_norm": 1.1734671077347993,
      "learning_rate": 0.00010231639613326847,
      "loss": 0.9375,
      "step": 6257
    },
    {
      "epoch": 4.144370860927152,
      "grad_norm": 0.9648746114153576,
      "learning_rate": 0.00010225311551685669,
      "loss": 0.6836,
      "step": 6258
    },
    {
      "epoch": 4.145033112582781,
      "grad_norm": 1.0811551504621864,
      "learning_rate": 0.0001021898443548753,
      "loss": 0.8203,
      "step": 6259
    },
    {
      "epoch": 4.145695364238411,
      "grad_norm": 1.0829789277214377,
      "learning_rate": 0.00010212658265985299,
      "loss": 0.8711,
      "step": 6260
    },
    {
      "epoch": 4.14635761589404,
      "grad_norm": 1.1097007770091567,
      "learning_rate": 0.0001020633304443161,
      "loss": 0.9258,
      "step": 6261
    },
    {
      "epoch": 4.147019867549669,
      "grad_norm": 0.9239493341310807,
      "learning_rate": 0.00010200008772078934,
      "loss": 0.6953,
      "step": 6262
    },
    {
      "epoch": 4.147682119205298,
      "grad_norm": 1.031851746059729,
      "learning_rate": 0.00010193685450179556,
      "loss": 0.793,
      "step": 6263
    },
    {
      "epoch": 4.1483443708609276,
      "grad_norm": 1.174899506033891,
      "learning_rate": 0.00010187363079985563,
      "loss": 0.9648,
      "step": 6264
    },
    {
      "epoch": 4.1490066225165565,
      "grad_norm": 1.1427609768080544,
      "learning_rate": 0.00010181041662748847,
      "loss": 1.0,
      "step": 6265
    },
    {
      "epoch": 4.149668874172185,
      "grad_norm": 0.9268195929719053,
      "learning_rate": 0.00010174721199721137,
      "loss": 0.7578,
      "step": 6266
    },
    {
      "epoch": 4.150331125827814,
      "grad_norm": 1.249271326803708,
      "learning_rate": 0.00010168401692153948,
      "loss": 1.0391,
      "step": 6267
    },
    {
      "epoch": 4.150993377483443,
      "grad_norm": 1.2507333029392143,
      "learning_rate": 0.00010162083141298611,
      "loss": 1.0,
      "step": 6268
    },
    {
      "epoch": 4.151655629139073,
      "grad_norm": 1.1300147134367888,
      "learning_rate": 0.00010155765548406282,
      "loss": 0.8984,
      "step": 6269
    },
    {
      "epoch": 4.152317880794702,
      "grad_norm": 1.2434929176061937,
      "learning_rate": 0.00010149448914727915,
      "loss": 1.0156,
      "step": 6270
    },
    {
      "epoch": 4.152980132450331,
      "grad_norm": 0.9021352461958396,
      "learning_rate": 0.00010143133241514267,
      "loss": 0.7969,
      "step": 6271
    },
    {
      "epoch": 4.15364238410596,
      "grad_norm": 1.2346634641243102,
      "learning_rate": 0.00010136818530015919,
      "loss": 1.0547,
      "step": 6272
    },
    {
      "epoch": 4.15430463576159,
      "grad_norm": 1.079344373981255,
      "learning_rate": 0.00010130504781483259,
      "loss": 0.9023,
      "step": 6273
    },
    {
      "epoch": 4.154966887417219,
      "grad_norm": 0.8907119629602627,
      "learning_rate": 0.00010124191997166475,
      "loss": 0.6133,
      "step": 6274
    },
    {
      "epoch": 4.155629139072848,
      "grad_norm": 1.2151958161001144,
      "learning_rate": 0.00010117880178315583,
      "loss": 1.0703,
      "step": 6275
    },
    {
      "epoch": 4.156291390728477,
      "grad_norm": 0.9244592713523764,
      "learning_rate": 0.00010111569326180385,
      "loss": 0.7188,
      "step": 6276
    },
    {
      "epoch": 4.156953642384106,
      "grad_norm": 1.1667763990718285,
      "learning_rate": 0.00010105259442010497,
      "loss": 1.0547,
      "step": 6277
    },
    {
      "epoch": 4.157615894039735,
      "grad_norm": 0.9052862583549124,
      "learning_rate": 0.00010098950527055362,
      "loss": 0.6836,
      "step": 6278
    },
    {
      "epoch": 4.158278145695364,
      "grad_norm": 1.1897168742282023,
      "learning_rate": 0.00010092642582564212,
      "loss": 0.9336,
      "step": 6279
    },
    {
      "epoch": 4.158940397350993,
      "grad_norm": 1.1915588558432273,
      "learning_rate": 0.00010086335609786086,
      "loss": 0.957,
      "step": 6280
    },
    {
      "epoch": 4.159602649006622,
      "grad_norm": 1.2391416344488286,
      "learning_rate": 0.00010080029609969851,
      "loss": 1.1094,
      "step": 6281
    },
    {
      "epoch": 4.160264900662252,
      "grad_norm": 1.2770166828212963,
      "learning_rate": 0.00010073724584364154,
      "loss": 1.0312,
      "step": 6282
    },
    {
      "epoch": 4.160927152317881,
      "grad_norm": 1.158290914590156,
      "learning_rate": 0.00010067420534217465,
      "loss": 0.9219,
      "step": 6283
    },
    {
      "epoch": 4.16158940397351,
      "grad_norm": 1.387910325986089,
      "learning_rate": 0.00010061117460778064,
      "loss": 1.1797,
      "step": 6284
    },
    {
      "epoch": 4.162251655629139,
      "grad_norm": 1.1790062732380073,
      "learning_rate": 0.00010054815365294032,
      "loss": 0.9492,
      "step": 6285
    },
    {
      "epoch": 4.162913907284768,
      "grad_norm": 0.9696999785500117,
      "learning_rate": 0.0001004851424901324,
      "loss": 0.6641,
      "step": 6286
    },
    {
      "epoch": 4.163576158940398,
      "grad_norm": 1.1182208162656357,
      "learning_rate": 0.00010042214113183405,
      "loss": 0.9336,
      "step": 6287
    },
    {
      "epoch": 4.164238410596027,
      "grad_norm": 1.2817489650473708,
      "learning_rate": 0.00010035914959052013,
      "loss": 1.1328,
      "step": 6288
    },
    {
      "epoch": 4.164900662251656,
      "grad_norm": 1.2236976878516763,
      "learning_rate": 0.00010029616787866365,
      "loss": 1.0078,
      "step": 6289
    },
    {
      "epoch": 4.1655629139072845,
      "grad_norm": 0.9817315841573214,
      "learning_rate": 0.0001002331960087358,
      "loss": 0.7227,
      "step": 6290
    },
    {
      "epoch": 4.166225165562914,
      "grad_norm": 1.0963752299713079,
      "learning_rate": 0.00010017023399320575,
      "loss": 0.875,
      "step": 6291
    },
    {
      "epoch": 4.166887417218543,
      "grad_norm": 1.0281943631886468,
      "learning_rate": 0.00010010728184454058,
      "loss": 0.8203,
      "step": 6292
    },
    {
      "epoch": 4.167549668874172,
      "grad_norm": 1.1157241205484807,
      "learning_rate": 0.00010004433957520561,
      "loss": 0.8906,
      "step": 6293
    },
    {
      "epoch": 4.168211920529801,
      "grad_norm": 1.064009821886658,
      "learning_rate": 9.998140719766417e-05,
      "loss": 0.8906,
      "step": 6294
    },
    {
      "epoch": 4.16887417218543,
      "grad_norm": 1.0426246555872825,
      "learning_rate": 9.991848472437749e-05,
      "loss": 0.7891,
      "step": 6295
    },
    {
      "epoch": 4.16953642384106,
      "grad_norm": 1.0609059648050276,
      "learning_rate": 9.985557216780502e-05,
      "loss": 0.8047,
      "step": 6296
    },
    {
      "epoch": 4.170198675496689,
      "grad_norm": 1.2392519524864485,
      "learning_rate": 9.97926695404042e-05,
      "loss": 1.0078,
      "step": 6297
    },
    {
      "epoch": 4.170860927152318,
      "grad_norm": 1.1414525468647627,
      "learning_rate": 9.972977685463035e-05,
      "loss": 0.9258,
      "step": 6298
    },
    {
      "epoch": 4.171523178807947,
      "grad_norm": 0.9348721484046824,
      "learning_rate": 9.966689412293706e-05,
      "loss": 0.625,
      "step": 6299
    },
    {
      "epoch": 4.172185430463577,
      "grad_norm": 1.0877397281266818,
      "learning_rate": 9.960402135777575e-05,
      "loss": 0.8086,
      "step": 6300
    },
    {
      "epoch": 4.1728476821192055,
      "grad_norm": 1.0346749464108425,
      "learning_rate": 9.954115857159604e-05,
      "loss": 0.8008,
      "step": 6301
    },
    {
      "epoch": 4.1735099337748345,
      "grad_norm": 1.0076580589762194,
      "learning_rate": 9.947830577684536e-05,
      "loss": 0.8047,
      "step": 6302
    },
    {
      "epoch": 4.174172185430463,
      "grad_norm": 1.1231548582329725,
      "learning_rate": 9.941546298596937e-05,
      "loss": 0.8789,
      "step": 6303
    },
    {
      "epoch": 4.174834437086092,
      "grad_norm": 0.9683365099245336,
      "learning_rate": 9.935263021141164e-05,
      "loss": 0.7227,
      "step": 6304
    },
    {
      "epoch": 4.175496688741722,
      "grad_norm": 0.9726749254564172,
      "learning_rate": 9.928980746561376e-05,
      "loss": 0.668,
      "step": 6305
    },
    {
      "epoch": 4.176158940397351,
      "grad_norm": 1.2424340084579304,
      "learning_rate": 9.922699476101541e-05,
      "loss": 1.0469,
      "step": 6306
    },
    {
      "epoch": 4.17682119205298,
      "grad_norm": 0.9880361068133409,
      "learning_rate": 9.916419211005422e-05,
      "loss": 0.7656,
      "step": 6307
    },
    {
      "epoch": 4.177483443708609,
      "grad_norm": 0.9433731696873005,
      "learning_rate": 9.910139952516573e-05,
      "loss": 0.7031,
      "step": 6308
    },
    {
      "epoch": 4.178145695364238,
      "grad_norm": 0.9009293169216358,
      "learning_rate": 9.903861701878369e-05,
      "loss": 0.7031,
      "step": 6309
    },
    {
      "epoch": 4.178807947019868,
      "grad_norm": 1.1414129875819874,
      "learning_rate": 9.897584460333976e-05,
      "loss": 0.8828,
      "step": 6310
    },
    {
      "epoch": 4.179470198675497,
      "grad_norm": 1.0062374265018146,
      "learning_rate": 9.891308229126355e-05,
      "loss": 0.7109,
      "step": 6311
    },
    {
      "epoch": 4.180132450331126,
      "grad_norm": 1.1162047658478456,
      "learning_rate": 9.885033009498278e-05,
      "loss": 0.8984,
      "step": 6312
    },
    {
      "epoch": 4.180794701986755,
      "grad_norm": 1.0102415149265176,
      "learning_rate": 9.878758802692303e-05,
      "loss": 0.8203,
      "step": 6313
    },
    {
      "epoch": 4.1814569536423845,
      "grad_norm": 1.3311752437753013,
      "learning_rate": 9.872485609950796e-05,
      "loss": 1.1094,
      "step": 6314
    },
    {
      "epoch": 4.182119205298013,
      "grad_norm": 0.9435009278926006,
      "learning_rate": 9.866213432515924e-05,
      "loss": 0.668,
      "step": 6315
    },
    {
      "epoch": 4.182781456953642,
      "grad_norm": 1.113166157839667,
      "learning_rate": 9.859942271629655e-05,
      "loss": 0.9961,
      "step": 6316
    },
    {
      "epoch": 4.183443708609271,
      "grad_norm": 1.0895652724239613,
      "learning_rate": 9.853672128533734e-05,
      "loss": 0.8672,
      "step": 6317
    },
    {
      "epoch": 4.184105960264901,
      "grad_norm": 1.167106363083232,
      "learning_rate": 9.847403004469742e-05,
      "loss": 0.9883,
      "step": 6318
    },
    {
      "epoch": 4.18476821192053,
      "grad_norm": 1.0912021191359709,
      "learning_rate": 9.841134900679025e-05,
      "loss": 0.9062,
      "step": 6319
    },
    {
      "epoch": 4.185430463576159,
      "grad_norm": 1.049493579459763,
      "learning_rate": 9.834867818402741e-05,
      "loss": 0.8398,
      "step": 6320
    },
    {
      "epoch": 4.186092715231788,
      "grad_norm": 1.3235529551613285,
      "learning_rate": 9.828601758881846e-05,
      "loss": 1.0781,
      "step": 6321
    },
    {
      "epoch": 4.186754966887417,
      "grad_norm": 0.9063972330932647,
      "learning_rate": 9.822336723357094e-05,
      "loss": 0.6797,
      "step": 6322
    },
    {
      "epoch": 4.187417218543047,
      "grad_norm": 1.2421613198266561,
      "learning_rate": 9.816072713069024e-05,
      "loss": 1.0312,
      "step": 6323
    },
    {
      "epoch": 4.188079470198676,
      "grad_norm": 1.204355571800546,
      "learning_rate": 9.809809729257994e-05,
      "loss": 0.9727,
      "step": 6324
    },
    {
      "epoch": 4.188741721854305,
      "grad_norm": 0.9782765307011785,
      "learning_rate": 9.80354777316414e-05,
      "loss": 0.7812,
      "step": 6325
    },
    {
      "epoch": 4.1894039735099335,
      "grad_norm": 0.9739953174615141,
      "learning_rate": 9.7972868460274e-05,
      "loss": 0.7578,
      "step": 6326
    },
    {
      "epoch": 4.1900662251655625,
      "grad_norm": 1.264720229956241,
      "learning_rate": 9.791026949087512e-05,
      "loss": 0.9961,
      "step": 6327
    },
    {
      "epoch": 4.190728476821192,
      "grad_norm": 0.9283017228927886,
      "learning_rate": 9.784768083584013e-05,
      "loss": 0.7422,
      "step": 6328
    },
    {
      "epoch": 4.191390728476821,
      "grad_norm": 1.0598234533667843,
      "learning_rate": 9.778510250756217e-05,
      "loss": 0.8828,
      "step": 6329
    },
    {
      "epoch": 4.19205298013245,
      "grad_norm": 0.9213391711206443,
      "learning_rate": 9.772253451843256e-05,
      "loss": 0.6484,
      "step": 6330
    },
    {
      "epoch": 4.192715231788079,
      "grad_norm": 0.9523749004630827,
      "learning_rate": 9.765997688084047e-05,
      "loss": 0.6758,
      "step": 6331
    },
    {
      "epoch": 4.193377483443709,
      "grad_norm": 1.157662953902676,
      "learning_rate": 9.759742960717298e-05,
      "loss": 0.8945,
      "step": 6332
    },
    {
      "epoch": 4.194039735099338,
      "grad_norm": 1.1331376497619277,
      "learning_rate": 9.753489270981524e-05,
      "loss": 0.8867,
      "step": 6333
    },
    {
      "epoch": 4.194701986754967,
      "grad_norm": 1.1845278704355464,
      "learning_rate": 9.747236620115026e-05,
      "loss": 0.9492,
      "step": 6334
    },
    {
      "epoch": 4.195364238410596,
      "grad_norm": 1.1973833363191007,
      "learning_rate": 9.740985009355893e-05,
      "loss": 1.0469,
      "step": 6335
    },
    {
      "epoch": 4.196026490066225,
      "grad_norm": 1.0864693226545,
      "learning_rate": 9.734734439942023e-05,
      "loss": 0.7422,
      "step": 6336
    },
    {
      "epoch": 4.196688741721855,
      "grad_norm": 1.1687791886864916,
      "learning_rate": 9.728484913111096e-05,
      "loss": 0.875,
      "step": 6337
    },
    {
      "epoch": 4.1973509933774835,
      "grad_norm": 0.9474838355614406,
      "learning_rate": 9.722236430100592e-05,
      "loss": 0.7422,
      "step": 6338
    },
    {
      "epoch": 4.1980132450331125,
      "grad_norm": 1.042709317715414,
      "learning_rate": 9.715988992147788e-05,
      "loss": 0.8516,
      "step": 6339
    },
    {
      "epoch": 4.198675496688741,
      "grad_norm": 1.248408007180437,
      "learning_rate": 9.70974260048974e-05,
      "loss": 1.0156,
      "step": 6340
    },
    {
      "epoch": 4.199337748344371,
      "grad_norm": 1.3485938364841386,
      "learning_rate": 9.703497256363306e-05,
      "loss": 1.0625,
      "step": 6341
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.1740129841952909,
      "learning_rate": 9.697252961005138e-05,
      "loss": 0.9844,
      "step": 6342
    },
    {
      "epoch": 4.200662251655629,
      "grad_norm": 1.1997132316208377,
      "learning_rate": 9.69100971565168e-05,
      "loss": 0.9609,
      "step": 6343
    },
    {
      "epoch": 4.201324503311258,
      "grad_norm": 1.2019450984595004,
      "learning_rate": 9.684767521539159e-05,
      "loss": 0.9648,
      "step": 6344
    },
    {
      "epoch": 4.201986754966887,
      "grad_norm": 1.1931857795889356,
      "learning_rate": 9.678526379903613e-05,
      "loss": 0.9805,
      "step": 6345
    },
    {
      "epoch": 4.202649006622517,
      "grad_norm": 1.1066204311843768,
      "learning_rate": 9.672286291980848e-05,
      "loss": 0.9023,
      "step": 6346
    },
    {
      "epoch": 4.203311258278146,
      "grad_norm": 1.0583767501303423,
      "learning_rate": 9.666047259006475e-05,
      "loss": 0.9414,
      "step": 6347
    },
    {
      "epoch": 4.203973509933775,
      "grad_norm": 1.1246865046668657,
      "learning_rate": 9.659809282215903e-05,
      "loss": 0.9102,
      "step": 6348
    },
    {
      "epoch": 4.204635761589404,
      "grad_norm": 1.2445124791927982,
      "learning_rate": 9.653572362844315e-05,
      "loss": 1.0781,
      "step": 6349
    },
    {
      "epoch": 4.2052980132450335,
      "grad_norm": 1.0025323216943478,
      "learning_rate": 9.647336502126686e-05,
      "loss": 0.8906,
      "step": 6350
    },
    {
      "epoch": 4.205960264900662,
      "grad_norm": 1.1377046472387322,
      "learning_rate": 9.641101701297806e-05,
      "loss": 0.9688,
      "step": 6351
    },
    {
      "epoch": 4.206622516556291,
      "grad_norm": 1.0892224327940327,
      "learning_rate": 9.634867961592223e-05,
      "loss": 1.0234,
      "step": 6352
    },
    {
      "epoch": 4.20728476821192,
      "grad_norm": 0.96781822687656,
      "learning_rate": 9.62863528424429e-05,
      "loss": 0.7539,
      "step": 6353
    },
    {
      "epoch": 4.207947019867549,
      "grad_norm": 1.2210425076185618,
      "learning_rate": 9.622403670488154e-05,
      "loss": 0.9727,
      "step": 6354
    },
    {
      "epoch": 4.208609271523179,
      "grad_norm": 1.0053235016991606,
      "learning_rate": 9.616173121557748e-05,
      "loss": 0.7812,
      "step": 6355
    },
    {
      "epoch": 4.209271523178808,
      "grad_norm": 1.115709294435586,
      "learning_rate": 9.609943638686783e-05,
      "loss": 0.9453,
      "step": 6356
    },
    {
      "epoch": 4.209933774834437,
      "grad_norm": 1.019291596128949,
      "learning_rate": 9.603715223108768e-05,
      "loss": 0.8359,
      "step": 6357
    },
    {
      "epoch": 4.210596026490066,
      "grad_norm": 1.0012472194257234,
      "learning_rate": 9.597487876057012e-05,
      "loss": 0.8281,
      "step": 6358
    },
    {
      "epoch": 4.211258278145696,
      "grad_norm": 1.181660273010959,
      "learning_rate": 9.591261598764595e-05,
      "loss": 0.9727,
      "step": 6359
    },
    {
      "epoch": 4.211920529801325,
      "grad_norm": 0.9677083153229692,
      "learning_rate": 9.585036392464381e-05,
      "loss": 0.793,
      "step": 6360
    },
    {
      "epoch": 4.212582781456954,
      "grad_norm": 1.182116519922221,
      "learning_rate": 9.578812258389052e-05,
      "loss": 1.0391,
      "step": 6361
    },
    {
      "epoch": 4.213245033112583,
      "grad_norm": 1.373992342997325,
      "learning_rate": 9.572589197771047e-05,
      "loss": 1.1406,
      "step": 6362
    },
    {
      "epoch": 4.2139072847682115,
      "grad_norm": 1.135145960929472,
      "learning_rate": 9.566367211842599e-05,
      "loss": 0.8867,
      "step": 6363
    },
    {
      "epoch": 4.214569536423841,
      "grad_norm": 0.9283921631574411,
      "learning_rate": 9.560146301835739e-05,
      "loss": 0.7227,
      "step": 6364
    },
    {
      "epoch": 4.21523178807947,
      "grad_norm": 1.1889076785462431,
      "learning_rate": 9.553926468982282e-05,
      "loss": 0.8555,
      "step": 6365
    },
    {
      "epoch": 4.215894039735099,
      "grad_norm": 0.9736363548200329,
      "learning_rate": 9.547707714513813e-05,
      "loss": 0.7578,
      "step": 6366
    },
    {
      "epoch": 4.216556291390728,
      "grad_norm": 1.0605341844736331,
      "learning_rate": 9.54149003966173e-05,
      "loss": 0.8398,
      "step": 6367
    },
    {
      "epoch": 4.217218543046358,
      "grad_norm": 1.1482125496212838,
      "learning_rate": 9.5352734456572e-05,
      "loss": 0.875,
      "step": 6368
    },
    {
      "epoch": 4.217880794701987,
      "grad_norm": 1.0687781528343903,
      "learning_rate": 9.529057933731174e-05,
      "loss": 0.9102,
      "step": 6369
    },
    {
      "epoch": 4.218543046357616,
      "grad_norm": 1.2079815294237133,
      "learning_rate": 9.522843505114405e-05,
      "loss": 0.9805,
      "step": 6370
    },
    {
      "epoch": 4.219205298013245,
      "grad_norm": 1.1816102529327877,
      "learning_rate": 9.516630161037416e-05,
      "loss": 0.8789,
      "step": 6371
    },
    {
      "epoch": 4.219867549668874,
      "grad_norm": 1.0295758200786869,
      "learning_rate": 9.510417902730512e-05,
      "loss": 0.8008,
      "step": 6372
    },
    {
      "epoch": 4.220529801324504,
      "grad_norm": 1.1124848887901126,
      "learning_rate": 9.504206731423805e-05,
      "loss": 0.8086,
      "step": 6373
    },
    {
      "epoch": 4.221192052980133,
      "grad_norm": 0.9688061778396168,
      "learning_rate": 9.497996648347172e-05,
      "loss": 0.8242,
      "step": 6374
    },
    {
      "epoch": 4.2218543046357615,
      "grad_norm": 0.9340412604967088,
      "learning_rate": 9.491787654730279e-05,
      "loss": 0.7148,
      "step": 6375
    },
    {
      "epoch": 4.22251655629139,
      "grad_norm": 1.0948110720077904,
      "learning_rate": 9.485579751802583e-05,
      "loss": 0.9102,
      "step": 6376
    },
    {
      "epoch": 4.22317880794702,
      "grad_norm": 1.0839621551664844,
      "learning_rate": 9.479372940793317e-05,
      "loss": 0.8984,
      "step": 6377
    },
    {
      "epoch": 4.223841059602649,
      "grad_norm": 1.0051276582715367,
      "learning_rate": 9.473167222931498e-05,
      "loss": 0.8203,
      "step": 6378
    },
    {
      "epoch": 4.224503311258278,
      "grad_norm": 1.3749651843688695,
      "learning_rate": 9.466962599445932e-05,
      "loss": 1.0703,
      "step": 6379
    },
    {
      "epoch": 4.225165562913907,
      "grad_norm": 1.045876160553046,
      "learning_rate": 9.46075907156521e-05,
      "loss": 0.793,
      "step": 6380
    },
    {
      "epoch": 4.225827814569536,
      "grad_norm": 1.0214340321984852,
      "learning_rate": 9.454556640517688e-05,
      "loss": 0.8047,
      "step": 6381
    },
    {
      "epoch": 4.226490066225166,
      "grad_norm": 1.0145596828749481,
      "learning_rate": 9.448355307531537e-05,
      "loss": 0.7617,
      "step": 6382
    },
    {
      "epoch": 4.227152317880795,
      "grad_norm": 1.184006144826606,
      "learning_rate": 9.442155073834678e-05,
      "loss": 0.8867,
      "step": 6383
    },
    {
      "epoch": 4.227814569536424,
      "grad_norm": 1.079674386578136,
      "learning_rate": 9.435955940654834e-05,
      "loss": 0.832,
      "step": 6384
    },
    {
      "epoch": 4.228476821192053,
      "grad_norm": 1.1813038490789038,
      "learning_rate": 9.429757909219502e-05,
      "loss": 0.9805,
      "step": 6385
    },
    {
      "epoch": 4.2291390728476825,
      "grad_norm": 0.8908454648317472,
      "learning_rate": 9.423560980755971e-05,
      "loss": 0.7031,
      "step": 6386
    },
    {
      "epoch": 4.2298013245033115,
      "grad_norm": 1.0673918406426417,
      "learning_rate": 9.417365156491286e-05,
      "loss": 0.8359,
      "step": 6387
    },
    {
      "epoch": 4.23046357615894,
      "grad_norm": 1.0422633164185293,
      "learning_rate": 9.411170437652314e-05,
      "loss": 0.8477,
      "step": 6388
    },
    {
      "epoch": 4.231125827814569,
      "grad_norm": 1.1336143596842143,
      "learning_rate": 9.404976825465665e-05,
      "loss": 0.8516,
      "step": 6389
    },
    {
      "epoch": 4.231788079470198,
      "grad_norm": 1.0407151053510952,
      "learning_rate": 9.398784321157746e-05,
      "loss": 0.7617,
      "step": 6390
    },
    {
      "epoch": 4.232450331125828,
      "grad_norm": 1.0587097752374712,
      "learning_rate": 9.392592925954752e-05,
      "loss": 0.7539,
      "step": 6391
    },
    {
      "epoch": 4.233112582781457,
      "grad_norm": 1.2943396970177838,
      "learning_rate": 9.386402641082647e-05,
      "loss": 0.9609,
      "step": 6392
    },
    {
      "epoch": 4.233774834437086,
      "grad_norm": 1.0051789268195652,
      "learning_rate": 9.380213467767172e-05,
      "loss": 0.8086,
      "step": 6393
    },
    {
      "epoch": 4.234437086092715,
      "grad_norm": 1.0391757209857218,
      "learning_rate": 9.374025407233859e-05,
      "loss": 0.7969,
      "step": 6394
    },
    {
      "epoch": 4.235099337748345,
      "grad_norm": 1.0042265652483715,
      "learning_rate": 9.367838460708017e-05,
      "loss": 0.8047,
      "step": 6395
    },
    {
      "epoch": 4.235761589403974,
      "grad_norm": 1.2312096007987243,
      "learning_rate": 9.361652629414728e-05,
      "loss": 0.9688,
      "step": 6396
    },
    {
      "epoch": 4.236423841059603,
      "grad_norm": 0.984468043810883,
      "learning_rate": 9.355467914578866e-05,
      "loss": 0.7539,
      "step": 6397
    },
    {
      "epoch": 4.237086092715232,
      "grad_norm": 1.1929119621431088,
      "learning_rate": 9.349284317425065e-05,
      "loss": 0.8789,
      "step": 6398
    },
    {
      "epoch": 4.237748344370861,
      "grad_norm": 1.2771271524430903,
      "learning_rate": 9.343101839177752e-05,
      "loss": 1.1094,
      "step": 6399
    },
    {
      "epoch": 4.23841059602649,
      "grad_norm": 1.2359803745335507,
      "learning_rate": 9.33692048106113e-05,
      "loss": 0.9648,
      "step": 6400
    },
    {
      "epoch": 4.239072847682119,
      "grad_norm": 1.0808978899817623,
      "learning_rate": 9.330740244299178e-05,
      "loss": 0.8164,
      "step": 6401
    },
    {
      "epoch": 4.239735099337748,
      "grad_norm": 1.0219057596122603,
      "learning_rate": 9.324561130115651e-05,
      "loss": 0.8086,
      "step": 6402
    },
    {
      "epoch": 4.240397350993377,
      "grad_norm": 1.1998339710722454,
      "learning_rate": 9.318383139734095e-05,
      "loss": 0.9492,
      "step": 6403
    },
    {
      "epoch": 4.241059602649006,
      "grad_norm": 1.189481935349903,
      "learning_rate": 9.312206274377808e-05,
      "loss": 0.957,
      "step": 6404
    },
    {
      "epoch": 4.241721854304636,
      "grad_norm": 1.0501328211492178,
      "learning_rate": 9.306030535269886e-05,
      "loss": 0.8086,
      "step": 6405
    },
    {
      "epoch": 4.242384105960265,
      "grad_norm": 1.2624321155956144,
      "learning_rate": 9.2998559236332e-05,
      "loss": 0.9727,
      "step": 6406
    },
    {
      "epoch": 4.243046357615894,
      "grad_norm": 1.2453895236937362,
      "learning_rate": 9.293682440690397e-05,
      "loss": 0.9375,
      "step": 6407
    },
    {
      "epoch": 4.243708609271523,
      "grad_norm": 1.2223612567697948,
      "learning_rate": 9.28751008766388e-05,
      "loss": 0.9062,
      "step": 6408
    },
    {
      "epoch": 4.244370860927153,
      "grad_norm": 1.2717282055522192,
      "learning_rate": 9.281338865775865e-05,
      "loss": 1.0469,
      "step": 6409
    },
    {
      "epoch": 4.245033112582782,
      "grad_norm": 1.0022742501782054,
      "learning_rate": 9.275168776248316e-05,
      "loss": 0.8008,
      "step": 6410
    },
    {
      "epoch": 4.2456953642384105,
      "grad_norm": 0.9582658288955709,
      "learning_rate": 9.268999820302978e-05,
      "loss": 0.7539,
      "step": 6411
    },
    {
      "epoch": 4.2463576158940395,
      "grad_norm": 1.0669570426076789,
      "learning_rate": 9.262831999161382e-05,
      "loss": 0.8828,
      "step": 6412
    },
    {
      "epoch": 4.247019867549669,
      "grad_norm": 1.1133437265346304,
      "learning_rate": 9.256665314044828e-05,
      "loss": 0.9453,
      "step": 6413
    },
    {
      "epoch": 4.247682119205298,
      "grad_norm": 1.1299656298220075,
      "learning_rate": 9.250499766174386e-05,
      "loss": 0.9141,
      "step": 6414
    },
    {
      "epoch": 4.248344370860927,
      "grad_norm": 1.1116719956950516,
      "learning_rate": 9.244335356770901e-05,
      "loss": 0.8672,
      "step": 6415
    },
    {
      "epoch": 4.249006622516556,
      "grad_norm": 1.292182090610611,
      "learning_rate": 9.238172087055007e-05,
      "loss": 1.125,
      "step": 6416
    },
    {
      "epoch": 4.249668874172185,
      "grad_norm": 1.0895784632044996,
      "learning_rate": 9.232009958247098e-05,
      "loss": 0.8125,
      "step": 6417
    },
    {
      "epoch": 4.250331125827815,
      "grad_norm": 1.0267292317527752,
      "learning_rate": 9.225848971567335e-05,
      "loss": 0.7109,
      "step": 6418
    },
    {
      "epoch": 4.250993377483444,
      "grad_norm": 0.9869951714916474,
      "learning_rate": 9.219689128235686e-05,
      "loss": 0.7773,
      "step": 6419
    },
    {
      "epoch": 4.251655629139073,
      "grad_norm": 1.0953755197497155,
      "learning_rate": 9.213530429471854e-05,
      "loss": 0.8672,
      "step": 6420
    },
    {
      "epoch": 4.252317880794702,
      "grad_norm": 1.1181636857780448,
      "learning_rate": 9.207372876495334e-05,
      "loss": 0.8984,
      "step": 6421
    },
    {
      "epoch": 4.252980132450331,
      "grad_norm": 1.2404276221615507,
      "learning_rate": 9.201216470525398e-05,
      "loss": 1.0,
      "step": 6422
    },
    {
      "epoch": 4.2536423841059605,
      "grad_norm": 1.1511897132471132,
      "learning_rate": 9.195061212781084e-05,
      "loss": 0.7969,
      "step": 6423
    },
    {
      "epoch": 4.2543046357615895,
      "grad_norm": 1.1372988302422178,
      "learning_rate": 9.188907104481196e-05,
      "loss": 0.7969,
      "step": 6424
    },
    {
      "epoch": 4.254966887417218,
      "grad_norm": 1.1535629779356065,
      "learning_rate": 9.182754146844327e-05,
      "loss": 0.9766,
      "step": 6425
    },
    {
      "epoch": 4.255629139072847,
      "grad_norm": 1.184966706426104,
      "learning_rate": 9.176602341088827e-05,
      "loss": 0.8359,
      "step": 6426
    },
    {
      "epoch": 4.256291390728477,
      "grad_norm": 1.2274812342797508,
      "learning_rate": 9.170451688432827e-05,
      "loss": 0.9414,
      "step": 6427
    },
    {
      "epoch": 4.256953642384106,
      "grad_norm": 0.9629787851634235,
      "learning_rate": 9.164302190094227e-05,
      "loss": 0.7344,
      "step": 6428
    },
    {
      "epoch": 4.257615894039735,
      "grad_norm": 1.1161199547174125,
      "learning_rate": 9.1581538472907e-05,
      "loss": 0.8672,
      "step": 6429
    },
    {
      "epoch": 4.258278145695364,
      "grad_norm": 1.2315211510513135,
      "learning_rate": 9.152006661239682e-05,
      "loss": 1.0156,
      "step": 6430
    },
    {
      "epoch": 4.258940397350994,
      "grad_norm": 1.0418513904872728,
      "learning_rate": 9.145860633158392e-05,
      "loss": 0.7344,
      "step": 6431
    },
    {
      "epoch": 4.259602649006623,
      "grad_norm": 1.2916533960519494,
      "learning_rate": 9.139715764263813e-05,
      "loss": 1.0703,
      "step": 6432
    },
    {
      "epoch": 4.260264900662252,
      "grad_norm": 0.9816934513317195,
      "learning_rate": 9.133572055772695e-05,
      "loss": 0.8203,
      "step": 6433
    },
    {
      "epoch": 4.260927152317881,
      "grad_norm": 1.0898418826367926,
      "learning_rate": 9.127429508901572e-05,
      "loss": 0.8828,
      "step": 6434
    },
    {
      "epoch": 4.26158940397351,
      "grad_norm": 1.233387710968643,
      "learning_rate": 9.12128812486673e-05,
      "loss": 1.0234,
      "step": 6435
    },
    {
      "epoch": 4.262251655629139,
      "grad_norm": 1.2284670848756418,
      "learning_rate": 9.115147904884232e-05,
      "loss": 1.0312,
      "step": 6436
    },
    {
      "epoch": 4.262913907284768,
      "grad_norm": 1.1599816403399947,
      "learning_rate": 9.109008850169923e-05,
      "loss": 0.9297,
      "step": 6437
    },
    {
      "epoch": 4.263576158940397,
      "grad_norm": 1.1273815978865702,
      "learning_rate": 9.102870961939399e-05,
      "loss": 0.8555,
      "step": 6438
    },
    {
      "epoch": 4.264238410596026,
      "grad_norm": 1.0727035563369833,
      "learning_rate": 9.096734241408027e-05,
      "loss": 0.832,
      "step": 6439
    },
    {
      "epoch": 4.264900662251655,
      "grad_norm": 1.2769597288504944,
      "learning_rate": 9.09059868979096e-05,
      "loss": 1.0391,
      "step": 6440
    },
    {
      "epoch": 4.265562913907285,
      "grad_norm": 1.0792071819946627,
      "learning_rate": 9.0844643083031e-05,
      "loss": 0.7852,
      "step": 6441
    },
    {
      "epoch": 4.266225165562914,
      "grad_norm": 0.9146475858802257,
      "learning_rate": 9.078331098159126e-05,
      "loss": 0.7578,
      "step": 6442
    },
    {
      "epoch": 4.266887417218543,
      "grad_norm": 1.044225761637071,
      "learning_rate": 9.072199060573484e-05,
      "loss": 0.8555,
      "step": 6443
    },
    {
      "epoch": 4.267549668874172,
      "grad_norm": 1.145641840910904,
      "learning_rate": 9.066068196760392e-05,
      "loss": 0.8516,
      "step": 6444
    },
    {
      "epoch": 4.268211920529802,
      "grad_norm": 1.046860453440571,
      "learning_rate": 9.059938507933819e-05,
      "loss": 0.793,
      "step": 6445
    },
    {
      "epoch": 4.268874172185431,
      "grad_norm": 1.1036731084195965,
      "learning_rate": 9.053809995307528e-05,
      "loss": 0.8672,
      "step": 6446
    },
    {
      "epoch": 4.26953642384106,
      "grad_norm": 1.0325140401122295,
      "learning_rate": 9.047682660095028e-05,
      "loss": 0.7969,
      "step": 6447
    },
    {
      "epoch": 4.2701986754966885,
      "grad_norm": 1.3903200297123874,
      "learning_rate": 9.0415565035096e-05,
      "loss": 1.1328,
      "step": 6448
    },
    {
      "epoch": 4.2708609271523175,
      "grad_norm": 1.1606864109989947,
      "learning_rate": 9.035431526764299e-05,
      "loss": 0.8516,
      "step": 6449
    },
    {
      "epoch": 4.271523178807947,
      "grad_norm": 1.0503341243233026,
      "learning_rate": 9.029307731071939e-05,
      "loss": 0.832,
      "step": 6450
    },
    {
      "epoch": 4.272185430463576,
      "grad_norm": 1.2073721314987047,
      "learning_rate": 9.023185117645095e-05,
      "loss": 0.9766,
      "step": 6451
    },
    {
      "epoch": 4.272847682119205,
      "grad_norm": 1.0775086388924306,
      "learning_rate": 9.017063687696119e-05,
      "loss": 0.7344,
      "step": 6452
    },
    {
      "epoch": 4.273509933774834,
      "grad_norm": 1.2922065922706014,
      "learning_rate": 9.01094344243713e-05,
      "loss": 1.0,
      "step": 6453
    },
    {
      "epoch": 4.274172185430464,
      "grad_norm": 1.0249221324117628,
      "learning_rate": 9.004824383079995e-05,
      "loss": 0.7812,
      "step": 6454
    },
    {
      "epoch": 4.274834437086093,
      "grad_norm": 1.1289590701775016,
      "learning_rate": 8.998706510836369e-05,
      "loss": 0.8633,
      "step": 6455
    },
    {
      "epoch": 4.275496688741722,
      "grad_norm": 1.1817607805243004,
      "learning_rate": 8.99258982691766e-05,
      "loss": 0.875,
      "step": 6456
    },
    {
      "epoch": 4.276158940397351,
      "grad_norm": 1.1859075635059466,
      "learning_rate": 8.986474332535032e-05,
      "loss": 0.918,
      "step": 6457
    },
    {
      "epoch": 4.27682119205298,
      "grad_norm": 1.2898863731799532,
      "learning_rate": 8.980360028899431e-05,
      "loss": 1.125,
      "step": 6458
    },
    {
      "epoch": 4.27748344370861,
      "grad_norm": 1.2921682210258052,
      "learning_rate": 8.974246917221557e-05,
      "loss": 1.0469,
      "step": 6459
    },
    {
      "epoch": 4.2781456953642385,
      "grad_norm": 1.0811499774537532,
      "learning_rate": 8.968134998711875e-05,
      "loss": 0.8281,
      "step": 6460
    },
    {
      "epoch": 4.278807947019867,
      "grad_norm": 1.1002991631122645,
      "learning_rate": 8.96202427458062e-05,
      "loss": 0.9062,
      "step": 6461
    },
    {
      "epoch": 4.279470198675496,
      "grad_norm": 1.0778643230130702,
      "learning_rate": 8.955914746037783e-05,
      "loss": 0.8516,
      "step": 6462
    },
    {
      "epoch": 4.280132450331126,
      "grad_norm": 1.0666939955318964,
      "learning_rate": 8.949806414293111e-05,
      "loss": 0.8633,
      "step": 6463
    },
    {
      "epoch": 4.280794701986755,
      "grad_norm": 1.2875522295064752,
      "learning_rate": 8.94369928055614e-05,
      "loss": 1.0391,
      "step": 6464
    },
    {
      "epoch": 4.281456953642384,
      "grad_norm": 1.2324830155065163,
      "learning_rate": 8.937593346036145e-05,
      "loss": 0.9688,
      "step": 6465
    },
    {
      "epoch": 4.282119205298013,
      "grad_norm": 1.209314565621922,
      "learning_rate": 8.931488611942162e-05,
      "loss": 0.9453,
      "step": 6466
    },
    {
      "epoch": 4.282781456953642,
      "grad_norm": 1.20535288296118,
      "learning_rate": 8.925385079483016e-05,
      "loss": 0.8906,
      "step": 6467
    },
    {
      "epoch": 4.283443708609272,
      "grad_norm": 1.247691862091088,
      "learning_rate": 8.919282749867266e-05,
      "loss": 1.0156,
      "step": 6468
    },
    {
      "epoch": 4.284105960264901,
      "grad_norm": 0.9585489735079032,
      "learning_rate": 8.91318162430324e-05,
      "loss": 0.7695,
      "step": 6469
    },
    {
      "epoch": 4.28476821192053,
      "grad_norm": 1.0582705729775692,
      "learning_rate": 8.907081703999038e-05,
      "loss": 0.9141,
      "step": 6470
    },
    {
      "epoch": 4.285430463576159,
      "grad_norm": 1.1233493330715014,
      "learning_rate": 8.900982990162516e-05,
      "loss": 0.8867,
      "step": 6471
    },
    {
      "epoch": 4.2860927152317885,
      "grad_norm": 1.0405564493758683,
      "learning_rate": 8.894885484001281e-05,
      "loss": 0.9102,
      "step": 6472
    },
    {
      "epoch": 4.286754966887417,
      "grad_norm": 1.0939263309416731,
      "learning_rate": 8.888789186722712e-05,
      "loss": 0.7891,
      "step": 6473
    },
    {
      "epoch": 4.287417218543046,
      "grad_norm": 1.03146645769966,
      "learning_rate": 8.882694099533948e-05,
      "loss": 0.8672,
      "step": 6474
    },
    {
      "epoch": 4.288079470198675,
      "grad_norm": 1.3800269250458674,
      "learning_rate": 8.876600223641886e-05,
      "loss": 1.1641,
      "step": 6475
    },
    {
      "epoch": 4.288741721854304,
      "grad_norm": 1.1548462236047603,
      "learning_rate": 8.870507560253176e-05,
      "loss": 0.9492,
      "step": 6476
    },
    {
      "epoch": 4.289403973509934,
      "grad_norm": 1.098766439727207,
      "learning_rate": 8.864416110574249e-05,
      "loss": 0.8789,
      "step": 6477
    },
    {
      "epoch": 4.290066225165563,
      "grad_norm": 1.138459459204018,
      "learning_rate": 8.858325875811273e-05,
      "loss": 0.9688,
      "step": 6478
    },
    {
      "epoch": 4.290728476821192,
      "grad_norm": 1.213706060706707,
      "learning_rate": 8.852236857170181e-05,
      "loss": 0.9883,
      "step": 6479
    },
    {
      "epoch": 4.291390728476821,
      "grad_norm": 1.0271159636446885,
      "learning_rate": 8.846149055856678e-05,
      "loss": 0.7305,
      "step": 6480
    },
    {
      "epoch": 4.292052980132451,
      "grad_norm": 1.2930202569586278,
      "learning_rate": 8.840062473076214e-05,
      "loss": 0.9258,
      "step": 6481
    },
    {
      "epoch": 4.29271523178808,
      "grad_norm": 1.1778441782102695,
      "learning_rate": 8.833977110033995e-05,
      "loss": 0.8867,
      "step": 6482
    },
    {
      "epoch": 4.293377483443709,
      "grad_norm": 1.1210598243631176,
      "learning_rate": 8.827892967935008e-05,
      "loss": 0.8984,
      "step": 6483
    },
    {
      "epoch": 4.294039735099338,
      "grad_norm": 1.073120318874248,
      "learning_rate": 8.82181004798397e-05,
      "loss": 0.832,
      "step": 6484
    },
    {
      "epoch": 4.2947019867549665,
      "grad_norm": 1.0846231265241009,
      "learning_rate": 8.815728351385371e-05,
      "loss": 0.7383,
      "step": 6485
    },
    {
      "epoch": 4.295364238410596,
      "grad_norm": 1.1419120784735433,
      "learning_rate": 8.809647879343462e-05,
      "loss": 0.9297,
      "step": 6486
    },
    {
      "epoch": 4.296026490066225,
      "grad_norm": 1.275142629972156,
      "learning_rate": 8.803568633062244e-05,
      "loss": 0.9688,
      "step": 6487
    },
    {
      "epoch": 4.296688741721854,
      "grad_norm": 1.2662426107608238,
      "learning_rate": 8.797490613745472e-05,
      "loss": 1.0078,
      "step": 6488
    },
    {
      "epoch": 4.297350993377483,
      "grad_norm": 1.2950733483329975,
      "learning_rate": 8.79141382259667e-05,
      "loss": 1.0391,
      "step": 6489
    },
    {
      "epoch": 4.298013245033113,
      "grad_norm": 1.2061348918891535,
      "learning_rate": 8.785338260819113e-05,
      "loss": 0.9922,
      "step": 6490
    },
    {
      "epoch": 4.298675496688742,
      "grad_norm": 1.1532494720991027,
      "learning_rate": 8.779263929615824e-05,
      "loss": 0.8594,
      "step": 6491
    },
    {
      "epoch": 4.299337748344371,
      "grad_norm": 1.1363772298905315,
      "learning_rate": 8.773190830189601e-05,
      "loss": 0.8633,
      "step": 6492
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.1327265795281738,
      "learning_rate": 8.767118963742979e-05,
      "loss": 0.9023,
      "step": 6493
    },
    {
      "epoch": 4.300662251655629,
      "grad_norm": 1.3511613611938742,
      "learning_rate": 8.761048331478256e-05,
      "loss": 1.1094,
      "step": 6494
    },
    {
      "epoch": 4.301324503311259,
      "grad_norm": 1.0499832504684283,
      "learning_rate": 8.754978934597499e-05,
      "loss": 0.8594,
      "step": 6495
    },
    {
      "epoch": 4.3019867549668875,
      "grad_norm": 1.0500849469597457,
      "learning_rate": 8.748910774302501e-05,
      "loss": 0.8047,
      "step": 6496
    },
    {
      "epoch": 4.3026490066225165,
      "grad_norm": 1.07718259033963,
      "learning_rate": 8.74284385179484e-05,
      "loss": 0.8242,
      "step": 6497
    },
    {
      "epoch": 4.303311258278145,
      "grad_norm": 1.1184508549964862,
      "learning_rate": 8.736778168275835e-05,
      "loss": 0.8984,
      "step": 6498
    },
    {
      "epoch": 4.303973509933774,
      "grad_norm": 1.0543322122960102,
      "learning_rate": 8.730713724946557e-05,
      "loss": 0.8359,
      "step": 6499
    },
    {
      "epoch": 4.304635761589404,
      "grad_norm": 0.9773925031059764,
      "learning_rate": 8.724650523007828e-05,
      "loss": 0.7617,
      "step": 6500
    },
    {
      "epoch": 4.305298013245033,
      "grad_norm": 1.116408911714799,
      "learning_rate": 8.71858856366025e-05,
      "loss": 0.9805,
      "step": 6501
    },
    {
      "epoch": 4.305960264900662,
      "grad_norm": 1.1578753148364553,
      "learning_rate": 8.712527848104152e-05,
      "loss": 0.8789,
      "step": 6502
    },
    {
      "epoch": 4.306622516556291,
      "grad_norm": 1.404442418044603,
      "learning_rate": 8.706468377539616e-05,
      "loss": 1.2109,
      "step": 6503
    },
    {
      "epoch": 4.307284768211921,
      "grad_norm": 1.2105688608967666,
      "learning_rate": 8.700410153166493e-05,
      "loss": 0.9961,
      "step": 6504
    },
    {
      "epoch": 4.30794701986755,
      "grad_norm": 1.0742676311385257,
      "learning_rate": 8.694353176184392e-05,
      "loss": 0.8008,
      "step": 6505
    },
    {
      "epoch": 4.308609271523179,
      "grad_norm": 1.308510772030092,
      "learning_rate": 8.688297447792645e-05,
      "loss": 0.9688,
      "step": 6506
    },
    {
      "epoch": 4.309271523178808,
      "grad_norm": 1.2319254428356263,
      "learning_rate": 8.682242969190371e-05,
      "loss": 0.9258,
      "step": 6507
    },
    {
      "epoch": 4.3099337748344375,
      "grad_norm": 1.048518455518634,
      "learning_rate": 8.676189741576414e-05,
      "loss": 0.8984,
      "step": 6508
    },
    {
      "epoch": 4.3105960264900665,
      "grad_norm": 0.9786386280947232,
      "learning_rate": 8.670137766149388e-05,
      "loss": 0.7617,
      "step": 6509
    },
    {
      "epoch": 4.311258278145695,
      "grad_norm": 0.9867022842855568,
      "learning_rate": 8.664087044107662e-05,
      "loss": 0.8047,
      "step": 6510
    },
    {
      "epoch": 4.311920529801324,
      "grad_norm": 1.204806630029523,
      "learning_rate": 8.658037576649338e-05,
      "loss": 1.0,
      "step": 6511
    },
    {
      "epoch": 4.312582781456953,
      "grad_norm": 1.2314258858619214,
      "learning_rate": 8.651989364972277e-05,
      "loss": 1.0859,
      "step": 6512
    },
    {
      "epoch": 4.313245033112583,
      "grad_norm": 0.931008347347989,
      "learning_rate": 8.645942410274098e-05,
      "loss": 0.6562,
      "step": 6513
    },
    {
      "epoch": 4.313907284768212,
      "grad_norm": 1.105687984199342,
      "learning_rate": 8.639896713752176e-05,
      "loss": 0.793,
      "step": 6514
    },
    {
      "epoch": 4.314569536423841,
      "grad_norm": 1.0126082460091115,
      "learning_rate": 8.633852276603616e-05,
      "loss": 0.7734,
      "step": 6515
    },
    {
      "epoch": 4.31523178807947,
      "grad_norm": 1.0765367173194051,
      "learning_rate": 8.627809100025287e-05,
      "loss": 0.8633,
      "step": 6516
    },
    {
      "epoch": 4.315894039735099,
      "grad_norm": 1.0457040590024647,
      "learning_rate": 8.621767185213819e-05,
      "loss": 0.7852,
      "step": 6517
    },
    {
      "epoch": 4.316556291390729,
      "grad_norm": 1.2584060044538594,
      "learning_rate": 8.615726533365568e-05,
      "loss": 1.0234,
      "step": 6518
    },
    {
      "epoch": 4.317218543046358,
      "grad_norm": 1.1690578053749487,
      "learning_rate": 8.60968714567666e-05,
      "loss": 0.9844,
      "step": 6519
    },
    {
      "epoch": 4.317880794701987,
      "grad_norm": 1.0993124487006185,
      "learning_rate": 8.603649023342955e-05,
      "loss": 0.9102,
      "step": 6520
    },
    {
      "epoch": 4.3185430463576155,
      "grad_norm": 1.114645141275922,
      "learning_rate": 8.597612167560075e-05,
      "loss": 0.8164,
      "step": 6521
    },
    {
      "epoch": 4.319205298013245,
      "grad_norm": 1.1224813662990656,
      "learning_rate": 8.591576579523393e-05,
      "loss": 0.875,
      "step": 6522
    },
    {
      "epoch": 4.319867549668874,
      "grad_norm": 1.0577479989547875,
      "learning_rate": 8.585542260428018e-05,
      "loss": 0.7344,
      "step": 6523
    },
    {
      "epoch": 4.320529801324503,
      "grad_norm": 1.1457339265308462,
      "learning_rate": 8.57950921146881e-05,
      "loss": 0.8906,
      "step": 6524
    },
    {
      "epoch": 4.321192052980132,
      "grad_norm": 1.070397577534459,
      "learning_rate": 8.573477433840385e-05,
      "loss": 0.8438,
      "step": 6525
    },
    {
      "epoch": 4.321854304635762,
      "grad_norm": 1.2811075748783753,
      "learning_rate": 8.567446928737114e-05,
      "loss": 0.9414,
      "step": 6526
    },
    {
      "epoch": 4.322516556291391,
      "grad_norm": 1.1951335390804283,
      "learning_rate": 8.561417697353095e-05,
      "loss": 1.0,
      "step": 6527
    },
    {
      "epoch": 4.32317880794702,
      "grad_norm": 1.2127491848071905,
      "learning_rate": 8.555389740882184e-05,
      "loss": 0.957,
      "step": 6528
    },
    {
      "epoch": 4.323841059602649,
      "grad_norm": 1.0165547434444686,
      "learning_rate": 8.549363060517988e-05,
      "loss": 0.7617,
      "step": 6529
    },
    {
      "epoch": 4.324503311258278,
      "grad_norm": 1.0386171334140835,
      "learning_rate": 8.543337657453868e-05,
      "loss": 0.7734,
      "step": 6530
    },
    {
      "epoch": 4.325165562913908,
      "grad_norm": 1.3543796781944457,
      "learning_rate": 8.537313532882907e-05,
      "loss": 1.0,
      "step": 6531
    },
    {
      "epoch": 4.325827814569537,
      "grad_norm": 1.1205183554875442,
      "learning_rate": 8.531290687997965e-05,
      "loss": 0.8945,
      "step": 6532
    },
    {
      "epoch": 4.3264900662251655,
      "grad_norm": 1.314765940935484,
      "learning_rate": 8.525269123991623e-05,
      "loss": 1.0234,
      "step": 6533
    },
    {
      "epoch": 4.3271523178807945,
      "grad_norm": 1.0472262039832598,
      "learning_rate": 8.519248842056225e-05,
      "loss": 0.7734,
      "step": 6534
    },
    {
      "epoch": 4.327814569536423,
      "grad_norm": 1.0417705748543051,
      "learning_rate": 8.513229843383859e-05,
      "loss": 0.8203,
      "step": 6535
    },
    {
      "epoch": 4.328476821192053,
      "grad_norm": 1.0109130075484016,
      "learning_rate": 8.507212129166358e-05,
      "loss": 0.75,
      "step": 6536
    },
    {
      "epoch": 4.329139072847682,
      "grad_norm": 1.1641529932921637,
      "learning_rate": 8.501195700595275e-05,
      "loss": 0.875,
      "step": 6537
    },
    {
      "epoch": 4.329801324503311,
      "grad_norm": 1.1050832243499442,
      "learning_rate": 8.495180558861965e-05,
      "loss": 0.8164,
      "step": 6538
    },
    {
      "epoch": 4.33046357615894,
      "grad_norm": 1.118964944831242,
      "learning_rate": 8.489166705157478e-05,
      "loss": 0.875,
      "step": 6539
    },
    {
      "epoch": 4.33112582781457,
      "grad_norm": 1.0871735755745358,
      "learning_rate": 8.483154140672623e-05,
      "loss": 0.8281,
      "step": 6540
    },
    {
      "epoch": 4.331788079470199,
      "grad_norm": 1.0563780685586572,
      "learning_rate": 8.47714286659796e-05,
      "loss": 0.8398,
      "step": 6541
    },
    {
      "epoch": 4.332450331125828,
      "grad_norm": 1.1522550878216666,
      "learning_rate": 8.471132884123801e-05,
      "loss": 0.9102,
      "step": 6542
    },
    {
      "epoch": 4.333112582781457,
      "grad_norm": 1.1986314259350248,
      "learning_rate": 8.465124194440171e-05,
      "loss": 0.9102,
      "step": 6543
    },
    {
      "epoch": 4.3337748344370866,
      "grad_norm": 1.1556693760458472,
      "learning_rate": 8.45911679873688e-05,
      "loss": 0.8594,
      "step": 6544
    },
    {
      "epoch": 4.3344370860927155,
      "grad_norm": 1.238035709832722,
      "learning_rate": 8.453110698203443e-05,
      "loss": 0.8984,
      "step": 6545
    },
    {
      "epoch": 4.335099337748344,
      "grad_norm": 1.171110689410266,
      "learning_rate": 8.447105894029147e-05,
      "loss": 0.875,
      "step": 6546
    },
    {
      "epoch": 4.335761589403973,
      "grad_norm": 1.0659584483463174,
      "learning_rate": 8.441102387403014e-05,
      "loss": 0.8672,
      "step": 6547
    },
    {
      "epoch": 4.336423841059602,
      "grad_norm": 1.2909584772547398,
      "learning_rate": 8.435100179513807e-05,
      "loss": 1.0078,
      "step": 6548
    },
    {
      "epoch": 4.337086092715232,
      "grad_norm": 1.1195846118465116,
      "learning_rate": 8.429099271550022e-05,
      "loss": 0.7383,
      "step": 6549
    },
    {
      "epoch": 4.337748344370861,
      "grad_norm": 1.1132429841498623,
      "learning_rate": 8.423099664699914e-05,
      "loss": 0.8047,
      "step": 6550
    },
    {
      "epoch": 4.33841059602649,
      "grad_norm": 1.3415162180677649,
      "learning_rate": 8.417101360151479e-05,
      "loss": 1.0547,
      "step": 6551
    },
    {
      "epoch": 4.339072847682119,
      "grad_norm": 1.0785021012284788,
      "learning_rate": 8.41110435909244e-05,
      "loss": 0.8789,
      "step": 6552
    },
    {
      "epoch": 4.339735099337748,
      "grad_norm": 1.2478490187056797,
      "learning_rate": 8.405108662710283e-05,
      "loss": 0.9492,
      "step": 6553
    },
    {
      "epoch": 4.340397350993378,
      "grad_norm": 1.1605860610427157,
      "learning_rate": 8.399114272192212e-05,
      "loss": 0.9219,
      "step": 6554
    },
    {
      "epoch": 4.341059602649007,
      "grad_norm": 1.073646000099421,
      "learning_rate": 8.393121188725193e-05,
      "loss": 0.8164,
      "step": 6555
    },
    {
      "epoch": 4.341721854304636,
      "grad_norm": 1.1154438850111101,
      "learning_rate": 8.387129413495928e-05,
      "loss": 0.8125,
      "step": 6556
    },
    {
      "epoch": 4.342384105960265,
      "grad_norm": 1.102894165221164,
      "learning_rate": 8.381138947690849e-05,
      "loss": 0.8242,
      "step": 6557
    },
    {
      "epoch": 4.343046357615894,
      "grad_norm": 1.4744681389366043,
      "learning_rate": 8.375149792496141e-05,
      "loss": 1.2734,
      "step": 6558
    },
    {
      "epoch": 4.343708609271523,
      "grad_norm": 1.0504640473517697,
      "learning_rate": 8.369161949097732e-05,
      "loss": 0.7578,
      "step": 6559
    },
    {
      "epoch": 4.344370860927152,
      "grad_norm": 1.2884883224488397,
      "learning_rate": 8.363175418681275e-05,
      "loss": 1.0469,
      "step": 6560
    },
    {
      "epoch": 4.345033112582781,
      "grad_norm": 1.214502760014441,
      "learning_rate": 8.35719020243217e-05,
      "loss": 0.9062,
      "step": 6561
    },
    {
      "epoch": 4.34569536423841,
      "grad_norm": 1.1592231811682447,
      "learning_rate": 8.351206301535561e-05,
      "loss": 0.9688,
      "step": 6562
    },
    {
      "epoch": 4.34635761589404,
      "grad_norm": 1.104092873247346,
      "learning_rate": 8.345223717176338e-05,
      "loss": 0.8828,
      "step": 6563
    },
    {
      "epoch": 4.347019867549669,
      "grad_norm": 1.0051248145490712,
      "learning_rate": 8.339242450539105e-05,
      "loss": 0.7852,
      "step": 6564
    },
    {
      "epoch": 4.347682119205298,
      "grad_norm": 1.192770158583818,
      "learning_rate": 8.333262502808237e-05,
      "loss": 0.9688,
      "step": 6565
    },
    {
      "epoch": 4.348344370860927,
      "grad_norm": 1.071704901045227,
      "learning_rate": 8.32728387516782e-05,
      "loss": 0.793,
      "step": 6566
    },
    {
      "epoch": 4.349006622516557,
      "grad_norm": 0.9664527996154553,
      "learning_rate": 8.321306568801695e-05,
      "loss": 0.7695,
      "step": 6567
    },
    {
      "epoch": 4.349668874172186,
      "grad_norm": 0.9561335542099324,
      "learning_rate": 8.315330584893444e-05,
      "loss": 0.7188,
      "step": 6568
    },
    {
      "epoch": 4.350331125827815,
      "grad_norm": 1.0957160678359565,
      "learning_rate": 8.309355924626376e-05,
      "loss": 0.8008,
      "step": 6569
    },
    {
      "epoch": 4.3509933774834435,
      "grad_norm": 1.1025295533706028,
      "learning_rate": 8.303382589183524e-05,
      "loss": 0.8398,
      "step": 6570
    },
    {
      "epoch": 4.3516556291390724,
      "grad_norm": 1.1950828869136805,
      "learning_rate": 8.29741057974771e-05,
      "loss": 0.9102,
      "step": 6571
    },
    {
      "epoch": 4.352317880794702,
      "grad_norm": 1.1006090408976663,
      "learning_rate": 8.291439897501444e-05,
      "loss": 0.7656,
      "step": 6572
    },
    {
      "epoch": 4.352980132450331,
      "grad_norm": 1.2121962503462504,
      "learning_rate": 8.285470543626981e-05,
      "loss": 0.8789,
      "step": 6573
    },
    {
      "epoch": 4.35364238410596,
      "grad_norm": 1.3151973402304271,
      "learning_rate": 8.279502519306332e-05,
      "loss": 1.0,
      "step": 6574
    },
    {
      "epoch": 4.354304635761589,
      "grad_norm": 1.3618572858536624,
      "learning_rate": 8.273535825721238e-05,
      "loss": 1.1016,
      "step": 6575
    },
    {
      "epoch": 4.354966887417219,
      "grad_norm": 1.300058585797083,
      "learning_rate": 8.267570464053158e-05,
      "loss": 0.9805,
      "step": 6576
    },
    {
      "epoch": 4.355629139072848,
      "grad_norm": 1.184031077250656,
      "learning_rate": 8.261606435483319e-05,
      "loss": 0.8633,
      "step": 6577
    },
    {
      "epoch": 4.356291390728477,
      "grad_norm": 1.1300417310460407,
      "learning_rate": 8.255643741192654e-05,
      "loss": 0.832,
      "step": 6578
    },
    {
      "epoch": 4.356953642384106,
      "grad_norm": 1.114157950226741,
      "learning_rate": 8.24968238236185e-05,
      "loss": 0.8164,
      "step": 6579
    },
    {
      "epoch": 4.357615894039735,
      "grad_norm": 1.0445024191681573,
      "learning_rate": 8.243722360171329e-05,
      "loss": 0.8398,
      "step": 6580
    },
    {
      "epoch": 4.3582781456953645,
      "grad_norm": 1.1685642525673738,
      "learning_rate": 8.237763675801238e-05,
      "loss": 0.9453,
      "step": 6581
    },
    {
      "epoch": 4.3589403973509935,
      "grad_norm": 1.25653984426189,
      "learning_rate": 8.23180633043146e-05,
      "loss": 0.9219,
      "step": 6582
    },
    {
      "epoch": 4.359602649006622,
      "grad_norm": 1.0367693161281861,
      "learning_rate": 8.225850325241626e-05,
      "loss": 0.7812,
      "step": 6583
    },
    {
      "epoch": 4.360264900662251,
      "grad_norm": 1.0947242453991388,
      "learning_rate": 8.219895661411092e-05,
      "loss": 0.8047,
      "step": 6584
    },
    {
      "epoch": 4.360927152317881,
      "grad_norm": 1.2326569821429494,
      "learning_rate": 8.213942340118951e-05,
      "loss": 0.9375,
      "step": 6585
    },
    {
      "epoch": 4.36158940397351,
      "grad_norm": 1.1785066419064292,
      "learning_rate": 8.207990362544022e-05,
      "loss": 0.8398,
      "step": 6586
    },
    {
      "epoch": 4.362251655629139,
      "grad_norm": 1.1873254867853171,
      "learning_rate": 8.202039729864867e-05,
      "loss": 0.9219,
      "step": 6587
    },
    {
      "epoch": 4.362913907284768,
      "grad_norm": 0.9762211203946645,
      "learning_rate": 8.196090443259787e-05,
      "loss": 0.6875,
      "step": 6588
    },
    {
      "epoch": 4.363576158940397,
      "grad_norm": 0.9711837611368349,
      "learning_rate": 8.190142503906798e-05,
      "loss": 0.7461,
      "step": 6589
    },
    {
      "epoch": 4.364238410596027,
      "grad_norm": 1.226985246997386,
      "learning_rate": 8.184195912983673e-05,
      "loss": 0.9141,
      "step": 6590
    },
    {
      "epoch": 4.364900662251656,
      "grad_norm": 1.0652454489846093,
      "learning_rate": 8.178250671667891e-05,
      "loss": 0.7734,
      "step": 6591
    },
    {
      "epoch": 4.365562913907285,
      "grad_norm": 1.1953601354619727,
      "learning_rate": 8.172306781136685e-05,
      "loss": 0.9414,
      "step": 6592
    },
    {
      "epoch": 4.366225165562914,
      "grad_norm": 1.2674960195595504,
      "learning_rate": 8.166364242567018e-05,
      "loss": 0.957,
      "step": 6593
    },
    {
      "epoch": 4.3668874172185435,
      "grad_norm": 1.1190671121053586,
      "learning_rate": 8.160423057135572e-05,
      "loss": 0.8945,
      "step": 6594
    },
    {
      "epoch": 4.367549668874172,
      "grad_norm": 1.193792833073534,
      "learning_rate": 8.154483226018773e-05,
      "loss": 0.9102,
      "step": 6595
    },
    {
      "epoch": 4.368211920529801,
      "grad_norm": 1.235329512097102,
      "learning_rate": 8.148544750392783e-05,
      "loss": 0.9375,
      "step": 6596
    },
    {
      "epoch": 4.36887417218543,
      "grad_norm": 1.1442611056660759,
      "learning_rate": 8.14260763143348e-05,
      "loss": 0.8867,
      "step": 6597
    },
    {
      "epoch": 4.369536423841059,
      "grad_norm": 1.0793076426892187,
      "learning_rate": 8.136671870316478e-05,
      "loss": 0.7891,
      "step": 6598
    },
    {
      "epoch": 4.370198675496689,
      "grad_norm": 1.124819227265678,
      "learning_rate": 8.130737468217133e-05,
      "loss": 0.8945,
      "step": 6599
    },
    {
      "epoch": 4.370860927152318,
      "grad_norm": 1.1622233077799642,
      "learning_rate": 8.124804426310527e-05,
      "loss": 0.9648,
      "step": 6600
    },
    {
      "epoch": 4.371523178807947,
      "grad_norm": 0.9545244146821957,
      "learning_rate": 8.118872745771461e-05,
      "loss": 0.8359,
      "step": 6601
    },
    {
      "epoch": 4.372185430463576,
      "grad_norm": 1.1032754763789947,
      "learning_rate": 8.112942427774488e-05,
      "loss": 0.8125,
      "step": 6602
    },
    {
      "epoch": 4.372847682119206,
      "grad_norm": 1.2388393400328686,
      "learning_rate": 8.107013473493863e-05,
      "loss": 0.957,
      "step": 6603
    },
    {
      "epoch": 4.373509933774835,
      "grad_norm": 1.183275925049816,
      "learning_rate": 8.101085884103598e-05,
      "loss": 0.8789,
      "step": 6604
    },
    {
      "epoch": 4.374172185430464,
      "grad_norm": 1.1528850570288558,
      "learning_rate": 8.095159660777426e-05,
      "loss": 0.9336,
      "step": 6605
    },
    {
      "epoch": 4.3748344370860925,
      "grad_norm": 1.090090580160501,
      "learning_rate": 8.089234804688803e-05,
      "loss": 0.8242,
      "step": 6606
    },
    {
      "epoch": 4.3754966887417215,
      "grad_norm": 1.12574167469852,
      "learning_rate": 8.083311317010914e-05,
      "loss": 0.8477,
      "step": 6607
    },
    {
      "epoch": 4.376158940397351,
      "grad_norm": 1.1478687611613405,
      "learning_rate": 8.077389198916681e-05,
      "loss": 0.8789,
      "step": 6608
    },
    {
      "epoch": 4.37682119205298,
      "grad_norm": 1.0679615791990253,
      "learning_rate": 8.071468451578754e-05,
      "loss": 0.7969,
      "step": 6609
    },
    {
      "epoch": 4.377483443708609,
      "grad_norm": 1.0942488433565987,
      "learning_rate": 8.065549076169506e-05,
      "loss": 0.7695,
      "step": 6610
    },
    {
      "epoch": 4.378145695364238,
      "grad_norm": 0.9454352174887917,
      "learning_rate": 8.059631073861038e-05,
      "loss": 0.6641,
      "step": 6611
    },
    {
      "epoch": 4.378807947019867,
      "grad_norm": 1.074625700476439,
      "learning_rate": 8.053714445825192e-05,
      "loss": 0.8125,
      "step": 6612
    },
    {
      "epoch": 4.379470198675497,
      "grad_norm": 1.0294798648306254,
      "learning_rate": 8.047799193233518e-05,
      "loss": 0.7383,
      "step": 6613
    },
    {
      "epoch": 4.380132450331126,
      "grad_norm": 1.1746390864502145,
      "learning_rate": 8.041885317257313e-05,
      "loss": 0.9258,
      "step": 6614
    },
    {
      "epoch": 4.380794701986755,
      "grad_norm": 1.0888892849645295,
      "learning_rate": 8.035972819067582e-05,
      "loss": 0.8555,
      "step": 6615
    },
    {
      "epoch": 4.381456953642384,
      "grad_norm": 1.2222679589653076,
      "learning_rate": 8.030061699835071e-05,
      "loss": 0.9062,
      "step": 6616
    },
    {
      "epoch": 4.382119205298014,
      "grad_norm": 1.2511019060538426,
      "learning_rate": 8.024151960730261e-05,
      "loss": 0.9727,
      "step": 6617
    },
    {
      "epoch": 4.3827814569536425,
      "grad_norm": 1.0117680564891882,
      "learning_rate": 8.018243602923335e-05,
      "loss": 0.7617,
      "step": 6618
    },
    {
      "epoch": 4.3834437086092715,
      "grad_norm": 1.312960781814946,
      "learning_rate": 8.012336627584214e-05,
      "loss": 0.9805,
      "step": 6619
    },
    {
      "epoch": 4.3841059602649,
      "grad_norm": 1.338241079961762,
      "learning_rate": 8.006431035882554e-05,
      "loss": 1.0156,
      "step": 6620
    },
    {
      "epoch": 4.38476821192053,
      "grad_norm": 1.0272388734369555,
      "learning_rate": 8.000526828987728e-05,
      "loss": 0.7344,
      "step": 6621
    },
    {
      "epoch": 4.385430463576159,
      "grad_norm": 1.0868562584749215,
      "learning_rate": 7.994624008068834e-05,
      "loss": 0.8086,
      "step": 6622
    },
    {
      "epoch": 4.386092715231788,
      "grad_norm": 1.0940004621320567,
      "learning_rate": 7.988722574294707e-05,
      "loss": 0.7344,
      "step": 6623
    },
    {
      "epoch": 4.386754966887417,
      "grad_norm": 1.3052027824056271,
      "learning_rate": 7.982822528833884e-05,
      "loss": 1.0078,
      "step": 6624
    },
    {
      "epoch": 4.387417218543046,
      "grad_norm": 1.0992004292268522,
      "learning_rate": 7.976923872854651e-05,
      "loss": 0.875,
      "step": 6625
    },
    {
      "epoch": 4.388079470198676,
      "grad_norm": 1.1505698007601282,
      "learning_rate": 7.971026607525014e-05,
      "loss": 0.8516,
      "step": 6626
    },
    {
      "epoch": 4.388741721854305,
      "grad_norm": 1.313791013621543,
      "learning_rate": 7.965130734012694e-05,
      "loss": 0.9922,
      "step": 6627
    },
    {
      "epoch": 4.389403973509934,
      "grad_norm": 1.3352336041295405,
      "learning_rate": 7.959236253485127e-05,
      "loss": 1.0312,
      "step": 6628
    },
    {
      "epoch": 4.390066225165563,
      "grad_norm": 1.2759967159200156,
      "learning_rate": 7.953343167109514e-05,
      "loss": 0.9805,
      "step": 6629
    },
    {
      "epoch": 4.390728476821192,
      "grad_norm": 1.0865507842986137,
      "learning_rate": 7.94745147605274e-05,
      "loss": 0.7344,
      "step": 6630
    },
    {
      "epoch": 4.391390728476821,
      "grad_norm": 1.0999034418079447,
      "learning_rate": 7.941561181481425e-05,
      "loss": 0.8125,
      "step": 6631
    },
    {
      "epoch": 4.39205298013245,
      "grad_norm": 1.1413390732396653,
      "learning_rate": 7.935672284561917e-05,
      "loss": 0.8438,
      "step": 6632
    },
    {
      "epoch": 4.392715231788079,
      "grad_norm": 1.068223005112426,
      "learning_rate": 7.929784786460293e-05,
      "loss": 0.793,
      "step": 6633
    },
    {
      "epoch": 4.393377483443708,
      "grad_norm": 1.2246871465100333,
      "learning_rate": 7.923898688342333e-05,
      "loss": 0.9688,
      "step": 6634
    },
    {
      "epoch": 4.394039735099338,
      "grad_norm": 1.052563400200038,
      "learning_rate": 7.918013991373562e-05,
      "loss": 0.7734,
      "step": 6635
    },
    {
      "epoch": 4.394701986754967,
      "grad_norm": 1.1681981956150516,
      "learning_rate": 7.912130696719208e-05,
      "loss": 0.8398,
      "step": 6636
    },
    {
      "epoch": 4.395364238410596,
      "grad_norm": 1.0091363390676518,
      "learning_rate": 7.906248805544236e-05,
      "loss": 0.707,
      "step": 6637
    },
    {
      "epoch": 4.396026490066225,
      "grad_norm": 1.0485955575230692,
      "learning_rate": 7.900368319013333e-05,
      "loss": 0.8359,
      "step": 6638
    },
    {
      "epoch": 4.396688741721855,
      "grad_norm": 1.1305418995374719,
      "learning_rate": 7.894489238290899e-05,
      "loss": 0.7969,
      "step": 6639
    },
    {
      "epoch": 4.397350993377484,
      "grad_norm": 1.2176688685773693,
      "learning_rate": 7.88861156454105e-05,
      "loss": 0.9766,
      "step": 6640
    },
    {
      "epoch": 4.398013245033113,
      "grad_norm": 1.0787966029316884,
      "learning_rate": 7.882735298927642e-05,
      "loss": 0.8086,
      "step": 6641
    },
    {
      "epoch": 4.398675496688742,
      "grad_norm": 1.0972246243590105,
      "learning_rate": 7.876860442614247e-05,
      "loss": 0.75,
      "step": 6642
    },
    {
      "epoch": 4.3993377483443705,
      "grad_norm": 1.138106355840279,
      "learning_rate": 7.870986996764149e-05,
      "loss": 0.8047,
      "step": 6643
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.2784869159372187,
      "learning_rate": 7.86511496254035e-05,
      "loss": 1.0078,
      "step": 6644
    },
    {
      "epoch": 4.400662251655629,
      "grad_norm": 1.1484106856217897,
      "learning_rate": 7.859244341105589e-05,
      "loss": 0.7969,
      "step": 6645
    },
    {
      "epoch": 4.401324503311258,
      "grad_norm": 1.2472132481840061,
      "learning_rate": 7.853375133622319e-05,
      "loss": 0.8477,
      "step": 6646
    },
    {
      "epoch": 4.401986754966887,
      "grad_norm": 1.1231491118048726,
      "learning_rate": 7.8475073412527e-05,
      "loss": 0.793,
      "step": 6647
    },
    {
      "epoch": 4.402649006622516,
      "grad_norm": 1.3065627736935417,
      "learning_rate": 7.841640965158634e-05,
      "loss": 0.9648,
      "step": 6648
    },
    {
      "epoch": 4.403311258278146,
      "grad_norm": 1.2019108809751762,
      "learning_rate": 7.83577600650172e-05,
      "loss": 0.8359,
      "step": 6649
    },
    {
      "epoch": 4.403973509933775,
      "grad_norm": 1.2282193266351717,
      "learning_rate": 7.829912466443291e-05,
      "loss": 0.8516,
      "step": 6650
    },
    {
      "epoch": 4.404635761589404,
      "grad_norm": 1.412753338831306,
      "learning_rate": 7.824050346144404e-05,
      "loss": 1.1719,
      "step": 6651
    },
    {
      "epoch": 4.405298013245033,
      "grad_norm": 1.1599696847232894,
      "learning_rate": 7.818189646765808e-05,
      "loss": 0.9492,
      "step": 6652
    },
    {
      "epoch": 4.405960264900663,
      "grad_norm": 1.187468974939121,
      "learning_rate": 7.812330369468002e-05,
      "loss": 0.9297,
      "step": 6653
    },
    {
      "epoch": 4.406622516556292,
      "grad_norm": 1.3430597505103727,
      "learning_rate": 7.80647251541119e-05,
      "loss": 1.0234,
      "step": 6654
    },
    {
      "epoch": 4.4072847682119205,
      "grad_norm": 1.0254189971110994,
      "learning_rate": 7.800616085755294e-05,
      "loss": 0.7695,
      "step": 6655
    },
    {
      "epoch": 4.407947019867549,
      "grad_norm": 1.0755704577343002,
      "learning_rate": 7.794761081659945e-05,
      "loss": 0.8633,
      "step": 6656
    },
    {
      "epoch": 4.408609271523179,
      "grad_norm": 1.206075096432033,
      "learning_rate": 7.788907504284505e-05,
      "loss": 0.9297,
      "step": 6657
    },
    {
      "epoch": 4.409271523178808,
      "grad_norm": 1.052875857059609,
      "learning_rate": 7.78305535478806e-05,
      "loss": 0.8242,
      "step": 6658
    },
    {
      "epoch": 4.409933774834437,
      "grad_norm": 1.0022569107733932,
      "learning_rate": 7.777204634329387e-05,
      "loss": 0.7422,
      "step": 6659
    },
    {
      "epoch": 4.410596026490066,
      "grad_norm": 1.0823396140997483,
      "learning_rate": 7.771355344067009e-05,
      "loss": 0.8594,
      "step": 6660
    },
    {
      "epoch": 4.411258278145695,
      "grad_norm": 1.0039750163776349,
      "learning_rate": 7.765507485159141e-05,
      "loss": 0.8242,
      "step": 6661
    },
    {
      "epoch": 4.411920529801325,
      "grad_norm": 1.2133633019107544,
      "learning_rate": 7.759661058763732e-05,
      "loss": 0.9258,
      "step": 6662
    },
    {
      "epoch": 4.412582781456954,
      "grad_norm": 1.2276589996590845,
      "learning_rate": 7.753816066038447e-05,
      "loss": 0.9219,
      "step": 6663
    },
    {
      "epoch": 4.413245033112583,
      "grad_norm": 1.1427179977175665,
      "learning_rate": 7.747972508140653e-05,
      "loss": 0.8281,
      "step": 6664
    },
    {
      "epoch": 4.413907284768212,
      "grad_norm": 1.014641238352878,
      "learning_rate": 7.742130386227436e-05,
      "loss": 0.7578,
      "step": 6665
    },
    {
      "epoch": 4.414569536423841,
      "grad_norm": 1.2518525749027354,
      "learning_rate": 7.736289701455622e-05,
      "loss": 0.8555,
      "step": 6666
    },
    {
      "epoch": 4.4152317880794705,
      "grad_norm": 1.00221930214085,
      "learning_rate": 7.730450454981723e-05,
      "loss": 0.75,
      "step": 6667
    },
    {
      "epoch": 4.415894039735099,
      "grad_norm": 1.1249399553512136,
      "learning_rate": 7.724612647961969e-05,
      "loss": 0.8945,
      "step": 6668
    },
    {
      "epoch": 4.416556291390728,
      "grad_norm": 1.22816578943164,
      "learning_rate": 7.718776281552321e-05,
      "loss": 0.8789,
      "step": 6669
    },
    {
      "epoch": 4.417218543046357,
      "grad_norm": 1.1337125458059392,
      "learning_rate": 7.712941356908454e-05,
      "loss": 0.7578,
      "step": 6670
    },
    {
      "epoch": 4.417880794701987,
      "grad_norm": 1.3149060329253341,
      "learning_rate": 7.707107875185734e-05,
      "loss": 0.9375,
      "step": 6671
    },
    {
      "epoch": 4.418543046357616,
      "grad_norm": 1.091746881321148,
      "learning_rate": 7.70127583753927e-05,
      "loss": 0.7852,
      "step": 6672
    },
    {
      "epoch": 4.419205298013245,
      "grad_norm": 1.3320524986648246,
      "learning_rate": 7.695445245123863e-05,
      "loss": 1.0391,
      "step": 6673
    },
    {
      "epoch": 4.419867549668874,
      "grad_norm": 1.1141833598775863,
      "learning_rate": 7.68961609909404e-05,
      "loss": 0.8242,
      "step": 6674
    },
    {
      "epoch": 4.420529801324503,
      "grad_norm": 1.1511271788463762,
      "learning_rate": 7.683788400604047e-05,
      "loss": 0.875,
      "step": 6675
    },
    {
      "epoch": 4.421192052980133,
      "grad_norm": 1.0550247354866535,
      "learning_rate": 7.67796215080783e-05,
      "loss": 0.8125,
      "step": 6676
    },
    {
      "epoch": 4.421854304635762,
      "grad_norm": 1.0532269711721365,
      "learning_rate": 7.672137350859043e-05,
      "loss": 0.8164,
      "step": 6677
    },
    {
      "epoch": 4.422516556291391,
      "grad_norm": 1.2058639555401087,
      "learning_rate": 7.666314001911073e-05,
      "loss": 0.8828,
      "step": 6678
    },
    {
      "epoch": 4.42317880794702,
      "grad_norm": 1.3803461597785633,
      "learning_rate": 7.660492105117015e-05,
      "loss": 1.1562,
      "step": 6679
    },
    {
      "epoch": 4.423841059602649,
      "grad_norm": 1.1376579035350038,
      "learning_rate": 7.654671661629657e-05,
      "loss": 0.9883,
      "step": 6680
    },
    {
      "epoch": 4.424503311258278,
      "grad_norm": 1.1803130353260232,
      "learning_rate": 7.648852672601529e-05,
      "loss": 0.8789,
      "step": 6681
    },
    {
      "epoch": 4.425165562913907,
      "grad_norm": 1.224438180548455,
      "learning_rate": 7.643035139184842e-05,
      "loss": 0.9883,
      "step": 6682
    },
    {
      "epoch": 4.425827814569536,
      "grad_norm": 1.1047768340042923,
      "learning_rate": 7.637219062531545e-05,
      "loss": 0.8008,
      "step": 6683
    },
    {
      "epoch": 4.426490066225165,
      "grad_norm": 0.9131842276168701,
      "learning_rate": 7.63140444379329e-05,
      "loss": 0.75,
      "step": 6684
    },
    {
      "epoch": 4.427152317880795,
      "grad_norm": 1.0965744953084597,
      "learning_rate": 7.625591284121433e-05,
      "loss": 0.8828,
      "step": 6685
    },
    {
      "epoch": 4.427814569536424,
      "grad_norm": 1.1492240174604782,
      "learning_rate": 7.619779584667035e-05,
      "loss": 0.918,
      "step": 6686
    },
    {
      "epoch": 4.428476821192053,
      "grad_norm": 1.2232356648972063,
      "learning_rate": 7.613969346580902e-05,
      "loss": 0.9922,
      "step": 6687
    },
    {
      "epoch": 4.429139072847682,
      "grad_norm": 1.1978655299990777,
      "learning_rate": 7.608160571013516e-05,
      "loss": 0.8711,
      "step": 6688
    },
    {
      "epoch": 4.429801324503312,
      "grad_norm": 0.9964243345615611,
      "learning_rate": 7.602353259115079e-05,
      "loss": 0.7852,
      "step": 6689
    },
    {
      "epoch": 4.430463576158941,
      "grad_norm": 1.0671894519135996,
      "learning_rate": 7.596547412035505e-05,
      "loss": 0.8242,
      "step": 6690
    },
    {
      "epoch": 4.4311258278145695,
      "grad_norm": 1.084647888872076,
      "learning_rate": 7.590743030924428e-05,
      "loss": 0.793,
      "step": 6691
    },
    {
      "epoch": 4.4317880794701985,
      "grad_norm": 1.0882801746530415,
      "learning_rate": 7.584940116931168e-05,
      "loss": 0.8047,
      "step": 6692
    },
    {
      "epoch": 4.432450331125827,
      "grad_norm": 1.1735188989127194,
      "learning_rate": 7.579138671204783e-05,
      "loss": 0.9141,
      "step": 6693
    },
    {
      "epoch": 4.433112582781457,
      "grad_norm": 1.0749665880380326,
      "learning_rate": 7.573338694894012e-05,
      "loss": 0.8438,
      "step": 6694
    },
    {
      "epoch": 4.433774834437086,
      "grad_norm": 1.0690543798102934,
      "learning_rate": 7.567540189147324e-05,
      "loss": 0.7656,
      "step": 6695
    },
    {
      "epoch": 4.434437086092715,
      "grad_norm": 1.1253094998014919,
      "learning_rate": 7.561743155112894e-05,
      "loss": 0.9102,
      "step": 6696
    },
    {
      "epoch": 4.435099337748344,
      "grad_norm": 1.127828035078651,
      "learning_rate": 7.555947593938596e-05,
      "loss": 0.7188,
      "step": 6697
    },
    {
      "epoch": 4.435761589403974,
      "grad_norm": 1.0916928110573747,
      "learning_rate": 7.550153506772015e-05,
      "loss": 0.8242,
      "step": 6698
    },
    {
      "epoch": 4.436423841059603,
      "grad_norm": 1.0912448228023155,
      "learning_rate": 7.544360894760443e-05,
      "loss": 0.8945,
      "step": 6699
    },
    {
      "epoch": 4.437086092715232,
      "grad_norm": 1.2017666477349516,
      "learning_rate": 7.538569759050901e-05,
      "loss": 0.8711,
      "step": 6700
    },
    {
      "epoch": 4.437748344370861,
      "grad_norm": 1.1768409939976037,
      "learning_rate": 7.532780100790088e-05,
      "loss": 0.8711,
      "step": 6701
    },
    {
      "epoch": 4.43841059602649,
      "grad_norm": 1.0554871421196965,
      "learning_rate": 7.526991921124418e-05,
      "loss": 0.8086,
      "step": 6702
    },
    {
      "epoch": 4.4390728476821195,
      "grad_norm": 1.0427813721943868,
      "learning_rate": 7.521205221200022e-05,
      "loss": 0.7461,
      "step": 6703
    },
    {
      "epoch": 4.4397350993377485,
      "grad_norm": 1.272818121417988,
      "learning_rate": 7.51542000216274e-05,
      "loss": 0.9023,
      "step": 6704
    },
    {
      "epoch": 4.440397350993377,
      "grad_norm": 1.1329384483991136,
      "learning_rate": 7.509636265158102e-05,
      "loss": 0.8867,
      "step": 6705
    },
    {
      "epoch": 4.441059602649006,
      "grad_norm": 1.0613040880330014,
      "learning_rate": 7.503854011331355e-05,
      "loss": 0.6953,
      "step": 6706
    },
    {
      "epoch": 4.441721854304636,
      "grad_norm": 1.114863945118355,
      "learning_rate": 7.49807324182746e-05,
      "loss": 0.7969,
      "step": 6707
    },
    {
      "epoch": 4.442384105960265,
      "grad_norm": 1.1381658912110388,
      "learning_rate": 7.492293957791066e-05,
      "loss": 0.8203,
      "step": 6708
    },
    {
      "epoch": 4.443046357615894,
      "grad_norm": 1.235176444381562,
      "learning_rate": 7.486516160366546e-05,
      "loss": 0.957,
      "step": 6709
    },
    {
      "epoch": 4.443708609271523,
      "grad_norm": 1.1615944496132282,
      "learning_rate": 7.480739850697959e-05,
      "loss": 0.8672,
      "step": 6710
    },
    {
      "epoch": 4.444370860927152,
      "grad_norm": 1.1004126624846313,
      "learning_rate": 7.474965029929088e-05,
      "loss": 0.8398,
      "step": 6711
    },
    {
      "epoch": 4.445033112582782,
      "grad_norm": 1.1926545309136802,
      "learning_rate": 7.469191699203419e-05,
      "loss": 0.8477,
      "step": 6712
    },
    {
      "epoch": 4.445695364238411,
      "grad_norm": 1.0778019110793824,
      "learning_rate": 7.46341985966413e-05,
      "loss": 0.7383,
      "step": 6713
    },
    {
      "epoch": 4.44635761589404,
      "grad_norm": 0.9834189626443061,
      "learning_rate": 7.457649512454108e-05,
      "loss": 0.6797,
      "step": 6714
    },
    {
      "epoch": 4.447019867549669,
      "grad_norm": 1.1704619249978323,
      "learning_rate": 7.451880658715952e-05,
      "loss": 0.8398,
      "step": 6715
    },
    {
      "epoch": 4.447682119205298,
      "grad_norm": 1.2497441038757344,
      "learning_rate": 7.446113299591969e-05,
      "loss": 0.9844,
      "step": 6716
    },
    {
      "epoch": 4.448344370860927,
      "grad_norm": 1.074660948360414,
      "learning_rate": 7.44034743622415e-05,
      "loss": 0.8281,
      "step": 6717
    },
    {
      "epoch": 4.449006622516556,
      "grad_norm": 1.106230103840178,
      "learning_rate": 7.43458306975421e-05,
      "loss": 0.7344,
      "step": 6718
    },
    {
      "epoch": 4.449668874172185,
      "grad_norm": 1.1566431650891433,
      "learning_rate": 7.428820201323553e-05,
      "loss": 0.9141,
      "step": 6719
    },
    {
      "epoch": 4.450331125827814,
      "grad_norm": 1.2154256666393897,
      "learning_rate": 7.423058832073298e-05,
      "loss": 0.8828,
      "step": 6720
    },
    {
      "epoch": 4.450993377483444,
      "grad_norm": 1.077345601887816,
      "learning_rate": 7.417298963144267e-05,
      "loss": 0.7969,
      "step": 6721
    },
    {
      "epoch": 4.451655629139073,
      "grad_norm": 1.1735585177017243,
      "learning_rate": 7.411540595676971e-05,
      "loss": 0.9062,
      "step": 6722
    },
    {
      "epoch": 4.452317880794702,
      "grad_norm": 1.0859617178511285,
      "learning_rate": 7.405783730811627e-05,
      "loss": 0.7266,
      "step": 6723
    },
    {
      "epoch": 4.452980132450331,
      "grad_norm": 1.1476941794566047,
      "learning_rate": 7.400028369688179e-05,
      "loss": 0.8359,
      "step": 6724
    },
    {
      "epoch": 4.45364238410596,
      "grad_norm": 1.1583232283756104,
      "learning_rate": 7.394274513446245e-05,
      "loss": 0.8438,
      "step": 6725
    },
    {
      "epoch": 4.45430463576159,
      "grad_norm": 1.197941225233906,
      "learning_rate": 7.388522163225148e-05,
      "loss": 0.8672,
      "step": 6726
    },
    {
      "epoch": 4.454966887417219,
      "grad_norm": 1.1234075819594622,
      "learning_rate": 7.382771320163928e-05,
      "loss": 0.832,
      "step": 6727
    },
    {
      "epoch": 4.4556291390728475,
      "grad_norm": 1.027335912731799,
      "learning_rate": 7.377021985401319e-05,
      "loss": 0.6875,
      "step": 6728
    },
    {
      "epoch": 4.4562913907284765,
      "grad_norm": 1.2154460488969059,
      "learning_rate": 7.371274160075745e-05,
      "loss": 0.9062,
      "step": 6729
    },
    {
      "epoch": 4.456953642384106,
      "grad_norm": 1.2811001253837753,
      "learning_rate": 7.365527845325355e-05,
      "loss": 0.9375,
      "step": 6730
    },
    {
      "epoch": 4.457615894039735,
      "grad_norm": 1.1846120451804887,
      "learning_rate": 7.359783042287972e-05,
      "loss": 0.793,
      "step": 6731
    },
    {
      "epoch": 4.458278145695364,
      "grad_norm": 1.3222825296284855,
      "learning_rate": 7.354039752101138e-05,
      "loss": 1.0547,
      "step": 6732
    },
    {
      "epoch": 4.458940397350993,
      "grad_norm": 1.2269381128740722,
      "learning_rate": 7.348297975902098e-05,
      "loss": 0.9062,
      "step": 6733
    },
    {
      "epoch": 4.459602649006623,
      "grad_norm": 1.1585013450316355,
      "learning_rate": 7.342557714827781e-05,
      "loss": 0.8555,
      "step": 6734
    },
    {
      "epoch": 4.460264900662252,
      "grad_norm": 1.1983171116815126,
      "learning_rate": 7.336818970014823e-05,
      "loss": 0.8633,
      "step": 6735
    },
    {
      "epoch": 4.460927152317881,
      "grad_norm": 1.0940528473858318,
      "learning_rate": 7.331081742599564e-05,
      "loss": 0.8008,
      "step": 6736
    },
    {
      "epoch": 4.46158940397351,
      "grad_norm": 1.028406584522927,
      "learning_rate": 7.325346033718044e-05,
      "loss": 0.668,
      "step": 6737
    },
    {
      "epoch": 4.462251655629139,
      "grad_norm": 1.1110168917534264,
      "learning_rate": 7.319611844505995e-05,
      "loss": 0.7852,
      "step": 6738
    },
    {
      "epoch": 4.4629139072847686,
      "grad_norm": 1.179180020674184,
      "learning_rate": 7.313879176098856e-05,
      "loss": 0.9531,
      "step": 6739
    },
    {
      "epoch": 4.4635761589403975,
      "grad_norm": 1.200395621021466,
      "learning_rate": 7.308148029631754e-05,
      "loss": 0.7773,
      "step": 6740
    },
    {
      "epoch": 4.464238410596026,
      "grad_norm": 1.0868604121971879,
      "learning_rate": 7.302418406239526e-05,
      "loss": 0.7344,
      "step": 6741
    },
    {
      "epoch": 4.464900662251655,
      "grad_norm": 1.262812488573143,
      "learning_rate": 7.29669030705671e-05,
      "loss": 0.9805,
      "step": 6742
    },
    {
      "epoch": 4.465562913907284,
      "grad_norm": 1.0886762113492363,
      "learning_rate": 7.290963733217522e-05,
      "loss": 0.7812,
      "step": 6743
    },
    {
      "epoch": 4.466225165562914,
      "grad_norm": 1.0936529973684774,
      "learning_rate": 7.285238685855897e-05,
      "loss": 0.8281,
      "step": 6744
    },
    {
      "epoch": 4.466887417218543,
      "grad_norm": 1.0590468483077213,
      "learning_rate": 7.279515166105463e-05,
      "loss": 0.8438,
      "step": 6745
    },
    {
      "epoch": 4.467549668874172,
      "grad_norm": 1.0984270583103386,
      "learning_rate": 7.273793175099541e-05,
      "loss": 0.8516,
      "step": 6746
    },
    {
      "epoch": 4.468211920529801,
      "grad_norm": 1.0644782168455313,
      "learning_rate": 7.268072713971141e-05,
      "loss": 0.7539,
      "step": 6747
    },
    {
      "epoch": 4.468874172185431,
      "grad_norm": 1.1834004448101916,
      "learning_rate": 7.262353783852988e-05,
      "loss": 0.9336,
      "step": 6748
    },
    {
      "epoch": 4.46953642384106,
      "grad_norm": 1.0951116809295849,
      "learning_rate": 7.256636385877502e-05,
      "loss": 0.8477,
      "step": 6749
    },
    {
      "epoch": 4.470198675496689,
      "grad_norm": 1.1975272090347655,
      "learning_rate": 7.250920521176778e-05,
      "loss": 0.8125,
      "step": 6750
    },
    {
      "epoch": 4.470860927152318,
      "grad_norm": 1.1658426534888067,
      "learning_rate": 7.245206190882639e-05,
      "loss": 0.8477,
      "step": 6751
    },
    {
      "epoch": 4.4715231788079475,
      "grad_norm": 1.137610707085754,
      "learning_rate": 7.239493396126572e-05,
      "loss": 0.7617,
      "step": 6752
    },
    {
      "epoch": 4.472185430463576,
      "grad_norm": 1.0951370152577016,
      "learning_rate": 7.233782138039784e-05,
      "loss": 0.7656,
      "step": 6753
    },
    {
      "epoch": 4.472847682119205,
      "grad_norm": 1.0504830303291108,
      "learning_rate": 7.228072417753179e-05,
      "loss": 0.7734,
      "step": 6754
    },
    {
      "epoch": 4.473509933774834,
      "grad_norm": 1.1675586275576926,
      "learning_rate": 7.222364236397334e-05,
      "loss": 0.8398,
      "step": 6755
    },
    {
      "epoch": 4.474172185430463,
      "grad_norm": 1.3056239946337038,
      "learning_rate": 7.216657595102533e-05,
      "loss": 0.9141,
      "step": 6756
    },
    {
      "epoch": 4.474834437086093,
      "grad_norm": 1.1389062157913172,
      "learning_rate": 7.210952494998762e-05,
      "loss": 0.8633,
      "step": 6757
    },
    {
      "epoch": 4.475496688741722,
      "grad_norm": 1.0760396513719077,
      "learning_rate": 7.2052489372157e-05,
      "loss": 0.6875,
      "step": 6758
    },
    {
      "epoch": 4.476158940397351,
      "grad_norm": 1.2624964724145045,
      "learning_rate": 7.199546922882713e-05,
      "loss": 0.9336,
      "step": 6759
    },
    {
      "epoch": 4.47682119205298,
      "grad_norm": 1.2817732428302149,
      "learning_rate": 7.193846453128852e-05,
      "loss": 0.9141,
      "step": 6760
    },
    {
      "epoch": 4.477483443708609,
      "grad_norm": 1.1028212360103709,
      "learning_rate": 7.1881475290829e-05,
      "loss": 0.7461,
      "step": 6761
    },
    {
      "epoch": 4.478145695364239,
      "grad_norm": 1.050629603175401,
      "learning_rate": 7.182450151873298e-05,
      "loss": 0.7812,
      "step": 6762
    },
    {
      "epoch": 4.478807947019868,
      "grad_norm": 1.285334177468639,
      "learning_rate": 7.176754322628185e-05,
      "loss": 0.9531,
      "step": 6763
    },
    {
      "epoch": 4.479470198675497,
      "grad_norm": 1.421649903462375,
      "learning_rate": 7.171060042475407e-05,
      "loss": 1.0469,
      "step": 6764
    },
    {
      "epoch": 4.4801324503311255,
      "grad_norm": 1.1753486887616844,
      "learning_rate": 7.165367312542503e-05,
      "loss": 0.8555,
      "step": 6765
    },
    {
      "epoch": 4.480794701986755,
      "grad_norm": 1.0959612777445915,
      "learning_rate": 7.159676133956685e-05,
      "loss": 0.7695,
      "step": 6766
    },
    {
      "epoch": 4.481456953642384,
      "grad_norm": 1.1437403953705947,
      "learning_rate": 7.153986507844884e-05,
      "loss": 0.832,
      "step": 6767
    },
    {
      "epoch": 4.482119205298013,
      "grad_norm": 1.0495238418550779,
      "learning_rate": 7.148298435333702e-05,
      "loss": 0.8125,
      "step": 6768
    },
    {
      "epoch": 4.482781456953642,
      "grad_norm": 0.9966713538813938,
      "learning_rate": 7.142611917549447e-05,
      "loss": 0.7617,
      "step": 6769
    },
    {
      "epoch": 4.483443708609272,
      "grad_norm": 1.2295140365073625,
      "learning_rate": 7.136926955618118e-05,
      "loss": 0.8672,
      "step": 6770
    },
    {
      "epoch": 4.484105960264901,
      "grad_norm": 1.1404065341567566,
      "learning_rate": 7.1312435506654e-05,
      "loss": 0.9219,
      "step": 6771
    },
    {
      "epoch": 4.48476821192053,
      "grad_norm": 1.0344974967528306,
      "learning_rate": 7.125561703816666e-05,
      "loss": 0.75,
      "step": 6772
    },
    {
      "epoch": 4.485430463576159,
      "grad_norm": 1.1612619473265504,
      "learning_rate": 7.119881416196993e-05,
      "loss": 0.7812,
      "step": 6773
    },
    {
      "epoch": 4.486092715231788,
      "grad_norm": 1.2940545465555338,
      "learning_rate": 7.114202688931146e-05,
      "loss": 0.9766,
      "step": 6774
    },
    {
      "epoch": 4.486754966887418,
      "grad_norm": 1.2044969788127284,
      "learning_rate": 7.108525523143569e-05,
      "loss": 0.8281,
      "step": 6775
    },
    {
      "epoch": 4.4874172185430465,
      "grad_norm": 1.1729881396806463,
      "learning_rate": 7.102849919958417e-05,
      "loss": 0.8047,
      "step": 6776
    },
    {
      "epoch": 4.4880794701986755,
      "grad_norm": 1.1130272635739702,
      "learning_rate": 7.097175880499514e-05,
      "loss": 0.7344,
      "step": 6777
    },
    {
      "epoch": 4.488741721854304,
      "grad_norm": 1.1647908140578158,
      "learning_rate": 7.091503405890388e-05,
      "loss": 0.8828,
      "step": 6778
    },
    {
      "epoch": 4.489403973509933,
      "grad_norm": 1.1348591213611854,
      "learning_rate": 7.085832497254263e-05,
      "loss": 0.7812,
      "step": 6779
    },
    {
      "epoch": 4.490066225165563,
      "grad_norm": 1.2101827530364666,
      "learning_rate": 7.080163155714035e-05,
      "loss": 0.8633,
      "step": 6780
    },
    {
      "epoch": 4.490728476821192,
      "grad_norm": 1.3384052932108452,
      "learning_rate": 7.07449538239229e-05,
      "loss": 1.1484,
      "step": 6781
    },
    {
      "epoch": 4.491390728476821,
      "grad_norm": 1.1710742975056254,
      "learning_rate": 7.068829178411332e-05,
      "loss": 0.8672,
      "step": 6782
    },
    {
      "epoch": 4.49205298013245,
      "grad_norm": 1.1323765258272818,
      "learning_rate": 7.063164544893126e-05,
      "loss": 0.8398,
      "step": 6783
    },
    {
      "epoch": 4.49271523178808,
      "grad_norm": 1.2199841579769128,
      "learning_rate": 7.057501482959326e-05,
      "loss": 0.8633,
      "step": 6784
    },
    {
      "epoch": 4.493377483443709,
      "grad_norm": 1.0787597980090142,
      "learning_rate": 7.051839993731293e-05,
      "loss": 0.7539,
      "step": 6785
    },
    {
      "epoch": 4.494039735099338,
      "grad_norm": 1.2222719846194399,
      "learning_rate": 7.046180078330067e-05,
      "loss": 0.875,
      "step": 6786
    },
    {
      "epoch": 4.494701986754967,
      "grad_norm": 1.0968657304756657,
      "learning_rate": 7.040521737876367e-05,
      "loss": 0.8555,
      "step": 6787
    },
    {
      "epoch": 4.495364238410596,
      "grad_norm": 1.1333335591216855,
      "learning_rate": 7.034864973490622e-05,
      "loss": 0.8594,
      "step": 6788
    },
    {
      "epoch": 4.4960264900662255,
      "grad_norm": 1.1599165635864697,
      "learning_rate": 7.029209786292925e-05,
      "loss": 0.9219,
      "step": 6789
    },
    {
      "epoch": 4.496688741721854,
      "grad_norm": 0.9955952001499914,
      "learning_rate": 7.023556177403072e-05,
      "loss": 0.7656,
      "step": 6790
    },
    {
      "epoch": 4.497350993377483,
      "grad_norm": 1.1663865905621547,
      "learning_rate": 7.017904147940546e-05,
      "loss": 0.8789,
      "step": 6791
    },
    {
      "epoch": 4.498013245033112,
      "grad_norm": 1.083419428931796,
      "learning_rate": 7.012253699024513e-05,
      "loss": 0.8359,
      "step": 6792
    },
    {
      "epoch": 4.498675496688742,
      "grad_norm": 1.0119618537706674,
      "learning_rate": 7.006604831773817e-05,
      "loss": 0.8008,
      "step": 6793
    },
    {
      "epoch": 4.499337748344371,
      "grad_norm": 1.2306832357193778,
      "learning_rate": 7.000957547307006e-05,
      "loss": 0.9297,
      "step": 6794
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.2989987952271396,
      "learning_rate": 6.995311846742314e-05,
      "loss": 0.9297,
      "step": 6795
    },
    {
      "epoch": 4.500662251655629,
      "grad_norm": 1.0654184224292123,
      "learning_rate": 6.98966773119764e-05,
      "loss": 0.8008,
      "step": 6796
    },
    {
      "epoch": 4.501324503311258,
      "grad_norm": 1.085661653977342,
      "learning_rate": 6.98402520179059e-05,
      "loss": 0.7695,
      "step": 6797
    },
    {
      "epoch": 4.501986754966888,
      "grad_norm": 1.1960083274566207,
      "learning_rate": 6.978384259638457e-05,
      "loss": 0.9102,
      "step": 6798
    },
    {
      "epoch": 4.502649006622517,
      "grad_norm": 1.2433217542594028,
      "learning_rate": 6.972744905858199e-05,
      "loss": 0.9023,
      "step": 6799
    },
    {
      "epoch": 4.503311258278146,
      "grad_norm": 1.448583137717848,
      "learning_rate": 6.967107141566485e-05,
      "loss": 1.1641,
      "step": 6800
    },
    {
      "epoch": 4.5039735099337745,
      "grad_norm": 1.3465719533460638,
      "learning_rate": 6.961470967879646e-05,
      "loss": 1.0078,
      "step": 6801
    },
    {
      "epoch": 4.5046357615894035,
      "grad_norm": 1.4410126817808542,
      "learning_rate": 6.955836385913713e-05,
      "loss": 1.1406,
      "step": 6802
    },
    {
      "epoch": 4.505298013245033,
      "grad_norm": 1.3144020289736424,
      "learning_rate": 6.950203396784407e-05,
      "loss": 1.0156,
      "step": 6803
    },
    {
      "epoch": 4.505960264900662,
      "grad_norm": 1.1479416533864808,
      "learning_rate": 6.944572001607114e-05,
      "loss": 0.8359,
      "step": 6804
    },
    {
      "epoch": 4.506622516556291,
      "grad_norm": 1.0720422372856222,
      "learning_rate": 6.938942201496912e-05,
      "loss": 0.7539,
      "step": 6805
    },
    {
      "epoch": 4.50728476821192,
      "grad_norm": 1.085355694569158,
      "learning_rate": 6.933313997568571e-05,
      "loss": 0.793,
      "step": 6806
    },
    {
      "epoch": 4.50794701986755,
      "grad_norm": 1.0212772133518924,
      "learning_rate": 6.927687390936545e-05,
      "loss": 0.7578,
      "step": 6807
    },
    {
      "epoch": 4.508609271523179,
      "grad_norm": 1.1719389709700936,
      "learning_rate": 6.922062382714958e-05,
      "loss": 0.9883,
      "step": 6808
    },
    {
      "epoch": 4.509271523178808,
      "grad_norm": 1.1502693021802814,
      "learning_rate": 6.916438974017635e-05,
      "loss": 0.8672,
      "step": 6809
    },
    {
      "epoch": 4.509933774834437,
      "grad_norm": 1.1651513095123094,
      "learning_rate": 6.910817165958064e-05,
      "loss": 0.9336,
      "step": 6810
    },
    {
      "epoch": 4.510596026490067,
      "grad_norm": 1.2441480749041118,
      "learning_rate": 6.905196959649432e-05,
      "loss": 0.918,
      "step": 6811
    },
    {
      "epoch": 4.511258278145696,
      "grad_norm": 1.2381093188067378,
      "learning_rate": 6.899578356204614e-05,
      "loss": 0.8633,
      "step": 6812
    },
    {
      "epoch": 4.5119205298013245,
      "grad_norm": 1.0841895781012965,
      "learning_rate": 6.893961356736147e-05,
      "loss": 0.7891,
      "step": 6813
    },
    {
      "epoch": 4.5125827814569535,
      "grad_norm": 1.1547671189767337,
      "learning_rate": 6.888345962356261e-05,
      "loss": 0.875,
      "step": 6814
    },
    {
      "epoch": 4.513245033112582,
      "grad_norm": 1.1335647726174272,
      "learning_rate": 6.88273217417687e-05,
      "loss": 0.7891,
      "step": 6815
    },
    {
      "epoch": 4.513907284768212,
      "grad_norm": 1.0964364839001548,
      "learning_rate": 6.877119993309576e-05,
      "loss": 0.8047,
      "step": 6816
    },
    {
      "epoch": 4.514569536423841,
      "grad_norm": 1.1540367303330639,
      "learning_rate": 6.87150942086565e-05,
      "loss": 0.9297,
      "step": 6817
    },
    {
      "epoch": 4.51523178807947,
      "grad_norm": 1.166514057517789,
      "learning_rate": 6.865900457956037e-05,
      "loss": 0.8359,
      "step": 6818
    },
    {
      "epoch": 4.515894039735099,
      "grad_norm": 1.0669191495638843,
      "learning_rate": 6.8602931056914e-05,
      "loss": 0.8125,
      "step": 6819
    },
    {
      "epoch": 4.516556291390728,
      "grad_norm": 1.1939004968943268,
      "learning_rate": 6.854687365182045e-05,
      "loss": 0.793,
      "step": 6820
    },
    {
      "epoch": 4.517218543046358,
      "grad_norm": 0.9413070467062662,
      "learning_rate": 6.849083237537969e-05,
      "loss": 0.6172,
      "step": 6821
    },
    {
      "epoch": 4.517880794701987,
      "grad_norm": 1.1258481418357573,
      "learning_rate": 6.84348072386886e-05,
      "loss": 0.8398,
      "step": 6822
    },
    {
      "epoch": 4.518543046357616,
      "grad_norm": 1.109209439715087,
      "learning_rate": 6.837879825284084e-05,
      "loss": 0.8438,
      "step": 6823
    },
    {
      "epoch": 4.519205298013245,
      "grad_norm": 1.1066448070357282,
      "learning_rate": 6.832280542892673e-05,
      "loss": 0.793,
      "step": 6824
    },
    {
      "epoch": 4.5198675496688745,
      "grad_norm": 1.1019115341386592,
      "learning_rate": 6.826682877803361e-05,
      "loss": 0.7812,
      "step": 6825
    },
    {
      "epoch": 4.520529801324503,
      "grad_norm": 1.1156086113145143,
      "learning_rate": 6.821086831124536e-05,
      "loss": 0.832,
      "step": 6826
    },
    {
      "epoch": 4.521192052980132,
      "grad_norm": 1.0841584365570298,
      "learning_rate": 6.815492403964288e-05,
      "loss": 0.7578,
      "step": 6827
    },
    {
      "epoch": 4.521854304635761,
      "grad_norm": 1.1674269118156773,
      "learning_rate": 6.80989959743038e-05,
      "loss": 0.8828,
      "step": 6828
    },
    {
      "epoch": 4.522516556291391,
      "grad_norm": 1.1424777461113194,
      "learning_rate": 6.804308412630251e-05,
      "loss": 0.7617,
      "step": 6829
    },
    {
      "epoch": 4.52317880794702,
      "grad_norm": 1.0897139457485012,
      "learning_rate": 6.798718850671009e-05,
      "loss": 0.7461,
      "step": 6830
    },
    {
      "epoch": 4.523841059602649,
      "grad_norm": 1.231808967696738,
      "learning_rate": 6.79313091265946e-05,
      "loss": 0.9141,
      "step": 6831
    },
    {
      "epoch": 4.524503311258278,
      "grad_norm": 1.1684173144862366,
      "learning_rate": 6.787544599702085e-05,
      "loss": 0.8398,
      "step": 6832
    },
    {
      "epoch": 4.525165562913907,
      "grad_norm": 1.0647251146661088,
      "learning_rate": 6.781959912905025e-05,
      "loss": 0.7109,
      "step": 6833
    },
    {
      "epoch": 4.525827814569537,
      "grad_norm": 1.0726829984318629,
      "learning_rate": 6.776376853374125e-05,
      "loss": 0.7656,
      "step": 6834
    },
    {
      "epoch": 4.526490066225166,
      "grad_norm": 1.3724215844872567,
      "learning_rate": 6.770795422214884e-05,
      "loss": 0.9648,
      "step": 6835
    },
    {
      "epoch": 4.527152317880795,
      "grad_norm": 1.1579244860238105,
      "learning_rate": 6.765215620532494e-05,
      "loss": 0.9414,
      "step": 6836
    },
    {
      "epoch": 4.527814569536424,
      "grad_norm": 1.0666909039984787,
      "learning_rate": 6.759637449431825e-05,
      "loss": 0.7109,
      "step": 6837
    },
    {
      "epoch": 4.5284768211920525,
      "grad_norm": 1.074793224029001,
      "learning_rate": 6.754060910017409e-05,
      "loss": 0.6797,
      "step": 6838
    },
    {
      "epoch": 4.529139072847682,
      "grad_norm": 1.0457583706817897,
      "learning_rate": 6.748486003393469e-05,
      "loss": 0.7734,
      "step": 6839
    },
    {
      "epoch": 4.529801324503311,
      "grad_norm": 1.1347520835838734,
      "learning_rate": 6.742912730663905e-05,
      "loss": 0.9141,
      "step": 6840
    },
    {
      "epoch": 4.53046357615894,
      "grad_norm": 1.1944120611865274,
      "learning_rate": 6.737341092932287e-05,
      "loss": 0.9492,
      "step": 6841
    },
    {
      "epoch": 4.531125827814569,
      "grad_norm": 1.2344141096249022,
      "learning_rate": 6.731771091301854e-05,
      "loss": 0.8711,
      "step": 6842
    },
    {
      "epoch": 4.531788079470199,
      "grad_norm": 1.1339812460824181,
      "learning_rate": 6.726202726875538e-05,
      "loss": 0.7422,
      "step": 6843
    },
    {
      "epoch": 4.532450331125828,
      "grad_norm": 1.2217190290496982,
      "learning_rate": 6.720636000755943e-05,
      "loss": 0.8984,
      "step": 6844
    },
    {
      "epoch": 4.533112582781457,
      "grad_norm": 1.3173167164927722,
      "learning_rate": 6.715070914045333e-05,
      "loss": 1.0078,
      "step": 6845
    },
    {
      "epoch": 4.533774834437086,
      "grad_norm": 1.0258589336170454,
      "learning_rate": 6.709507467845673e-05,
      "loss": 0.7109,
      "step": 6846
    },
    {
      "epoch": 4.534437086092716,
      "grad_norm": 1.1636126402442104,
      "learning_rate": 6.703945663258575e-05,
      "loss": 0.7891,
      "step": 6847
    },
    {
      "epoch": 4.535099337748345,
      "grad_norm": 1.2037385659158286,
      "learning_rate": 6.698385501385347e-05,
      "loss": 0.8828,
      "step": 6848
    },
    {
      "epoch": 4.535761589403974,
      "grad_norm": 1.314627748759114,
      "learning_rate": 6.692826983326972e-05,
      "loss": 0.8828,
      "step": 6849
    },
    {
      "epoch": 4.5364238410596025,
      "grad_norm": 1.1871950037209105,
      "learning_rate": 6.687270110184091e-05,
      "loss": 0.9023,
      "step": 6850
    },
    {
      "epoch": 4.5370860927152314,
      "grad_norm": 1.3138599145713863,
      "learning_rate": 6.681714883057024e-05,
      "loss": 0.9805,
      "step": 6851
    },
    {
      "epoch": 4.537748344370861,
      "grad_norm": 1.0206689946922418,
      "learning_rate": 6.676161303045778e-05,
      "loss": 0.7383,
      "step": 6852
    },
    {
      "epoch": 4.53841059602649,
      "grad_norm": 1.1105534934872179,
      "learning_rate": 6.670609371250029e-05,
      "loss": 0.7812,
      "step": 6853
    },
    {
      "epoch": 4.539072847682119,
      "grad_norm": 1.2410765309943481,
      "learning_rate": 6.665059088769112e-05,
      "loss": 0.8984,
      "step": 6854
    },
    {
      "epoch": 4.539735099337748,
      "grad_norm": 1.1785640148949748,
      "learning_rate": 6.659510456702052e-05,
      "loss": 0.918,
      "step": 6855
    },
    {
      "epoch": 4.540397350993377,
      "grad_norm": 1.2171637470599097,
      "learning_rate": 6.653963476147549e-05,
      "loss": 0.9219,
      "step": 6856
    },
    {
      "epoch": 4.541059602649007,
      "grad_norm": 1.095911838517165,
      "learning_rate": 6.648418148203956e-05,
      "loss": 0.8125,
      "step": 6857
    },
    {
      "epoch": 4.541721854304636,
      "grad_norm": 0.9708478586338003,
      "learning_rate": 6.642874473969322e-05,
      "loss": 0.6875,
      "step": 6858
    },
    {
      "epoch": 4.542384105960265,
      "grad_norm": 1.0448915886735928,
      "learning_rate": 6.637332454541349e-05,
      "loss": 0.8047,
      "step": 6859
    },
    {
      "epoch": 4.543046357615894,
      "grad_norm": 1.0169702498477373,
      "learning_rate": 6.631792091017423e-05,
      "loss": 0.7852,
      "step": 6860
    },
    {
      "epoch": 4.5437086092715235,
      "grad_norm": 1.1482212478413842,
      "learning_rate": 6.626253384494608e-05,
      "loss": 0.7539,
      "step": 6861
    },
    {
      "epoch": 4.5443708609271525,
      "grad_norm": 1.1786362622834043,
      "learning_rate": 6.620716336069627e-05,
      "loss": 0.8984,
      "step": 6862
    },
    {
      "epoch": 4.545033112582781,
      "grad_norm": 1.2607232566072555,
      "learning_rate": 6.615180946838868e-05,
      "loss": 0.8711,
      "step": 6863
    },
    {
      "epoch": 4.54569536423841,
      "grad_norm": 1.253239580661758,
      "learning_rate": 6.609647217898412e-05,
      "loss": 0.9688,
      "step": 6864
    },
    {
      "epoch": 4.54635761589404,
      "grad_norm": 1.118434113834205,
      "learning_rate": 6.604115150344004e-05,
      "loss": 0.875,
      "step": 6865
    },
    {
      "epoch": 4.547019867549669,
      "grad_norm": 1.254557868386404,
      "learning_rate": 6.598584745271046e-05,
      "loss": 0.9531,
      "step": 6866
    },
    {
      "epoch": 4.547682119205298,
      "grad_norm": 1.2197448558170783,
      "learning_rate": 6.593056003774634e-05,
      "loss": 0.8867,
      "step": 6867
    },
    {
      "epoch": 4.548344370860927,
      "grad_norm": 1.2009430698081902,
      "learning_rate": 6.58752892694951e-05,
      "loss": 0.8555,
      "step": 6868
    },
    {
      "epoch": 4.549006622516556,
      "grad_norm": 1.0658656506151505,
      "learning_rate": 6.58200351589011e-05,
      "loss": 0.7344,
      "step": 6869
    },
    {
      "epoch": 4.549668874172186,
      "grad_norm": 1.1686946069001052,
      "learning_rate": 6.576479771690518e-05,
      "loss": 0.8164,
      "step": 6870
    },
    {
      "epoch": 4.550331125827815,
      "grad_norm": 1.123056721411962,
      "learning_rate": 6.570957695444508e-05,
      "loss": 0.9258,
      "step": 6871
    },
    {
      "epoch": 4.550993377483444,
      "grad_norm": 1.166466314338416,
      "learning_rate": 6.565437288245505e-05,
      "loss": 0.8555,
      "step": 6872
    },
    {
      "epoch": 4.551655629139073,
      "grad_norm": 1.1603772619032338,
      "learning_rate": 6.559918551186617e-05,
      "loss": 0.7617,
      "step": 6873
    },
    {
      "epoch": 4.552317880794702,
      "grad_norm": 1.044582636785393,
      "learning_rate": 6.554401485360625e-05,
      "loss": 0.7422,
      "step": 6874
    },
    {
      "epoch": 4.552980132450331,
      "grad_norm": 1.0568924394017138,
      "learning_rate": 6.548886091859966e-05,
      "loss": 0.8164,
      "step": 6875
    },
    {
      "epoch": 4.55364238410596,
      "grad_norm": 0.9340536546779169,
      "learning_rate": 6.543372371776735e-05,
      "loss": 0.7422,
      "step": 6876
    },
    {
      "epoch": 4.554304635761589,
      "grad_norm": 1.1000348977210492,
      "learning_rate": 6.53786032620274e-05,
      "loss": 0.8086,
      "step": 6877
    },
    {
      "epoch": 4.554966887417218,
      "grad_norm": 1.1191784108113767,
      "learning_rate": 6.532349956229414e-05,
      "loss": 0.8242,
      "step": 6878
    },
    {
      "epoch": 4.555629139072848,
      "grad_norm": 1.2077328747670735,
      "learning_rate": 6.526841262947872e-05,
      "loss": 0.8203,
      "step": 6879
    },
    {
      "epoch": 4.556291390728477,
      "grad_norm": 1.139683439202912,
      "learning_rate": 6.521334247448901e-05,
      "loss": 0.7656,
      "step": 6880
    },
    {
      "epoch": 4.556953642384106,
      "grad_norm": 1.0407816168972148,
      "learning_rate": 6.515828910822957e-05,
      "loss": 0.7422,
      "step": 6881
    },
    {
      "epoch": 4.557615894039735,
      "grad_norm": 1.1903400792587258,
      "learning_rate": 6.510325254160154e-05,
      "loss": 0.832,
      "step": 6882
    },
    {
      "epoch": 4.558278145695365,
      "grad_norm": 1.408375708416154,
      "learning_rate": 6.504823278550285e-05,
      "loss": 1.1094,
      "step": 6883
    },
    {
      "epoch": 4.558940397350994,
      "grad_norm": 1.246862266247841,
      "learning_rate": 6.499322985082795e-05,
      "loss": 0.8711,
      "step": 6884
    },
    {
      "epoch": 4.559602649006623,
      "grad_norm": 1.0271767698613408,
      "learning_rate": 6.493824374846813e-05,
      "loss": 0.6953,
      "step": 6885
    },
    {
      "epoch": 4.5602649006622515,
      "grad_norm": 1.176743044749555,
      "learning_rate": 6.48832744893113e-05,
      "loss": 0.8594,
      "step": 6886
    },
    {
      "epoch": 4.5609271523178805,
      "grad_norm": 1.6108906824081386,
      "learning_rate": 6.482832208424195e-05,
      "loss": 1.2031,
      "step": 6887
    },
    {
      "epoch": 4.56158940397351,
      "grad_norm": 1.0828732377861132,
      "learning_rate": 6.477338654414122e-05,
      "loss": 0.7266,
      "step": 6888
    },
    {
      "epoch": 4.562251655629139,
      "grad_norm": 1.003110770144394,
      "learning_rate": 6.471846787988707e-05,
      "loss": 0.6797,
      "step": 6889
    },
    {
      "epoch": 4.562913907284768,
      "grad_norm": 0.9914673537821227,
      "learning_rate": 6.466356610235403e-05,
      "loss": 0.7188,
      "step": 6890
    },
    {
      "epoch": 4.563576158940397,
      "grad_norm": 1.1746249355854508,
      "learning_rate": 6.460868122241321e-05,
      "loss": 0.8906,
      "step": 6891
    },
    {
      "epoch": 4.564238410596026,
      "grad_norm": 1.2805054146315866,
      "learning_rate": 6.45538132509325e-05,
      "loss": 0.9766,
      "step": 6892
    },
    {
      "epoch": 4.564900662251656,
      "grad_norm": 1.0000477651904094,
      "learning_rate": 6.449896219877642e-05,
      "loss": 0.7148,
      "step": 6893
    },
    {
      "epoch": 4.565562913907285,
      "grad_norm": 1.0863621330245323,
      "learning_rate": 6.444412807680599e-05,
      "loss": 0.8555,
      "step": 6894
    },
    {
      "epoch": 4.566225165562914,
      "grad_norm": 1.1008891132711685,
      "learning_rate": 6.438931089587913e-05,
      "loss": 0.8008,
      "step": 6895
    },
    {
      "epoch": 4.566887417218543,
      "grad_norm": 1.123672323139189,
      "learning_rate": 6.433451066685017e-05,
      "loss": 0.8008,
      "step": 6896
    },
    {
      "epoch": 4.567549668874172,
      "grad_norm": 1.156650678347421,
      "learning_rate": 6.42797274005702e-05,
      "loss": 0.8516,
      "step": 6897
    },
    {
      "epoch": 4.5682119205298015,
      "grad_norm": 1.044532393397309,
      "learning_rate": 6.422496110788702e-05,
      "loss": 0.6992,
      "step": 6898
    },
    {
      "epoch": 4.5688741721854305,
      "grad_norm": 1.3027445811147818,
      "learning_rate": 6.417021179964491e-05,
      "loss": 0.8906,
      "step": 6899
    },
    {
      "epoch": 4.569536423841059,
      "grad_norm": 1.3392551441256109,
      "learning_rate": 6.411547948668483e-05,
      "loss": 0.9258,
      "step": 6900
    },
    {
      "epoch": 4.570198675496689,
      "grad_norm": 1.108687095495432,
      "learning_rate": 6.406076417984444e-05,
      "loss": 0.793,
      "step": 6901
    },
    {
      "epoch": 4.570860927152318,
      "grad_norm": 1.1370684483777989,
      "learning_rate": 6.400606588995805e-05,
      "loss": 0.875,
      "step": 6902
    },
    {
      "epoch": 4.571523178807947,
      "grad_norm": 1.0704211262512364,
      "learning_rate": 6.395138462785647e-05,
      "loss": 0.7461,
      "step": 6903
    },
    {
      "epoch": 4.572185430463576,
      "grad_norm": 1.3932304684590486,
      "learning_rate": 6.389672040436729e-05,
      "loss": 1.0859,
      "step": 6904
    },
    {
      "epoch": 4.572847682119205,
      "grad_norm": 1.1987664205999906,
      "learning_rate": 6.384207323031456e-05,
      "loss": 0.9648,
      "step": 6905
    },
    {
      "epoch": 4.573509933774835,
      "grad_norm": 1.1741022043078602,
      "learning_rate": 6.378744311651908e-05,
      "loss": 0.8477,
      "step": 6906
    },
    {
      "epoch": 4.574172185430464,
      "grad_norm": 1.1480275997435436,
      "learning_rate": 6.373283007379833e-05,
      "loss": 0.7695,
      "step": 6907
    },
    {
      "epoch": 4.574834437086093,
      "grad_norm": 1.165518318396098,
      "learning_rate": 6.367823411296625e-05,
      "loss": 0.8086,
      "step": 6908
    },
    {
      "epoch": 4.575496688741722,
      "grad_norm": 1.003600885806037,
      "learning_rate": 6.362365524483332e-05,
      "loss": 0.7656,
      "step": 6909
    },
    {
      "epoch": 4.576158940397351,
      "grad_norm": 1.226389408711565,
      "learning_rate": 6.356909348020707e-05,
      "loss": 0.9688,
      "step": 6910
    },
    {
      "epoch": 4.57682119205298,
      "grad_norm": 1.0687900581503968,
      "learning_rate": 6.35145488298912e-05,
      "loss": 0.7734,
      "step": 6911
    },
    {
      "epoch": 4.577483443708609,
      "grad_norm": 1.1816655492699555,
      "learning_rate": 6.346002130468613e-05,
      "loss": 0.8164,
      "step": 6912
    },
    {
      "epoch": 4.578145695364238,
      "grad_norm": 1.1135066462507233,
      "learning_rate": 6.340551091538898e-05,
      "loss": 0.7812,
      "step": 6913
    },
    {
      "epoch": 4.578807947019867,
      "grad_norm": 1.1638899796333226,
      "learning_rate": 6.335101767279351e-05,
      "loss": 0.832,
      "step": 6914
    },
    {
      "epoch": 4.579470198675496,
      "grad_norm": 1.121946526144474,
      "learning_rate": 6.329654158768988e-05,
      "loss": 0.8281,
      "step": 6915
    },
    {
      "epoch": 4.580132450331126,
      "grad_norm": 1.2845482062303137,
      "learning_rate": 6.32420826708651e-05,
      "loss": 0.9531,
      "step": 6916
    },
    {
      "epoch": 4.580794701986755,
      "grad_norm": 1.2162225708784902,
      "learning_rate": 6.318764093310253e-05,
      "loss": 0.8789,
      "step": 6917
    },
    {
      "epoch": 4.581456953642384,
      "grad_norm": 1.0874712271288844,
      "learning_rate": 6.313321638518235e-05,
      "loss": 0.793,
      "step": 6918
    },
    {
      "epoch": 4.582119205298013,
      "grad_norm": 1.068498489983017,
      "learning_rate": 6.307880903788125e-05,
      "loss": 0.7383,
      "step": 6919
    },
    {
      "epoch": 4.582781456953643,
      "grad_norm": 1.065664103307406,
      "learning_rate": 6.302441890197249e-05,
      "loss": 0.7773,
      "step": 6920
    },
    {
      "epoch": 4.583443708609272,
      "grad_norm": 1.2050575922136004,
      "learning_rate": 6.297004598822587e-05,
      "loss": 0.8477,
      "step": 6921
    },
    {
      "epoch": 4.584105960264901,
      "grad_norm": 1.0857044924821755,
      "learning_rate": 6.291569030740792e-05,
      "loss": 0.7656,
      "step": 6922
    },
    {
      "epoch": 4.5847682119205295,
      "grad_norm": 1.2560705097155973,
      "learning_rate": 6.286135187028171e-05,
      "loss": 0.8906,
      "step": 6923
    },
    {
      "epoch": 4.585430463576159,
      "grad_norm": 1.0581088153393603,
      "learning_rate": 6.28070306876068e-05,
      "loss": 0.7031,
      "step": 6924
    },
    {
      "epoch": 4.586092715231788,
      "grad_norm": 1.174399165521076,
      "learning_rate": 6.275272677013951e-05,
      "loss": 0.8008,
      "step": 6925
    },
    {
      "epoch": 4.586754966887417,
      "grad_norm": 1.166594953066894,
      "learning_rate": 6.269844012863253e-05,
      "loss": 0.9219,
      "step": 6926
    },
    {
      "epoch": 4.587417218543046,
      "grad_norm": 1.0806904357595433,
      "learning_rate": 6.264417077383532e-05,
      "loss": 0.7734,
      "step": 6927
    },
    {
      "epoch": 4.588079470198675,
      "grad_norm": 1.161460304934366,
      "learning_rate": 6.258991871649376e-05,
      "loss": 0.793,
      "step": 6928
    },
    {
      "epoch": 4.588741721854305,
      "grad_norm": 1.165820562666789,
      "learning_rate": 6.253568396735046e-05,
      "loss": 0.875,
      "step": 6929
    },
    {
      "epoch": 4.589403973509934,
      "grad_norm": 1.1285372264594138,
      "learning_rate": 6.248146653714443e-05,
      "loss": 0.8047,
      "step": 6930
    },
    {
      "epoch": 4.590066225165563,
      "grad_norm": 1.1387866223413086,
      "learning_rate": 6.242726643661138e-05,
      "loss": 0.9141,
      "step": 6931
    },
    {
      "epoch": 4.590728476821192,
      "grad_norm": 1.2812130839716394,
      "learning_rate": 6.237308367648365e-05,
      "loss": 0.9492,
      "step": 6932
    },
    {
      "epoch": 4.591390728476821,
      "grad_norm": 1.3182182679587982,
      "learning_rate": 6.231891826748987e-05,
      "loss": 0.9805,
      "step": 6933
    },
    {
      "epoch": 4.592052980132451,
      "grad_norm": 1.0947501243642095,
      "learning_rate": 6.226477022035554e-05,
      "loss": 0.793,
      "step": 6934
    },
    {
      "epoch": 4.5927152317880795,
      "grad_norm": 1.121534402990068,
      "learning_rate": 6.221063954580259e-05,
      "loss": 0.7617,
      "step": 6935
    },
    {
      "epoch": 4.593377483443708,
      "grad_norm": 1.226784154126119,
      "learning_rate": 6.215652625454949e-05,
      "loss": 0.9414,
      "step": 6936
    },
    {
      "epoch": 4.594039735099337,
      "grad_norm": 1.1229386473547285,
      "learning_rate": 6.210243035731121e-05,
      "loss": 0.8594,
      "step": 6937
    },
    {
      "epoch": 4.594701986754967,
      "grad_norm": 1.0468893990884205,
      "learning_rate": 6.204835186479945e-05,
      "loss": 0.7539,
      "step": 6938
    },
    {
      "epoch": 4.595364238410596,
      "grad_norm": 1.265856887695957,
      "learning_rate": 6.199429078772242e-05,
      "loss": 0.9062,
      "step": 6939
    },
    {
      "epoch": 4.596026490066225,
      "grad_norm": 1.1130122323313665,
      "learning_rate": 6.19402471367847e-05,
      "loss": 0.7656,
      "step": 6940
    },
    {
      "epoch": 4.596688741721854,
      "grad_norm": 1.3464327885009477,
      "learning_rate": 6.188622092268768e-05,
      "loss": 1.0234,
      "step": 6941
    },
    {
      "epoch": 4.597350993377484,
      "grad_norm": 0.9566015283958451,
      "learning_rate": 6.183221215612904e-05,
      "loss": 0.6055,
      "step": 6942
    },
    {
      "epoch": 4.598013245033113,
      "grad_norm": 1.1607718191218497,
      "learning_rate": 6.17782208478032e-05,
      "loss": 0.8047,
      "step": 6943
    },
    {
      "epoch": 4.598675496688742,
      "grad_norm": 1.076377943875386,
      "learning_rate": 6.172424700840114e-05,
      "loss": 0.7188,
      "step": 6944
    },
    {
      "epoch": 4.599337748344371,
      "grad_norm": 1.0446350237542574,
      "learning_rate": 6.16702906486102e-05,
      "loss": 0.7617,
      "step": 6945
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.3065927141980471,
      "learning_rate": 6.161635177911428e-05,
      "loss": 0.9688,
      "step": 6946
    },
    {
      "epoch": 4.6006622516556295,
      "grad_norm": 1.1141310748179272,
      "learning_rate": 6.15624304105941e-05,
      "loss": 0.8242,
      "step": 6947
    },
    {
      "epoch": 4.601324503311258,
      "grad_norm": 1.008078207130439,
      "learning_rate": 6.15085265537266e-05,
      "loss": 0.7109,
      "step": 6948
    },
    {
      "epoch": 4.601986754966887,
      "grad_norm": 1.0819306682510477,
      "learning_rate": 6.145464021918531e-05,
      "loss": 0.75,
      "step": 6949
    },
    {
      "epoch": 4.602649006622516,
      "grad_norm": 1.1334017945720658,
      "learning_rate": 6.14007714176404e-05,
      "loss": 0.7773,
      "step": 6950
    },
    {
      "epoch": 4.603311258278145,
      "grad_norm": 1.2391789108066615,
      "learning_rate": 6.134692015975856e-05,
      "loss": 0.9023,
      "step": 6951
    },
    {
      "epoch": 4.603973509933775,
      "grad_norm": 1.118121300960935,
      "learning_rate": 6.129308645620284e-05,
      "loss": 0.7461,
      "step": 6952
    },
    {
      "epoch": 4.604635761589404,
      "grad_norm": 1.2347511166336906,
      "learning_rate": 6.123927031763307e-05,
      "loss": 0.8633,
      "step": 6953
    },
    {
      "epoch": 4.605298013245033,
      "grad_norm": 1.1258387026714443,
      "learning_rate": 6.11854717547053e-05,
      "loss": 0.793,
      "step": 6954
    },
    {
      "epoch": 4.605960264900662,
      "grad_norm": 1.2023163018378926,
      "learning_rate": 6.113169077807238e-05,
      "loss": 0.8906,
      "step": 6955
    },
    {
      "epoch": 4.606622516556292,
      "grad_norm": 1.2214320573615893,
      "learning_rate": 6.107792739838355e-05,
      "loss": 0.8594,
      "step": 6956
    },
    {
      "epoch": 4.607284768211921,
      "grad_norm": 1.0806020221419912,
      "learning_rate": 6.102418162628457e-05,
      "loss": 0.7578,
      "step": 6957
    },
    {
      "epoch": 4.60794701986755,
      "grad_norm": 1.1798177414234585,
      "learning_rate": 6.097045347241766e-05,
      "loss": 0.8086,
      "step": 6958
    },
    {
      "epoch": 4.608609271523179,
      "grad_norm": 1.3769076843443844,
      "learning_rate": 6.091674294742164e-05,
      "loss": 1.0859,
      "step": 6959
    },
    {
      "epoch": 4.609271523178808,
      "grad_norm": 1.0639306015472156,
      "learning_rate": 6.086305006193187e-05,
      "loss": 0.8281,
      "step": 6960
    },
    {
      "epoch": 4.609933774834437,
      "grad_norm": 1.1445094011489165,
      "learning_rate": 6.080937482658006e-05,
      "loss": 0.832,
      "step": 6961
    },
    {
      "epoch": 4.610596026490066,
      "grad_norm": 1.2250906383901035,
      "learning_rate": 6.075571725199463e-05,
      "loss": 0.8438,
      "step": 6962
    },
    {
      "epoch": 4.611258278145695,
      "grad_norm": 1.1676497718174408,
      "learning_rate": 6.070207734880028e-05,
      "loss": 0.8828,
      "step": 6963
    },
    {
      "epoch": 4.611920529801324,
      "grad_norm": 1.197848513711673,
      "learning_rate": 6.064845512761839e-05,
      "loss": 0.8633,
      "step": 6964
    },
    {
      "epoch": 4.612582781456954,
      "grad_norm": 1.253446138575748,
      "learning_rate": 6.0594850599066805e-05,
      "loss": 0.9258,
      "step": 6965
    },
    {
      "epoch": 4.613245033112583,
      "grad_norm": 1.1844430572747644,
      "learning_rate": 6.0541263773759815e-05,
      "loss": 0.8281,
      "step": 6966
    },
    {
      "epoch": 4.613907284768212,
      "grad_norm": 1.0961258586393516,
      "learning_rate": 6.048769466230808e-05,
      "loss": 0.7422,
      "step": 6967
    },
    {
      "epoch": 4.614569536423841,
      "grad_norm": 0.99955978161838,
      "learning_rate": 6.043414327531915e-05,
      "loss": 0.6328,
      "step": 6968
    },
    {
      "epoch": 4.61523178807947,
      "grad_norm": 1.1660423726874696,
      "learning_rate": 6.0380609623396687e-05,
      "loss": 0.8633,
      "step": 6969
    },
    {
      "epoch": 4.6158940397351,
      "grad_norm": 1.2048777759940674,
      "learning_rate": 6.032709371714092e-05,
      "loss": 0.8438,
      "step": 6970
    },
    {
      "epoch": 4.6165562913907285,
      "grad_norm": 1.0701535994214333,
      "learning_rate": 6.027359556714867e-05,
      "loss": 0.7188,
      "step": 6971
    },
    {
      "epoch": 4.6172185430463575,
      "grad_norm": 1.1684494217386998,
      "learning_rate": 6.0220115184013216e-05,
      "loss": 0.8164,
      "step": 6972
    },
    {
      "epoch": 4.617880794701986,
      "grad_norm": 1.3791863451149293,
      "learning_rate": 6.0166652578324196e-05,
      "loss": 0.9609,
      "step": 6973
    },
    {
      "epoch": 4.618543046357616,
      "grad_norm": 1.2076100126399816,
      "learning_rate": 6.011320776066794e-05,
      "loss": 0.8633,
      "step": 6974
    },
    {
      "epoch": 4.619205298013245,
      "grad_norm": 1.3822303782541994,
      "learning_rate": 6.0059780741627e-05,
      "loss": 1.0938,
      "step": 6975
    },
    {
      "epoch": 4.619867549668874,
      "grad_norm": 1.048350775721481,
      "learning_rate": 6.000637153178061e-05,
      "loss": 0.7383,
      "step": 6976
    },
    {
      "epoch": 4.620529801324503,
      "grad_norm": 1.2078187909579006,
      "learning_rate": 5.995298014170445e-05,
      "loss": 0.8867,
      "step": 6977
    },
    {
      "epoch": 4.621192052980133,
      "grad_norm": 1.159730581457243,
      "learning_rate": 5.989960658197057e-05,
      "loss": 0.8828,
      "step": 6978
    },
    {
      "epoch": 4.621854304635762,
      "grad_norm": 1.1941997606678523,
      "learning_rate": 5.984625086314751e-05,
      "loss": 0.8516,
      "step": 6979
    },
    {
      "epoch": 4.622516556291391,
      "grad_norm": 1.2307401843524093,
      "learning_rate": 5.979291299580036e-05,
      "loss": 0.7812,
      "step": 6980
    },
    {
      "epoch": 4.62317880794702,
      "grad_norm": 1.0396628310227214,
      "learning_rate": 5.973959299049067e-05,
      "loss": 0.7266,
      "step": 6981
    },
    {
      "epoch": 4.623841059602649,
      "grad_norm": 1.0570827076998779,
      "learning_rate": 5.9686290857776385e-05,
      "loss": 0.75,
      "step": 6982
    },
    {
      "epoch": 4.6245033112582785,
      "grad_norm": 1.3495148598448052,
      "learning_rate": 5.963300660821185e-05,
      "loss": 0.9453,
      "step": 6983
    },
    {
      "epoch": 4.6251655629139075,
      "grad_norm": 1.0885162229533623,
      "learning_rate": 5.957974025234804e-05,
      "loss": 0.7461,
      "step": 6984
    },
    {
      "epoch": 4.625827814569536,
      "grad_norm": 1.1654507895282973,
      "learning_rate": 5.952649180073236e-05,
      "loss": 0.7852,
      "step": 6985
    },
    {
      "epoch": 4.626490066225165,
      "grad_norm": 1.2356631339621922,
      "learning_rate": 5.947326126390848e-05,
      "loss": 0.8281,
      "step": 6986
    },
    {
      "epoch": 4.627152317880794,
      "grad_norm": 1.433047929277579,
      "learning_rate": 5.942004865241672e-05,
      "loss": 1.0312,
      "step": 6987
    },
    {
      "epoch": 4.627814569536424,
      "grad_norm": 1.0982917907059395,
      "learning_rate": 5.936685397679387e-05,
      "loss": 0.7969,
      "step": 6988
    },
    {
      "epoch": 4.628476821192053,
      "grad_norm": 1.3756166861808772,
      "learning_rate": 5.9313677247572936e-05,
      "loss": 1.0391,
      "step": 6989
    },
    {
      "epoch": 4.629139072847682,
      "grad_norm": 1.2531345233597178,
      "learning_rate": 5.926051847528365e-05,
      "loss": 0.9688,
      "step": 6990
    },
    {
      "epoch": 4.629801324503311,
      "grad_norm": 1.2044522845626304,
      "learning_rate": 5.9207377670451943e-05,
      "loss": 0.832,
      "step": 6991
    },
    {
      "epoch": 4.630463576158941,
      "grad_norm": 1.0982888098807004,
      "learning_rate": 5.9154254843600354e-05,
      "loss": 0.7305,
      "step": 6992
    },
    {
      "epoch": 4.63112582781457,
      "grad_norm": 1.4406317105376167,
      "learning_rate": 5.91011500052479e-05,
      "loss": 1.0391,
      "step": 6993
    },
    {
      "epoch": 4.631788079470199,
      "grad_norm": 1.1117843218100936,
      "learning_rate": 5.904806316590984e-05,
      "loss": 0.7578,
      "step": 6994
    },
    {
      "epoch": 4.632450331125828,
      "grad_norm": 1.0910931878544285,
      "learning_rate": 5.899499433609795e-05,
      "loss": 0.7969,
      "step": 6995
    },
    {
      "epoch": 4.633112582781457,
      "grad_norm": 1.251473743233115,
      "learning_rate": 5.894194352632053e-05,
      "loss": 0.8906,
      "step": 6996
    },
    {
      "epoch": 4.633774834437086,
      "grad_norm": 1.1302000634674498,
      "learning_rate": 5.888891074708229e-05,
      "loss": 0.8242,
      "step": 6997
    },
    {
      "epoch": 4.634437086092715,
      "grad_norm": 1.0437022533348599,
      "learning_rate": 5.8835896008884216e-05,
      "loss": 0.7344,
      "step": 6998
    },
    {
      "epoch": 4.635099337748344,
      "grad_norm": 1.0939240578619611,
      "learning_rate": 5.878289932222395e-05,
      "loss": 0.7578,
      "step": 6999
    },
    {
      "epoch": 4.635761589403973,
      "grad_norm": 0.9955241038263763,
      "learning_rate": 5.8729920697595345e-05,
      "loss": 0.7109,
      "step": 7000
    },
    {
      "epoch": 4.636423841059603,
      "grad_norm": 1.0723762294825157,
      "learning_rate": 5.8676960145488816e-05,
      "loss": 0.75,
      "step": 7001
    },
    {
      "epoch": 4.637086092715232,
      "grad_norm": 1.0176393943293747,
      "learning_rate": 5.8624017676391214e-05,
      "loss": 0.7773,
      "step": 7002
    },
    {
      "epoch": 4.637748344370861,
      "grad_norm": 1.1535670273445204,
      "learning_rate": 5.8571093300785696e-05,
      "loss": 0.8594,
      "step": 7003
    },
    {
      "epoch": 4.63841059602649,
      "grad_norm": 1.1339423634110104,
      "learning_rate": 5.85181870291518e-05,
      "loss": 0.8672,
      "step": 7004
    },
    {
      "epoch": 4.639072847682119,
      "grad_norm": 1.1846229497516882,
      "learning_rate": 5.8465298871965794e-05,
      "loss": 0.8672,
      "step": 7005
    },
    {
      "epoch": 4.639735099337749,
      "grad_norm": 1.013882583684185,
      "learning_rate": 5.841242883970001e-05,
      "loss": 0.7109,
      "step": 7006
    },
    {
      "epoch": 4.640397350993378,
      "grad_norm": 1.1042227067314965,
      "learning_rate": 5.835957694282329e-05,
      "loss": 0.7812,
      "step": 7007
    },
    {
      "epoch": 4.6410596026490065,
      "grad_norm": 1.3040755706230582,
      "learning_rate": 5.830674319180094e-05,
      "loss": 0.9688,
      "step": 7008
    },
    {
      "epoch": 4.6417218543046355,
      "grad_norm": 1.0116156431345809,
      "learning_rate": 5.825392759709473e-05,
      "loss": 0.6797,
      "step": 7009
    },
    {
      "epoch": 4.642384105960264,
      "grad_norm": 1.0228567670487176,
      "learning_rate": 5.8201130169162654e-05,
      "loss": 0.6602,
      "step": 7010
    },
    {
      "epoch": 4.643046357615894,
      "grad_norm": 1.2078493017085987,
      "learning_rate": 5.814835091845928e-05,
      "loss": 0.9453,
      "step": 7011
    },
    {
      "epoch": 4.643708609271523,
      "grad_norm": 1.0966112598112292,
      "learning_rate": 5.80955898554354e-05,
      "loss": 0.8203,
      "step": 7012
    },
    {
      "epoch": 4.644370860927152,
      "grad_norm": 1.1553230993936505,
      "learning_rate": 5.804284699053842e-05,
      "loss": 0.7461,
      "step": 7013
    },
    {
      "epoch": 4.645033112582782,
      "grad_norm": 1.1065082754681614,
      "learning_rate": 5.7990122334212014e-05,
      "loss": 0.8086,
      "step": 7014
    },
    {
      "epoch": 4.645695364238411,
      "grad_norm": 1.2010700836145147,
      "learning_rate": 5.7937415896896245e-05,
      "loss": 0.8672,
      "step": 7015
    },
    {
      "epoch": 4.64635761589404,
      "grad_norm": 1.1840075470929352,
      "learning_rate": 5.788472768902754e-05,
      "loss": 0.8594,
      "step": 7016
    },
    {
      "epoch": 4.647019867549669,
      "grad_norm": 1.1971245083269113,
      "learning_rate": 5.783205772103883e-05,
      "loss": 0.8086,
      "step": 7017
    },
    {
      "epoch": 4.647682119205298,
      "grad_norm": 1.1794082049395023,
      "learning_rate": 5.77794060033594e-05,
      "loss": 0.8008,
      "step": 7018
    },
    {
      "epoch": 4.6483443708609276,
      "grad_norm": 1.1188360751481452,
      "learning_rate": 5.772677254641483e-05,
      "loss": 0.75,
      "step": 7019
    },
    {
      "epoch": 4.6490066225165565,
      "grad_norm": 1.035714479378241,
      "learning_rate": 5.767415736062719e-05,
      "loss": 0.6875,
      "step": 7020
    },
    {
      "epoch": 4.649668874172185,
      "grad_norm": 1.1032633046306326,
      "learning_rate": 5.762156045641484e-05,
      "loss": 0.7109,
      "step": 7021
    },
    {
      "epoch": 4.650331125827814,
      "grad_norm": 1.2112131069540477,
      "learning_rate": 5.756898184419262e-05,
      "loss": 0.8008,
      "step": 7022
    },
    {
      "epoch": 4.650993377483443,
      "grad_norm": 1.185459114317371,
      "learning_rate": 5.751642153437172e-05,
      "loss": 0.8008,
      "step": 7023
    },
    {
      "epoch": 4.651655629139073,
      "grad_norm": 1.4126557688963395,
      "learning_rate": 5.746387953735964e-05,
      "loss": 1.0547,
      "step": 7024
    },
    {
      "epoch": 4.652317880794702,
      "grad_norm": 1.2706397110687273,
      "learning_rate": 5.7411355863560194e-05,
      "loss": 0.8828,
      "step": 7025
    },
    {
      "epoch": 4.652980132450331,
      "grad_norm": 1.066286819523345,
      "learning_rate": 5.735885052337391e-05,
      "loss": 0.6289,
      "step": 7026
    },
    {
      "epoch": 4.65364238410596,
      "grad_norm": 1.1743937643410667,
      "learning_rate": 5.7306363527197294e-05,
      "loss": 0.8516,
      "step": 7027
    },
    {
      "epoch": 4.654304635761589,
      "grad_norm": 1.1574547977324783,
      "learning_rate": 5.7253894885423355e-05,
      "loss": 0.7812,
      "step": 7028
    },
    {
      "epoch": 4.654966887417219,
      "grad_norm": 1.0429272796260174,
      "learning_rate": 5.7201444608441524e-05,
      "loss": 0.668,
      "step": 7029
    },
    {
      "epoch": 4.655629139072848,
      "grad_norm": 1.1223914158006674,
      "learning_rate": 5.714901270663761e-05,
      "loss": 0.8438,
      "step": 7030
    },
    {
      "epoch": 4.656291390728477,
      "grad_norm": 1.1556065914148936,
      "learning_rate": 5.709659919039364e-05,
      "loss": 0.7461,
      "step": 7031
    },
    {
      "epoch": 4.656953642384106,
      "grad_norm": 1.3819540553465524,
      "learning_rate": 5.70442040700882e-05,
      "loss": 1.0,
      "step": 7032
    },
    {
      "epoch": 4.657615894039735,
      "grad_norm": 1.1879022462781499,
      "learning_rate": 5.699182735609597e-05,
      "loss": 0.8047,
      "step": 7033
    },
    {
      "epoch": 4.658278145695364,
      "grad_norm": 1.0812793072490416,
      "learning_rate": 5.6939469058788255e-05,
      "loss": 0.7305,
      "step": 7034
    },
    {
      "epoch": 4.658940397350993,
      "grad_norm": 1.097442110783161,
      "learning_rate": 5.688712918853261e-05,
      "loss": 0.8242,
      "step": 7035
    },
    {
      "epoch": 4.659602649006622,
      "grad_norm": 1.0584212059227276,
      "learning_rate": 5.683480775569289e-05,
      "loss": 0.7227,
      "step": 7036
    },
    {
      "epoch": 4.660264900662252,
      "grad_norm": 1.1828382647985438,
      "learning_rate": 5.678250477062929e-05,
      "loss": 0.8477,
      "step": 7037
    },
    {
      "epoch": 4.660927152317881,
      "grad_norm": 1.2563832106080912,
      "learning_rate": 5.6730220243698456e-05,
      "loss": 0.9297,
      "step": 7038
    },
    {
      "epoch": 4.66158940397351,
      "grad_norm": 1.306344873903242,
      "learning_rate": 5.667795418525335e-05,
      "loss": 0.8359,
      "step": 7039
    },
    {
      "epoch": 4.662251655629139,
      "grad_norm": 1.1662518949848335,
      "learning_rate": 5.662570660564323e-05,
      "loss": 0.7891,
      "step": 7040
    },
    {
      "epoch": 4.662913907284768,
      "grad_norm": 0.99820324650275,
      "learning_rate": 5.65734775152136e-05,
      "loss": 0.5664,
      "step": 7041
    },
    {
      "epoch": 4.663576158940398,
      "grad_norm": 1.0741395049460742,
      "learning_rate": 5.652126692430663e-05,
      "loss": 0.75,
      "step": 7042
    },
    {
      "epoch": 4.664238410596027,
      "grad_norm": 1.067751145176672,
      "learning_rate": 5.6469074843260506e-05,
      "loss": 0.7891,
      "step": 7043
    },
    {
      "epoch": 4.664900662251656,
      "grad_norm": 1.113621769657866,
      "learning_rate": 5.6416901282409785e-05,
      "loss": 0.7383,
      "step": 7044
    },
    {
      "epoch": 4.6655629139072845,
      "grad_norm": 1.1224087805689527,
      "learning_rate": 5.6364746252085535e-05,
      "loss": 0.7461,
      "step": 7045
    },
    {
      "epoch": 4.6662251655629134,
      "grad_norm": 1.1093097628627466,
      "learning_rate": 5.631260976261507e-05,
      "loss": 0.8125,
      "step": 7046
    },
    {
      "epoch": 4.666887417218543,
      "grad_norm": 1.0724445127372304,
      "learning_rate": 5.6260491824321897e-05,
      "loss": 0.7461,
      "step": 7047
    },
    {
      "epoch": 4.667549668874172,
      "grad_norm": 1.1884541826221118,
      "learning_rate": 5.620839244752608e-05,
      "loss": 0.9062,
      "step": 7048
    },
    {
      "epoch": 4.668211920529801,
      "grad_norm": 1.0855767259900309,
      "learning_rate": 5.61563116425438e-05,
      "loss": 0.6758,
      "step": 7049
    },
    {
      "epoch": 4.66887417218543,
      "grad_norm": 1.112874019715515,
      "learning_rate": 5.6104249419687706e-05,
      "loss": 0.7773,
      "step": 7050
    },
    {
      "epoch": 4.66953642384106,
      "grad_norm": 1.1766604990221852,
      "learning_rate": 5.6052205789266756e-05,
      "loss": 0.8633,
      "step": 7051
    },
    {
      "epoch": 4.670198675496689,
      "grad_norm": 1.1506621307446079,
      "learning_rate": 5.6000180761586125e-05,
      "loss": 0.7617,
      "step": 7052
    },
    {
      "epoch": 4.670860927152318,
      "grad_norm": 1.1873435749128756,
      "learning_rate": 5.594817434694733e-05,
      "loss": 0.8242,
      "step": 7053
    },
    {
      "epoch": 4.671523178807947,
      "grad_norm": 1.1243927461307628,
      "learning_rate": 5.5896186555648284e-05,
      "loss": 0.8398,
      "step": 7054
    },
    {
      "epoch": 4.672185430463577,
      "grad_norm": 1.5505940287125382,
      "learning_rate": 5.584421739798322e-05,
      "loss": 1.1875,
      "step": 7055
    },
    {
      "epoch": 4.6728476821192055,
      "grad_norm": 1.3330479328130966,
      "learning_rate": 5.579226688424251e-05,
      "loss": 1.0234,
      "step": 7056
    },
    {
      "epoch": 4.6735099337748345,
      "grad_norm": 1.1569371029468127,
      "learning_rate": 5.5740335024713075e-05,
      "loss": 0.7969,
      "step": 7057
    },
    {
      "epoch": 4.674172185430463,
      "grad_norm": 1.22583991876913,
      "learning_rate": 5.56884218296779e-05,
      "loss": 0.8008,
      "step": 7058
    },
    {
      "epoch": 4.674834437086092,
      "grad_norm": 1.059964586175887,
      "learning_rate": 5.563652730941645e-05,
      "loss": 0.7344,
      "step": 7059
    },
    {
      "epoch": 4.675496688741722,
      "grad_norm": 1.3669486261302486,
      "learning_rate": 5.558465147420449e-05,
      "loss": 0.9844,
      "step": 7060
    },
    {
      "epoch": 4.676158940397351,
      "grad_norm": 1.3586032853928351,
      "learning_rate": 5.5532794334313965e-05,
      "loss": 1.0078,
      "step": 7061
    },
    {
      "epoch": 4.67682119205298,
      "grad_norm": 1.2169532919154653,
      "learning_rate": 5.5480955900013106e-05,
      "loss": 0.9688,
      "step": 7062
    },
    {
      "epoch": 4.677483443708609,
      "grad_norm": 1.1036798576105105,
      "learning_rate": 5.54291361815667e-05,
      "loss": 0.7695,
      "step": 7063
    },
    {
      "epoch": 4.678145695364238,
      "grad_norm": 1.155226298889485,
      "learning_rate": 5.537733518923554e-05,
      "loss": 0.7734,
      "step": 7064
    },
    {
      "epoch": 4.678807947019868,
      "grad_norm": 1.2334251481222667,
      "learning_rate": 5.532555293327678e-05,
      "loss": 0.9453,
      "step": 7065
    },
    {
      "epoch": 4.679470198675497,
      "grad_norm": 1.1813151653522997,
      "learning_rate": 5.527378942394396e-05,
      "loss": 0.8047,
      "step": 7066
    },
    {
      "epoch": 4.680132450331126,
      "grad_norm": 1.1558483712216368,
      "learning_rate": 5.522204467148687e-05,
      "loss": 0.875,
      "step": 7067
    },
    {
      "epoch": 4.680794701986755,
      "grad_norm": 1.1692424355441124,
      "learning_rate": 5.5170318686151485e-05,
      "loss": 0.8242,
      "step": 7068
    },
    {
      "epoch": 4.6814569536423845,
      "grad_norm": 1.2167354805394077,
      "learning_rate": 5.511861147818023e-05,
      "loss": 0.8867,
      "step": 7069
    },
    {
      "epoch": 4.682119205298013,
      "grad_norm": 1.3018414722199823,
      "learning_rate": 5.5066923057811634e-05,
      "loss": 1.0156,
      "step": 7070
    },
    {
      "epoch": 4.682781456953642,
      "grad_norm": 1.1344952685949268,
      "learning_rate": 5.5015253435280656e-05,
      "loss": 0.832,
      "step": 7071
    },
    {
      "epoch": 4.683443708609271,
      "grad_norm": 1.2467812924210178,
      "learning_rate": 5.49636026208185e-05,
      "loss": 0.9336,
      "step": 7072
    },
    {
      "epoch": 4.684105960264901,
      "grad_norm": 1.2580954193186533,
      "learning_rate": 5.491197062465258e-05,
      "loss": 0.918,
      "step": 7073
    },
    {
      "epoch": 4.68476821192053,
      "grad_norm": 1.1215631231511882,
      "learning_rate": 5.4860357457006594e-05,
      "loss": 0.8125,
      "step": 7074
    },
    {
      "epoch": 4.685430463576159,
      "grad_norm": 1.0978783136959187,
      "learning_rate": 5.4808763128100555e-05,
      "loss": 0.8086,
      "step": 7075
    },
    {
      "epoch": 4.686092715231788,
      "grad_norm": 1.2622610259997786,
      "learning_rate": 5.4757187648150806e-05,
      "loss": 0.957,
      "step": 7076
    },
    {
      "epoch": 4.686754966887417,
      "grad_norm": 1.120968151397482,
      "learning_rate": 5.4705631027369774e-05,
      "loss": 0.6992,
      "step": 7077
    },
    {
      "epoch": 4.687417218543047,
      "grad_norm": 1.1085147152685302,
      "learning_rate": 5.4654093275966367e-05,
      "loss": 0.832,
      "step": 7078
    },
    {
      "epoch": 4.688079470198676,
      "grad_norm": 1.1402191673054127,
      "learning_rate": 5.4602574404145535e-05,
      "loss": 0.8242,
      "step": 7079
    },
    {
      "epoch": 4.688741721854305,
      "grad_norm": 1.2032560947634245,
      "learning_rate": 5.455107442210867e-05,
      "loss": 0.793,
      "step": 7080
    },
    {
      "epoch": 4.6894039735099335,
      "grad_norm": 1.0947289845339399,
      "learning_rate": 5.449959334005341e-05,
      "loss": 0.7305,
      "step": 7081
    },
    {
      "epoch": 4.6900662251655625,
      "grad_norm": 1.0495007219487504,
      "learning_rate": 5.444813116817349e-05,
      "loss": 0.6836,
      "step": 7082
    },
    {
      "epoch": 4.690728476821192,
      "grad_norm": 1.2230074675120852,
      "learning_rate": 5.439668791665906e-05,
      "loss": 0.8398,
      "step": 7083
    },
    {
      "epoch": 4.691390728476821,
      "grad_norm": 1.1468296423903233,
      "learning_rate": 5.434526359569654e-05,
      "loss": 0.7969,
      "step": 7084
    },
    {
      "epoch": 4.69205298013245,
      "grad_norm": 1.1667174000097313,
      "learning_rate": 5.4293858215468475e-05,
      "loss": 0.8164,
      "step": 7085
    },
    {
      "epoch": 4.692715231788079,
      "grad_norm": 1.0918912375854812,
      "learning_rate": 5.4242471786153676e-05,
      "loss": 0.668,
      "step": 7086
    },
    {
      "epoch": 4.693377483443709,
      "grad_norm": 1.121677768080626,
      "learning_rate": 5.419110431792726e-05,
      "loss": 0.7461,
      "step": 7087
    },
    {
      "epoch": 4.694039735099338,
      "grad_norm": 1.0847243475563966,
      "learning_rate": 5.413975582096067e-05,
      "loss": 0.7812,
      "step": 7088
    },
    {
      "epoch": 4.694701986754967,
      "grad_norm": 1.0367631214908204,
      "learning_rate": 5.4088426305421385e-05,
      "loss": 0.6797,
      "step": 7089
    },
    {
      "epoch": 4.695364238410596,
      "grad_norm": 1.0772441961787123,
      "learning_rate": 5.403711578147333e-05,
      "loss": 0.8047,
      "step": 7090
    },
    {
      "epoch": 4.696026490066226,
      "grad_norm": 1.21661670190246,
      "learning_rate": 5.398582425927646e-05,
      "loss": 0.9883,
      "step": 7091
    },
    {
      "epoch": 4.696688741721855,
      "grad_norm": 1.1931615473169208,
      "learning_rate": 5.393455174898718e-05,
      "loss": 0.832,
      "step": 7092
    },
    {
      "epoch": 4.6973509933774835,
      "grad_norm": 1.3109990322608516,
      "learning_rate": 5.3883298260758037e-05,
      "loss": 0.9727,
      "step": 7093
    },
    {
      "epoch": 4.6980132450331125,
      "grad_norm": 1.114938914632661,
      "learning_rate": 5.383206380473779e-05,
      "loss": 0.7539,
      "step": 7094
    },
    {
      "epoch": 4.698675496688741,
      "grad_norm": 1.0586752490670794,
      "learning_rate": 5.378084839107133e-05,
      "loss": 0.7227,
      "step": 7095
    },
    {
      "epoch": 4.699337748344371,
      "grad_norm": 1.1078226393374917,
      "learning_rate": 5.3729652029900124e-05,
      "loss": 0.7539,
      "step": 7096
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.2215658070496844,
      "learning_rate": 5.3678474731361505e-05,
      "loss": 0.8672,
      "step": 7097
    },
    {
      "epoch": 4.700662251655629,
      "grad_norm": 1.2491521687973715,
      "learning_rate": 5.362731650558919e-05,
      "loss": 0.8125,
      "step": 7098
    },
    {
      "epoch": 4.701324503311258,
      "grad_norm": 1.3257241365235994,
      "learning_rate": 5.357617736271296e-05,
      "loss": 0.9648,
      "step": 7099
    },
    {
      "epoch": 4.701986754966887,
      "grad_norm": 1.1708630982472266,
      "learning_rate": 5.3525057312859204e-05,
      "loss": 0.7852,
      "step": 7100
    },
    {
      "epoch": 4.702649006622517,
      "grad_norm": 1.0418924288205724,
      "learning_rate": 5.347395636615013e-05,
      "loss": 0.6875,
      "step": 7101
    },
    {
      "epoch": 4.703311258278146,
      "grad_norm": 1.2697256010250566,
      "learning_rate": 5.342287453270429e-05,
      "loss": 0.8633,
      "step": 7102
    },
    {
      "epoch": 4.703973509933775,
      "grad_norm": 1.1446557965870396,
      "learning_rate": 5.33718118226365e-05,
      "loss": 0.8125,
      "step": 7103
    },
    {
      "epoch": 4.704635761589404,
      "grad_norm": 0.9749906697692001,
      "learning_rate": 5.332076824605784e-05,
      "loss": 0.6094,
      "step": 7104
    },
    {
      "epoch": 4.7052980132450335,
      "grad_norm": 1.090784089440086,
      "learning_rate": 5.3269743813075397e-05,
      "loss": 0.7539,
      "step": 7105
    },
    {
      "epoch": 4.705960264900662,
      "grad_norm": 1.1440954459408965,
      "learning_rate": 5.321873853379272e-05,
      "loss": 0.8086,
      "step": 7106
    },
    {
      "epoch": 4.706622516556291,
      "grad_norm": 1.2214631534523952,
      "learning_rate": 5.316775241830932e-05,
      "loss": 0.8984,
      "step": 7107
    },
    {
      "epoch": 4.70728476821192,
      "grad_norm": 1.2165313386065622,
      "learning_rate": 5.311678547672108e-05,
      "loss": 0.9336,
      "step": 7108
    },
    {
      "epoch": 4.70794701986755,
      "grad_norm": 1.015415883944974,
      "learning_rate": 5.30658377191201e-05,
      "loss": 0.7188,
      "step": 7109
    },
    {
      "epoch": 4.708609271523179,
      "grad_norm": 1.215510323040822,
      "learning_rate": 5.301490915559457e-05,
      "loss": 0.8789,
      "step": 7110
    },
    {
      "epoch": 4.709271523178808,
      "grad_norm": 1.1603161379768387,
      "learning_rate": 5.296399979622887e-05,
      "loss": 0.8633,
      "step": 7111
    },
    {
      "epoch": 4.709933774834437,
      "grad_norm": 1.2345625844744474,
      "learning_rate": 5.291310965110368e-05,
      "loss": 0.9375,
      "step": 7112
    },
    {
      "epoch": 4.710596026490066,
      "grad_norm": 1.0987886549309647,
      "learning_rate": 5.286223873029591e-05,
      "loss": 0.75,
      "step": 7113
    },
    {
      "epoch": 4.711258278145696,
      "grad_norm": 0.9715037384497646,
      "learning_rate": 5.281138704387846e-05,
      "loss": 0.6836,
      "step": 7114
    },
    {
      "epoch": 4.711920529801325,
      "grad_norm": 1.055482340302242,
      "learning_rate": 5.2760554601920666e-05,
      "loss": 0.6641,
      "step": 7115
    },
    {
      "epoch": 4.712582781456954,
      "grad_norm": 1.1258207864591117,
      "learning_rate": 5.2709741414487814e-05,
      "loss": 0.7656,
      "step": 7116
    },
    {
      "epoch": 4.713245033112583,
      "grad_norm": 0.9993116413421633,
      "learning_rate": 5.265894749164157e-05,
      "loss": 0.6758,
      "step": 7117
    },
    {
      "epoch": 4.7139072847682115,
      "grad_norm": 1.1995881161745794,
      "learning_rate": 5.2608172843439744e-05,
      "loss": 0.875,
      "step": 7118
    },
    {
      "epoch": 4.714569536423841,
      "grad_norm": 1.212359892525248,
      "learning_rate": 5.255741747993625e-05,
      "loss": 0.9062,
      "step": 7119
    },
    {
      "epoch": 4.71523178807947,
      "grad_norm": 1.237288541773043,
      "learning_rate": 5.2506681411181144e-05,
      "loss": 0.8867,
      "step": 7120
    },
    {
      "epoch": 4.715894039735099,
      "grad_norm": 1.032104432730371,
      "learning_rate": 5.245596464722095e-05,
      "loss": 0.7109,
      "step": 7121
    },
    {
      "epoch": 4.716556291390728,
      "grad_norm": 1.2014457968659664,
      "learning_rate": 5.240526719809805e-05,
      "loss": 0.8477,
      "step": 7122
    },
    {
      "epoch": 4.717218543046357,
      "grad_norm": 1.193918382555,
      "learning_rate": 5.23545890738511e-05,
      "loss": 0.832,
      "step": 7123
    },
    {
      "epoch": 4.717880794701987,
      "grad_norm": 1.0920126587261676,
      "learning_rate": 5.230393028451495e-05,
      "loss": 0.8047,
      "step": 7124
    },
    {
      "epoch": 4.718543046357616,
      "grad_norm": 1.1612780794891169,
      "learning_rate": 5.225329084012073e-05,
      "loss": 0.7578,
      "step": 7125
    },
    {
      "epoch": 4.719205298013245,
      "grad_norm": 1.179832100183041,
      "learning_rate": 5.220267075069549e-05,
      "loss": 0.7578,
      "step": 7126
    },
    {
      "epoch": 4.719867549668874,
      "grad_norm": 1.1393915666642696,
      "learning_rate": 5.2152070026262713e-05,
      "loss": 0.8164,
      "step": 7127
    },
    {
      "epoch": 4.720529801324504,
      "grad_norm": 1.0795246085473353,
      "learning_rate": 5.21014886768418e-05,
      "loss": 0.7539,
      "step": 7128
    },
    {
      "epoch": 4.721192052980133,
      "grad_norm": 1.1642049867785575,
      "learning_rate": 5.2050926712448503e-05,
      "loss": 0.7969,
      "step": 7129
    },
    {
      "epoch": 4.7218543046357615,
      "grad_norm": 1.1726173359119982,
      "learning_rate": 5.200038414309471e-05,
      "loss": 0.7891,
      "step": 7130
    },
    {
      "epoch": 4.72251655629139,
      "grad_norm": 1.1135955208566248,
      "learning_rate": 5.194986097878839e-05,
      "loss": 0.7422,
      "step": 7131
    },
    {
      "epoch": 4.72317880794702,
      "grad_norm": 1.2309882883062173,
      "learning_rate": 5.189935722953367e-05,
      "loss": 0.9062,
      "step": 7132
    },
    {
      "epoch": 4.723841059602649,
      "grad_norm": 1.2099927357816274,
      "learning_rate": 5.1848872905330885e-05,
      "loss": 0.8203,
      "step": 7133
    },
    {
      "epoch": 4.724503311258278,
      "grad_norm": 1.3481521140344148,
      "learning_rate": 5.1798408016176585e-05,
      "loss": 0.957,
      "step": 7134
    },
    {
      "epoch": 4.725165562913907,
      "grad_norm": 1.074083335688575,
      "learning_rate": 5.1747962572063294e-05,
      "loss": 0.6641,
      "step": 7135
    },
    {
      "epoch": 4.725827814569536,
      "grad_norm": 1.2718830611086729,
      "learning_rate": 5.169753658297984e-05,
      "loss": 0.8711,
      "step": 7136
    },
    {
      "epoch": 4.726490066225166,
      "grad_norm": 1.1755623476665542,
      "learning_rate": 5.164713005891119e-05,
      "loss": 0.7773,
      "step": 7137
    },
    {
      "epoch": 4.727152317880795,
      "grad_norm": 1.2852782952449455,
      "learning_rate": 5.159674300983832e-05,
      "loss": 0.8594,
      "step": 7138
    },
    {
      "epoch": 4.727814569536424,
      "grad_norm": 1.095535218691356,
      "learning_rate": 5.1546375445738535e-05,
      "loss": 0.7812,
      "step": 7139
    },
    {
      "epoch": 4.728476821192053,
      "grad_norm": 1.0662498717182023,
      "learning_rate": 5.1496027376585106e-05,
      "loss": 0.7188,
      "step": 7140
    },
    {
      "epoch": 4.729139072847682,
      "grad_norm": 1.2747544030337339,
      "learning_rate": 5.144569881234756e-05,
      "loss": 0.9297,
      "step": 7141
    },
    {
      "epoch": 4.7298013245033115,
      "grad_norm": 1.0955048279431736,
      "learning_rate": 5.139538976299158e-05,
      "loss": 0.7188,
      "step": 7142
    },
    {
      "epoch": 4.73046357615894,
      "grad_norm": 1.0978554846546678,
      "learning_rate": 5.1345100238478905e-05,
      "loss": 0.7344,
      "step": 7143
    },
    {
      "epoch": 4.731125827814569,
      "grad_norm": 1.0603748675151685,
      "learning_rate": 5.1294830248767374e-05,
      "loss": 0.7852,
      "step": 7144
    },
    {
      "epoch": 4.731788079470198,
      "grad_norm": 1.0095152885970147,
      "learning_rate": 5.1244579803811055e-05,
      "loss": 0.7344,
      "step": 7145
    },
    {
      "epoch": 4.732450331125828,
      "grad_norm": 1.3373337810808805,
      "learning_rate": 5.119434891356019e-05,
      "loss": 0.9766,
      "step": 7146
    },
    {
      "epoch": 4.733112582781457,
      "grad_norm": 1.0346943298261204,
      "learning_rate": 5.114413758796096e-05,
      "loss": 0.7852,
      "step": 7147
    },
    {
      "epoch": 4.733774834437086,
      "grad_norm": 1.0369904510703065,
      "learning_rate": 5.109394583695589e-05,
      "loss": 0.6953,
      "step": 7148
    },
    {
      "epoch": 4.734437086092715,
      "grad_norm": 1.1388018166029203,
      "learning_rate": 5.104377367048342e-05,
      "loss": 0.8164,
      "step": 7149
    },
    {
      "epoch": 4.735099337748345,
      "grad_norm": 1.0910870857532011,
      "learning_rate": 5.099362109847824e-05,
      "loss": 0.8125,
      "step": 7150
    },
    {
      "epoch": 4.735761589403974,
      "grad_norm": 1.0975274580958678,
      "learning_rate": 5.094348813087121e-05,
      "loss": 0.7578,
      "step": 7151
    },
    {
      "epoch": 4.736423841059603,
      "grad_norm": 1.0581097904987777,
      "learning_rate": 5.089337477758919e-05,
      "loss": 0.7305,
      "step": 7152
    },
    {
      "epoch": 4.737086092715232,
      "grad_norm": 1.1201964295424447,
      "learning_rate": 5.084328104855511e-05,
      "loss": 0.8633,
      "step": 7153
    },
    {
      "epoch": 4.737748344370861,
      "grad_norm": 1.2435739535741215,
      "learning_rate": 5.079320695368819e-05,
      "loss": 0.8906,
      "step": 7154
    },
    {
      "epoch": 4.73841059602649,
      "grad_norm": 1.231746864094534,
      "learning_rate": 5.074315250290371e-05,
      "loss": 0.8672,
      "step": 7155
    },
    {
      "epoch": 4.739072847682119,
      "grad_norm": 1.1583337821362971,
      "learning_rate": 5.069311770611298e-05,
      "loss": 0.8125,
      "step": 7156
    },
    {
      "epoch": 4.739735099337748,
      "grad_norm": 1.1955960800541316,
      "learning_rate": 5.064310257322336e-05,
      "loss": 0.8125,
      "step": 7157
    },
    {
      "epoch": 4.740397350993377,
      "grad_norm": 1.1938773778790783,
      "learning_rate": 5.059310711413861e-05,
      "loss": 0.9297,
      "step": 7158
    },
    {
      "epoch": 4.741059602649006,
      "grad_norm": 1.230874836972059,
      "learning_rate": 5.054313133875833e-05,
      "loss": 0.8633,
      "step": 7159
    },
    {
      "epoch": 4.741721854304636,
      "grad_norm": 1.1991547294273075,
      "learning_rate": 5.049317525697823e-05,
      "loss": 0.8086,
      "step": 7160
    },
    {
      "epoch": 4.742384105960265,
      "grad_norm": 1.066769628467188,
      "learning_rate": 5.0443238878690217e-05,
      "loss": 0.6992,
      "step": 7161
    },
    {
      "epoch": 4.743046357615894,
      "grad_norm": 1.1067580674252477,
      "learning_rate": 5.0393322213782365e-05,
      "loss": 0.7578,
      "step": 7162
    },
    {
      "epoch": 4.743708609271523,
      "grad_norm": 1.2021088167374228,
      "learning_rate": 5.0343425272138606e-05,
      "loss": 0.875,
      "step": 7163
    },
    {
      "epoch": 4.744370860927153,
      "grad_norm": 1.348539761863938,
      "learning_rate": 5.029354806363921e-05,
      "loss": 1.0234,
      "step": 7164
    },
    {
      "epoch": 4.745033112582782,
      "grad_norm": 1.1541055714489195,
      "learning_rate": 5.024369059816035e-05,
      "loss": 0.8086,
      "step": 7165
    },
    {
      "epoch": 4.7456953642384105,
      "grad_norm": 1.2174528328538805,
      "learning_rate": 5.019385288557443e-05,
      "loss": 0.8672,
      "step": 7166
    },
    {
      "epoch": 4.7463576158940395,
      "grad_norm": 1.2117772520516918,
      "learning_rate": 5.0144034935749924e-05,
      "loss": 0.8516,
      "step": 7167
    },
    {
      "epoch": 4.747019867549669,
      "grad_norm": 1.1674270718010678,
      "learning_rate": 5.00942367585513e-05,
      "loss": 0.8828,
      "step": 7168
    },
    {
      "epoch": 4.747682119205298,
      "grad_norm": 1.0832283664466127,
      "learning_rate": 5.0044458363839144e-05,
      "loss": 0.7383,
      "step": 7169
    },
    {
      "epoch": 4.748344370860927,
      "grad_norm": 1.570906943764126,
      "learning_rate": 4.999469976147018e-05,
      "loss": 1.1094,
      "step": 7170
    },
    {
      "epoch": 4.749006622516556,
      "grad_norm": 1.1680387424937768,
      "learning_rate": 4.994496096129724e-05,
      "loss": 0.832,
      "step": 7171
    },
    {
      "epoch": 4.749668874172185,
      "grad_norm": 1.2290367897883943,
      "learning_rate": 4.9895241973169064e-05,
      "loss": 0.8477,
      "step": 7172
    },
    {
      "epoch": 4.750331125827815,
      "grad_norm": 1.1643297827947199,
      "learning_rate": 4.984554280693069e-05,
      "loss": 0.793,
      "step": 7173
    },
    {
      "epoch": 4.750993377483444,
      "grad_norm": 1.171335135311921,
      "learning_rate": 4.979586347242304e-05,
      "loss": 0.8242,
      "step": 7174
    },
    {
      "epoch": 4.751655629139073,
      "grad_norm": 1.0771202991260578,
      "learning_rate": 4.974620397948321e-05,
      "loss": 0.6875,
      "step": 7175
    },
    {
      "epoch": 4.752317880794702,
      "grad_norm": 1.1206934814728844,
      "learning_rate": 4.969656433794442e-05,
      "loss": 0.7617,
      "step": 7176
    },
    {
      "epoch": 4.752980132450331,
      "grad_norm": 1.1536339424463673,
      "learning_rate": 4.964694455763579e-05,
      "loss": 0.7852,
      "step": 7177
    },
    {
      "epoch": 4.7536423841059605,
      "grad_norm": 1.282018150722163,
      "learning_rate": 4.959734464838263e-05,
      "loss": 0.9492,
      "step": 7178
    },
    {
      "epoch": 4.7543046357615895,
      "grad_norm": 1.1007523001660744,
      "learning_rate": 4.9547764620006374e-05,
      "loss": 0.7344,
      "step": 7179
    },
    {
      "epoch": 4.754966887417218,
      "grad_norm": 1.2453785299082134,
      "learning_rate": 4.949820448232437e-05,
      "loss": 0.9297,
      "step": 7180
    },
    {
      "epoch": 4.755629139072847,
      "grad_norm": 1.277048308645179,
      "learning_rate": 4.944866424515003e-05,
      "loss": 0.9531,
      "step": 7181
    },
    {
      "epoch": 4.756291390728477,
      "grad_norm": 1.1832355031340849,
      "learning_rate": 4.939914391829296e-05,
      "loss": 0.8281,
      "step": 7182
    },
    {
      "epoch": 4.756953642384106,
      "grad_norm": 1.1944618209184032,
      "learning_rate": 4.93496435115588e-05,
      "loss": 0.8164,
      "step": 7183
    },
    {
      "epoch": 4.757615894039735,
      "grad_norm": 1.0784770790747988,
      "learning_rate": 4.93001630347491e-05,
      "loss": 0.8125,
      "step": 7184
    },
    {
      "epoch": 4.758278145695364,
      "grad_norm": 1.1928876695646673,
      "learning_rate": 4.925070249766164e-05,
      "loss": 0.9102,
      "step": 7185
    },
    {
      "epoch": 4.758940397350994,
      "grad_norm": 1.1339832149548208,
      "learning_rate": 4.92012619100901e-05,
      "loss": 0.8555,
      "step": 7186
    },
    {
      "epoch": 4.759602649006623,
      "grad_norm": 1.2044585172247242,
      "learning_rate": 4.9151841281824306e-05,
      "loss": 0.8125,
      "step": 7187
    },
    {
      "epoch": 4.760264900662252,
      "grad_norm": 1.1524824445690633,
      "learning_rate": 4.910244062265018e-05,
      "loss": 0.8438,
      "step": 7188
    },
    {
      "epoch": 4.760927152317881,
      "grad_norm": 1.2065007595189678,
      "learning_rate": 4.905305994234958e-05,
      "loss": 0.8242,
      "step": 7189
    },
    {
      "epoch": 4.76158940397351,
      "grad_norm": 1.0803387360724992,
      "learning_rate": 4.900369925070031e-05,
      "loss": 0.707,
      "step": 7190
    },
    {
      "epoch": 4.762251655629139,
      "grad_norm": 1.0858832587653307,
      "learning_rate": 4.895435855747659e-05,
      "loss": 0.7461,
      "step": 7191
    },
    {
      "epoch": 4.762913907284768,
      "grad_norm": 1.142773232752676,
      "learning_rate": 4.8905037872448315e-05,
      "loss": 0.832,
      "step": 7192
    },
    {
      "epoch": 4.763576158940397,
      "grad_norm": 1.3004154049541379,
      "learning_rate": 4.8855737205381503e-05,
      "loss": 0.9766,
      "step": 7193
    },
    {
      "epoch": 4.764238410596026,
      "grad_norm": 1.0666638591675446,
      "learning_rate": 4.880645656603831e-05,
      "loss": 0.7773,
      "step": 7194
    },
    {
      "epoch": 4.764900662251655,
      "grad_norm": 1.143019185769525,
      "learning_rate": 4.875719596417692e-05,
      "loss": 0.8906,
      "step": 7195
    },
    {
      "epoch": 4.765562913907285,
      "grad_norm": 1.202084605849404,
      "learning_rate": 4.870795540955138e-05,
      "loss": 0.8945,
      "step": 7196
    },
    {
      "epoch": 4.766225165562914,
      "grad_norm": 1.314272584439363,
      "learning_rate": 4.8658734911912014e-05,
      "loss": 0.8984,
      "step": 7197
    },
    {
      "epoch": 4.766887417218543,
      "grad_norm": 1.2843177845513354,
      "learning_rate": 4.86095344810049e-05,
      "loss": 0.9297,
      "step": 7198
    },
    {
      "epoch": 4.767549668874172,
      "grad_norm": 1.0176165726182163,
      "learning_rate": 4.8560354126572375e-05,
      "loss": 0.6602,
      "step": 7199
    },
    {
      "epoch": 4.768211920529802,
      "grad_norm": 1.1734179397178321,
      "learning_rate": 4.851119385835274e-05,
      "loss": 0.8359,
      "step": 7200
    },
    {
      "epoch": 4.768874172185431,
      "grad_norm": 1.0447069450270847,
      "learning_rate": 4.8462053686080264e-05,
      "loss": 0.6797,
      "step": 7201
    },
    {
      "epoch": 4.76953642384106,
      "grad_norm": 1.206181089514286,
      "learning_rate": 4.841293361948521e-05,
      "loss": 0.7656,
      "step": 7202
    },
    {
      "epoch": 4.7701986754966885,
      "grad_norm": 1.2038718853843677,
      "learning_rate": 4.836383366829395e-05,
      "loss": 0.8359,
      "step": 7203
    },
    {
      "epoch": 4.770860927152318,
      "grad_norm": 1.072471860513805,
      "learning_rate": 4.8314753842228886e-05,
      "loss": 0.6953,
      "step": 7204
    },
    {
      "epoch": 4.771523178807947,
      "grad_norm": 1.3284983442020013,
      "learning_rate": 4.82656941510083e-05,
      "loss": 0.9805,
      "step": 7205
    },
    {
      "epoch": 4.772185430463576,
      "grad_norm": 1.131439219344057,
      "learning_rate": 4.821665460434667e-05,
      "loss": 0.793,
      "step": 7206
    },
    {
      "epoch": 4.772847682119205,
      "grad_norm": 1.1942074755603045,
      "learning_rate": 4.8167635211954286e-05,
      "loss": 0.8477,
      "step": 7207
    },
    {
      "epoch": 4.773509933774834,
      "grad_norm": 1.222358636204658,
      "learning_rate": 4.81186359835376e-05,
      "loss": 0.8398,
      "step": 7208
    },
    {
      "epoch": 4.774172185430464,
      "grad_norm": 1.2014557929778804,
      "learning_rate": 4.806965692879907e-05,
      "loss": 0.8047,
      "step": 7209
    },
    {
      "epoch": 4.774834437086093,
      "grad_norm": 1.1699279034341865,
      "learning_rate": 4.8020698057437054e-05,
      "loss": 0.7227,
      "step": 7210
    },
    {
      "epoch": 4.775496688741722,
      "grad_norm": 1.1270894561086955,
      "learning_rate": 4.797175937914593e-05,
      "loss": 0.7227,
      "step": 7211
    },
    {
      "epoch": 4.776158940397351,
      "grad_norm": 1.2429824851769182,
      "learning_rate": 4.792284090361617e-05,
      "loss": 0.8242,
      "step": 7212
    },
    {
      "epoch": 4.77682119205298,
      "grad_norm": 1.3799122542420719,
      "learning_rate": 4.7873942640534234e-05,
      "loss": 0.9453,
      "step": 7213
    },
    {
      "epoch": 4.77748344370861,
      "grad_norm": 1.141064224729231,
      "learning_rate": 4.782506459958244e-05,
      "loss": 0.7344,
      "step": 7214
    },
    {
      "epoch": 4.7781456953642385,
      "grad_norm": 1.160508310016984,
      "learning_rate": 4.7776206790439256e-05,
      "loss": 0.7617,
      "step": 7215
    },
    {
      "epoch": 4.778807947019867,
      "grad_norm": 1.2270888783739182,
      "learning_rate": 4.772736922277915e-05,
      "loss": 0.8867,
      "step": 7216
    },
    {
      "epoch": 4.779470198675496,
      "grad_norm": 1.0712416443160644,
      "learning_rate": 4.7678551906272426e-05,
      "loss": 0.7188,
      "step": 7217
    },
    {
      "epoch": 4.780132450331126,
      "grad_norm": 1.1849827168098768,
      "learning_rate": 4.7629754850585475e-05,
      "loss": 0.8398,
      "step": 7218
    },
    {
      "epoch": 4.780794701986755,
      "grad_norm": 1.0937599741585422,
      "learning_rate": 4.758097806538069e-05,
      "loss": 0.707,
      "step": 7219
    },
    {
      "epoch": 4.781456953642384,
      "grad_norm": 1.1853483038709904,
      "learning_rate": 4.75322215603165e-05,
      "loss": 0.7773,
      "step": 7220
    },
    {
      "epoch": 4.782119205298013,
      "grad_norm": 1.164385890556684,
      "learning_rate": 4.748348534504714e-05,
      "loss": 0.7617,
      "step": 7221
    },
    {
      "epoch": 4.782781456953643,
      "grad_norm": 1.0737112184816073,
      "learning_rate": 4.743476942922305e-05,
      "loss": 0.6562,
      "step": 7222
    },
    {
      "epoch": 4.783443708609272,
      "grad_norm": 1.1136643304812823,
      "learning_rate": 4.7386073822490435e-05,
      "loss": 0.7461,
      "step": 7223
    },
    {
      "epoch": 4.784105960264901,
      "grad_norm": 1.1232421727022595,
      "learning_rate": 4.733739853449161e-05,
      "loss": 0.8008,
      "step": 7224
    },
    {
      "epoch": 4.78476821192053,
      "grad_norm": 1.126474321587426,
      "learning_rate": 4.728874357486495e-05,
      "loss": 0.7812,
      "step": 7225
    },
    {
      "epoch": 4.785430463576159,
      "grad_norm": 1.1873502910203428,
      "learning_rate": 4.724010895324458e-05,
      "loss": 0.8008,
      "step": 7226
    },
    {
      "epoch": 4.7860927152317885,
      "grad_norm": 1.0430732437740926,
      "learning_rate": 4.7191494679260705e-05,
      "loss": 0.7422,
      "step": 7227
    },
    {
      "epoch": 4.786754966887417,
      "grad_norm": 1.2281989077400766,
      "learning_rate": 4.714290076253955e-05,
      "loss": 0.7812,
      "step": 7228
    },
    {
      "epoch": 4.787417218543046,
      "grad_norm": 1.267796257062858,
      "learning_rate": 4.7094327212703296e-05,
      "loss": 0.832,
      "step": 7229
    },
    {
      "epoch": 4.788079470198675,
      "grad_norm": 1.0957312526178395,
      "learning_rate": 4.7045774039369995e-05,
      "loss": 0.7539,
      "step": 7230
    },
    {
      "epoch": 4.788741721854304,
      "grad_norm": 1.129782064356979,
      "learning_rate": 4.699724125215376e-05,
      "loss": 0.7695,
      "step": 7231
    },
    {
      "epoch": 4.789403973509934,
      "grad_norm": 1.2872351134949995,
      "learning_rate": 4.69487288606647e-05,
      "loss": 0.875,
      "step": 7232
    },
    {
      "epoch": 4.790066225165563,
      "grad_norm": 1.2303684799279542,
      "learning_rate": 4.6900236874508704e-05,
      "loss": 0.8242,
      "step": 7233
    },
    {
      "epoch": 4.790728476821192,
      "grad_norm": 1.108360759388955,
      "learning_rate": 4.685176530328787e-05,
      "loss": 0.7148,
      "step": 7234
    },
    {
      "epoch": 4.791390728476821,
      "grad_norm": 1.1420107292553023,
      "learning_rate": 4.6803314156600005e-05,
      "loss": 0.7695,
      "step": 7235
    },
    {
      "epoch": 4.79205298013245,
      "grad_norm": 1.0606104148340048,
      "learning_rate": 4.675488344403906e-05,
      "loss": 0.6836,
      "step": 7236
    },
    {
      "epoch": 4.79271523178808,
      "grad_norm": 1.0929066685688216,
      "learning_rate": 4.67064731751949e-05,
      "loss": 0.7422,
      "step": 7237
    },
    {
      "epoch": 4.793377483443709,
      "grad_norm": 1.130344448663671,
      "learning_rate": 4.665808335965327e-05,
      "loss": 0.7031,
      "step": 7238
    },
    {
      "epoch": 4.794039735099338,
      "grad_norm": 1.2692255276753004,
      "learning_rate": 4.660971400699586e-05,
      "loss": 0.8086,
      "step": 7239
    },
    {
      "epoch": 4.7947019867549665,
      "grad_norm": 1.1811844635449245,
      "learning_rate": 4.6561365126800414e-05,
      "loss": 0.8086,
      "step": 7240
    },
    {
      "epoch": 4.795364238410596,
      "grad_norm": 1.412055619775279,
      "learning_rate": 4.6513036728640616e-05,
      "loss": 0.9375,
      "step": 7241
    },
    {
      "epoch": 4.796026490066225,
      "grad_norm": 1.1680503005248548,
      "learning_rate": 4.6464728822085915e-05,
      "loss": 0.8242,
      "step": 7242
    },
    {
      "epoch": 4.796688741721854,
      "grad_norm": 1.227137338284097,
      "learning_rate": 4.6416441416701975e-05,
      "loss": 0.8008,
      "step": 7243
    },
    {
      "epoch": 4.797350993377483,
      "grad_norm": 1.1027400623574954,
      "learning_rate": 4.636817452205012e-05,
      "loss": 0.7227,
      "step": 7244
    },
    {
      "epoch": 4.798013245033113,
      "grad_norm": 1.2064146052760607,
      "learning_rate": 4.6319928147687804e-05,
      "loss": 0.8672,
      "step": 7245
    },
    {
      "epoch": 4.798675496688742,
      "grad_norm": 1.0645242665138956,
      "learning_rate": 4.6271702303168425e-05,
      "loss": 0.6914,
      "step": 7246
    },
    {
      "epoch": 4.799337748344371,
      "grad_norm": 1.1412294577166242,
      "learning_rate": 4.622349699804119e-05,
      "loss": 0.7734,
      "step": 7247
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.055176889692667,
      "learning_rate": 4.6175312241851214e-05,
      "loss": 0.6641,
      "step": 7248
    },
    {
      "epoch": 4.800662251655629,
      "grad_norm": 1.144096998756092,
      "learning_rate": 4.612714804413982e-05,
      "loss": 0.7383,
      "step": 7249
    },
    {
      "epoch": 4.801324503311259,
      "grad_norm": 1.1725531716423667,
      "learning_rate": 4.607900441444397e-05,
      "loss": 0.8555,
      "step": 7250
    },
    {
      "epoch": 4.8019867549668875,
      "grad_norm": 1.477249261971805,
      "learning_rate": 4.6030881362296633e-05,
      "loss": 1.1094,
      "step": 7251
    },
    {
      "epoch": 4.8026490066225165,
      "grad_norm": 1.103034167549984,
      "learning_rate": 4.598277889722674e-05,
      "loss": 0.7109,
      "step": 7252
    },
    {
      "epoch": 4.803311258278145,
      "grad_norm": 1.191031974826129,
      "learning_rate": 4.5934697028759185e-05,
      "loss": 0.7578,
      "step": 7253
    },
    {
      "epoch": 4.803973509933774,
      "grad_norm": 1.049916896903748,
      "learning_rate": 4.588663576641466e-05,
      "loss": 0.7227,
      "step": 7254
    },
    {
      "epoch": 4.804635761589404,
      "grad_norm": 1.106471165351405,
      "learning_rate": 4.583859511970991e-05,
      "loss": 0.7461,
      "step": 7255
    },
    {
      "epoch": 4.805298013245033,
      "grad_norm": 1.1065069565393768,
      "learning_rate": 4.5790575098157444e-05,
      "loss": 0.7422,
      "step": 7256
    },
    {
      "epoch": 4.805960264900662,
      "grad_norm": 1.0676115276861464,
      "learning_rate": 4.574257571126585e-05,
      "loss": 0.793,
      "step": 7257
    },
    {
      "epoch": 4.806622516556291,
      "grad_norm": 1.3485860573481734,
      "learning_rate": 4.5694596968539556e-05,
      "loss": 0.9492,
      "step": 7258
    },
    {
      "epoch": 4.807284768211921,
      "grad_norm": 1.0946302294894306,
      "learning_rate": 4.5646638879478905e-05,
      "loss": 0.7422,
      "step": 7259
    },
    {
      "epoch": 4.80794701986755,
      "grad_norm": 1.1644860782479798,
      "learning_rate": 4.5598701453580076e-05,
      "loss": 0.8398,
      "step": 7260
    },
    {
      "epoch": 4.808609271523179,
      "grad_norm": 1.2460177599886844,
      "learning_rate": 4.5550784700335285e-05,
      "loss": 0.8477,
      "step": 7261
    },
    {
      "epoch": 4.809271523178808,
      "grad_norm": 1.1573926437812292,
      "learning_rate": 4.550288862923264e-05,
      "loss": 0.832,
      "step": 7262
    },
    {
      "epoch": 4.8099337748344375,
      "grad_norm": 1.3365551600813912,
      "learning_rate": 4.545501324975601e-05,
      "loss": 0.9922,
      "step": 7263
    },
    {
      "epoch": 4.8105960264900665,
      "grad_norm": 1.41429628585556,
      "learning_rate": 4.540715857138539e-05,
      "loss": 0.9883,
      "step": 7264
    },
    {
      "epoch": 4.811258278145695,
      "grad_norm": 1.4715731329949375,
      "learning_rate": 4.535932460359643e-05,
      "loss": 1.1406,
      "step": 7265
    },
    {
      "epoch": 4.811920529801324,
      "grad_norm": 1.119232244641852,
      "learning_rate": 4.531151135586086e-05,
      "loss": 0.75,
      "step": 7266
    },
    {
      "epoch": 4.812582781456953,
      "grad_norm": 1.0968290081627292,
      "learning_rate": 4.526371883764632e-05,
      "loss": 0.7188,
      "step": 7267
    },
    {
      "epoch": 4.813245033112583,
      "grad_norm": 1.1985449875114043,
      "learning_rate": 4.521594705841615e-05,
      "loss": 0.7773,
      "step": 7268
    },
    {
      "epoch": 4.813907284768212,
      "grad_norm": 1.4390652035768554,
      "learning_rate": 4.516819602762983e-05,
      "loss": 1.0781,
      "step": 7269
    },
    {
      "epoch": 4.814569536423841,
      "grad_norm": 1.239334825772509,
      "learning_rate": 4.5120465754742514e-05,
      "loss": 0.8867,
      "step": 7270
    },
    {
      "epoch": 4.81523178807947,
      "grad_norm": 1.0662899560050692,
      "learning_rate": 4.5072756249205414e-05,
      "loss": 0.7305,
      "step": 7271
    },
    {
      "epoch": 4.815894039735099,
      "grad_norm": 1.211115363466212,
      "learning_rate": 4.5025067520465486e-05,
      "loss": 0.8672,
      "step": 7272
    },
    {
      "epoch": 4.816556291390729,
      "grad_norm": 1.1501225565940723,
      "learning_rate": 4.49773995779657e-05,
      "loss": 0.8203,
      "step": 7273
    },
    {
      "epoch": 4.817218543046358,
      "grad_norm": 1.0520112958921706,
      "learning_rate": 4.492975243114488e-05,
      "loss": 0.7148,
      "step": 7274
    },
    {
      "epoch": 4.817880794701987,
      "grad_norm": 1.15823393551042,
      "learning_rate": 4.488212608943765e-05,
      "loss": 0.7344,
      "step": 7275
    },
    {
      "epoch": 4.8185430463576155,
      "grad_norm": 1.1941209897866594,
      "learning_rate": 4.4834520562274565e-05,
      "loss": 0.9102,
      "step": 7276
    },
    {
      "epoch": 4.819205298013245,
      "grad_norm": 1.1144820690331463,
      "learning_rate": 4.47869358590821e-05,
      "loss": 0.7578,
      "step": 7277
    },
    {
      "epoch": 4.819867549668874,
      "grad_norm": 1.2278086652452413,
      "learning_rate": 4.4739371989282605e-05,
      "loss": 0.8047,
      "step": 7278
    },
    {
      "epoch": 4.820529801324503,
      "grad_norm": 1.133994046167678,
      "learning_rate": 4.469182896229417e-05,
      "loss": 0.7305,
      "step": 7279
    },
    {
      "epoch": 4.821192052980132,
      "grad_norm": 1.2798745930153606,
      "learning_rate": 4.464430678753096e-05,
      "loss": 0.8242,
      "step": 7280
    },
    {
      "epoch": 4.821854304635762,
      "grad_norm": 1.3500526098963184,
      "learning_rate": 4.4596805474402845e-05,
      "loss": 0.9648,
      "step": 7281
    },
    {
      "epoch": 4.822516556291391,
      "grad_norm": 1.2230366733534423,
      "learning_rate": 4.454932503231564e-05,
      "loss": 0.8594,
      "step": 7282
    },
    {
      "epoch": 4.82317880794702,
      "grad_norm": 1.1459466475464861,
      "learning_rate": 4.4501865470671095e-05,
      "loss": 0.8867,
      "step": 7283
    },
    {
      "epoch": 4.823841059602649,
      "grad_norm": 1.2965641426234706,
      "learning_rate": 4.445442679886667e-05,
      "loss": 0.9688,
      "step": 7284
    },
    {
      "epoch": 4.824503311258278,
      "grad_norm": 1.148667004616082,
      "learning_rate": 4.440700902629568e-05,
      "loss": 0.7461,
      "step": 7285
    },
    {
      "epoch": 4.825165562913908,
      "grad_norm": 1.1450806733741905,
      "learning_rate": 4.4359612162347594e-05,
      "loss": 0.875,
      "step": 7286
    },
    {
      "epoch": 4.825827814569537,
      "grad_norm": 1.2967620261005925,
      "learning_rate": 4.431223621640743e-05,
      "loss": 0.8945,
      "step": 7287
    },
    {
      "epoch": 4.8264900662251655,
      "grad_norm": 1.2618002949089853,
      "learning_rate": 4.4264881197856093e-05,
      "loss": 0.793,
      "step": 7288
    },
    {
      "epoch": 4.8271523178807945,
      "grad_norm": 1.0833462742744386,
      "learning_rate": 4.421754711607052e-05,
      "loss": 0.7305,
      "step": 7289
    },
    {
      "epoch": 4.827814569536423,
      "grad_norm": 1.2721430255211277,
      "learning_rate": 4.4170233980423414e-05,
      "loss": 0.8789,
      "step": 7290
    },
    {
      "epoch": 4.828476821192053,
      "grad_norm": 1.117055495391883,
      "learning_rate": 4.412294180028322e-05,
      "loss": 0.7891,
      "step": 7291
    },
    {
      "epoch": 4.829139072847682,
      "grad_norm": 1.063283806776567,
      "learning_rate": 4.407567058501445e-05,
      "loss": 0.6875,
      "step": 7292
    },
    {
      "epoch": 4.829801324503311,
      "grad_norm": 1.3350242838116635,
      "learning_rate": 4.4028420343977235e-05,
      "loss": 0.9336,
      "step": 7293
    },
    {
      "epoch": 4.83046357615894,
      "grad_norm": 1.1351646141885594,
      "learning_rate": 4.398119108652772e-05,
      "loss": 0.7695,
      "step": 7294
    },
    {
      "epoch": 4.83112582781457,
      "grad_norm": 1.0186945442175683,
      "learning_rate": 4.3933982822017876e-05,
      "loss": 0.6914,
      "step": 7295
    },
    {
      "epoch": 4.831788079470199,
      "grad_norm": 1.1620782701115138,
      "learning_rate": 4.388679555979544e-05,
      "loss": 0.7656,
      "step": 7296
    },
    {
      "epoch": 4.832450331125828,
      "grad_norm": 1.2153072972189078,
      "learning_rate": 4.383962930920399e-05,
      "loss": 0.8359,
      "step": 7297
    },
    {
      "epoch": 4.833112582781457,
      "grad_norm": 1.0455454477178099,
      "learning_rate": 4.3792484079583036e-05,
      "loss": 0.7266,
      "step": 7298
    },
    {
      "epoch": 4.8337748344370866,
      "grad_norm": 1.1255336400653617,
      "learning_rate": 4.3745359880267895e-05,
      "loss": 0.75,
      "step": 7299
    },
    {
      "epoch": 4.8344370860927155,
      "grad_norm": 1.3057101323437887,
      "learning_rate": 4.369825672058962e-05,
      "loss": 0.8125,
      "step": 7300
    },
    {
      "epoch": 4.835099337748344,
      "grad_norm": 1.1708970991693863,
      "learning_rate": 4.3651174609875284e-05,
      "loss": 0.7891,
      "step": 7301
    },
    {
      "epoch": 4.835761589403973,
      "grad_norm": 1.0417176198254414,
      "learning_rate": 4.360411355744756e-05,
      "loss": 0.7578,
      "step": 7302
    },
    {
      "epoch": 4.836423841059602,
      "grad_norm": 1.1206252567475228,
      "learning_rate": 4.355707357262515e-05,
      "loss": 0.6875,
      "step": 7303
    },
    {
      "epoch": 4.837086092715232,
      "grad_norm": 1.201093092778024,
      "learning_rate": 4.3510054664722546e-05,
      "loss": 0.832,
      "step": 7304
    },
    {
      "epoch": 4.837748344370861,
      "grad_norm": 1.0382061224308494,
      "learning_rate": 4.3463056843049995e-05,
      "loss": 0.6406,
      "step": 7305
    },
    {
      "epoch": 4.83841059602649,
      "grad_norm": 1.1789375649684357,
      "learning_rate": 4.341608011691349e-05,
      "loss": 0.7695,
      "step": 7306
    },
    {
      "epoch": 4.839072847682119,
      "grad_norm": 1.1584086134294098,
      "learning_rate": 4.3369124495615176e-05,
      "loss": 0.6914,
      "step": 7307
    },
    {
      "epoch": 4.839735099337748,
      "grad_norm": 1.1118714008213975,
      "learning_rate": 4.3322189988452674e-05,
      "loss": 0.7227,
      "step": 7308
    },
    {
      "epoch": 4.840397350993378,
      "grad_norm": 1.0953891782432186,
      "learning_rate": 4.327527660471954e-05,
      "loss": 0.7656,
      "step": 7309
    },
    {
      "epoch": 4.841059602649007,
      "grad_norm": 1.1831892743111774,
      "learning_rate": 4.3228384353705194e-05,
      "loss": 0.8242,
      "step": 7310
    },
    {
      "epoch": 4.841721854304636,
      "grad_norm": 1.1211450736125075,
      "learning_rate": 4.3181513244694905e-05,
      "loss": 0.6914,
      "step": 7311
    },
    {
      "epoch": 4.842384105960265,
      "grad_norm": 1.1579053118449358,
      "learning_rate": 4.313466328696955e-05,
      "loss": 0.8008,
      "step": 7312
    },
    {
      "epoch": 4.843046357615894,
      "grad_norm": 1.1254179317054087,
      "learning_rate": 4.30878344898061e-05,
      "loss": 0.7578,
      "step": 7313
    },
    {
      "epoch": 4.843708609271523,
      "grad_norm": 1.1586489666176043,
      "learning_rate": 4.304102686247708e-05,
      "loss": 0.8555,
      "step": 7314
    },
    {
      "epoch": 4.844370860927152,
      "grad_norm": 1.3016385643646549,
      "learning_rate": 4.299424041425099e-05,
      "loss": 0.8867,
      "step": 7315
    },
    {
      "epoch": 4.845033112582781,
      "grad_norm": 1.104516746616668,
      "learning_rate": 4.2947475154392124e-05,
      "loss": 0.75,
      "step": 7316
    },
    {
      "epoch": 4.845695364238411,
      "grad_norm": 1.3259959689923941,
      "learning_rate": 4.29007310921605e-05,
      "loss": 0.8867,
      "step": 7317
    },
    {
      "epoch": 4.84635761589404,
      "grad_norm": 1.2595064765106416,
      "learning_rate": 4.2854008236811915e-05,
      "loss": 0.8086,
      "step": 7318
    },
    {
      "epoch": 4.847019867549669,
      "grad_norm": 1.0927094435415716,
      "learning_rate": 4.28073065975981e-05,
      "loss": 0.6875,
      "step": 7319
    },
    {
      "epoch": 4.847682119205298,
      "grad_norm": 1.0176775356399204,
      "learning_rate": 4.276062618376654e-05,
      "loss": 0.7109,
      "step": 7320
    },
    {
      "epoch": 4.848344370860927,
      "grad_norm": 1.1131454656987099,
      "learning_rate": 4.27139670045604e-05,
      "loss": 0.7227,
      "step": 7321
    },
    {
      "epoch": 4.849006622516557,
      "grad_norm": 1.1894464678290488,
      "learning_rate": 4.2667329069218845e-05,
      "loss": 0.8008,
      "step": 7322
    },
    {
      "epoch": 4.849668874172186,
      "grad_norm": 1.2134539698532596,
      "learning_rate": 4.262071238697663e-05,
      "loss": 0.7812,
      "step": 7323
    },
    {
      "epoch": 4.850331125827815,
      "grad_norm": 1.1812973051351516,
      "learning_rate": 4.257411696706445e-05,
      "loss": 0.8047,
      "step": 7324
    },
    {
      "epoch": 4.8509933774834435,
      "grad_norm": 1.2616527120329524,
      "learning_rate": 4.252754281870867e-05,
      "loss": 0.8125,
      "step": 7325
    },
    {
      "epoch": 4.8516556291390724,
      "grad_norm": 1.023269705573654,
      "learning_rate": 4.248098995113155e-05,
      "loss": 0.6445,
      "step": 7326
    },
    {
      "epoch": 4.852317880794702,
      "grad_norm": 1.0977692850328138,
      "learning_rate": 4.243445837355111e-05,
      "loss": 0.7656,
      "step": 7327
    },
    {
      "epoch": 4.852980132450331,
      "grad_norm": 1.1585361827293845,
      "learning_rate": 4.2387948095181085e-05,
      "loss": 0.9336,
      "step": 7328
    },
    {
      "epoch": 4.85364238410596,
      "grad_norm": 1.0955392079935942,
      "learning_rate": 4.2341459125231106e-05,
      "loss": 0.7461,
      "step": 7329
    },
    {
      "epoch": 4.854304635761589,
      "grad_norm": 1.1209557351927768,
      "learning_rate": 4.229499147290645e-05,
      "loss": 0.7109,
      "step": 7330
    },
    {
      "epoch": 4.854966887417218,
      "grad_norm": 1.2349125637116352,
      "learning_rate": 4.224854514740827e-05,
      "loss": 0.8516,
      "step": 7331
    },
    {
      "epoch": 4.855629139072848,
      "grad_norm": 1.1365769374893557,
      "learning_rate": 4.220212015793353e-05,
      "loss": 0.7031,
      "step": 7332
    },
    {
      "epoch": 4.856291390728477,
      "grad_norm": 1.23597994581889,
      "learning_rate": 4.215571651367486e-05,
      "loss": 0.8164,
      "step": 7333
    },
    {
      "epoch": 4.856953642384106,
      "grad_norm": 1.2618447845655199,
      "learning_rate": 4.210933422382067e-05,
      "loss": 0.8633,
      "step": 7334
    },
    {
      "epoch": 4.857615894039736,
      "grad_norm": 1.2240117497689331,
      "learning_rate": 4.206297329755523e-05,
      "loss": 0.8594,
      "step": 7335
    },
    {
      "epoch": 4.8582781456953645,
      "grad_norm": 1.2345581190940509,
      "learning_rate": 4.201663374405857e-05,
      "loss": 0.832,
      "step": 7336
    },
    {
      "epoch": 4.8589403973509935,
      "grad_norm": 1.1263304846181992,
      "learning_rate": 4.197031557250638e-05,
      "loss": 0.8125,
      "step": 7337
    },
    {
      "epoch": 4.859602649006622,
      "grad_norm": 1.1565296619018002,
      "learning_rate": 4.1924018792070283e-05,
      "loss": 0.7891,
      "step": 7338
    },
    {
      "epoch": 4.860264900662251,
      "grad_norm": 1.1231202633561745,
      "learning_rate": 4.187774341191745e-05,
      "loss": 0.6758,
      "step": 7339
    },
    {
      "epoch": 4.860927152317881,
      "grad_norm": 1.2867019548779108,
      "learning_rate": 4.1831489441211016e-05,
      "loss": 0.8945,
      "step": 7340
    },
    {
      "epoch": 4.86158940397351,
      "grad_norm": 1.191010978987119,
      "learning_rate": 4.1785256889109825e-05,
      "loss": 0.8203,
      "step": 7341
    },
    {
      "epoch": 4.862251655629139,
      "grad_norm": 1.1319905310306495,
      "learning_rate": 4.1739045764768414e-05,
      "loss": 0.7578,
      "step": 7342
    },
    {
      "epoch": 4.862913907284768,
      "grad_norm": 1.0638498665992946,
      "learning_rate": 4.169285607733703e-05,
      "loss": 0.6641,
      "step": 7343
    },
    {
      "epoch": 4.863576158940397,
      "grad_norm": 1.161734758139168,
      "learning_rate": 4.1646687835961915e-05,
      "loss": 0.7656,
      "step": 7344
    },
    {
      "epoch": 4.864238410596027,
      "grad_norm": 1.1280494274083037,
      "learning_rate": 4.160054104978485e-05,
      "loss": 0.7539,
      "step": 7345
    },
    {
      "epoch": 4.864900662251656,
      "grad_norm": 1.0537209988343053,
      "learning_rate": 4.155441572794338e-05,
      "loss": 0.7383,
      "step": 7346
    },
    {
      "epoch": 4.865562913907285,
      "grad_norm": 1.2399869568515673,
      "learning_rate": 4.150831187957088e-05,
      "loss": 0.8008,
      "step": 7347
    },
    {
      "epoch": 4.866225165562914,
      "grad_norm": 1.1484586856293266,
      "learning_rate": 4.1462229513796474e-05,
      "loss": 0.8047,
      "step": 7348
    },
    {
      "epoch": 4.866887417218543,
      "grad_norm": 1.195994033044378,
      "learning_rate": 4.141616863974494e-05,
      "loss": 0.793,
      "step": 7349
    },
    {
      "epoch": 4.867549668874172,
      "grad_norm": 1.0873323855434467,
      "learning_rate": 4.1370129266536926e-05,
      "loss": 0.7305,
      "step": 7350
    },
    {
      "epoch": 4.868211920529801,
      "grad_norm": 1.1855084135421061,
      "learning_rate": 4.132411140328868e-05,
      "loss": 0.7891,
      "step": 7351
    },
    {
      "epoch": 4.86887417218543,
      "grad_norm": 1.1897271924176702,
      "learning_rate": 4.127811505911229e-05,
      "loss": 0.8047,
      "step": 7352
    },
    {
      "epoch": 4.869536423841059,
      "grad_norm": 1.2456769290462968,
      "learning_rate": 4.123214024311564e-05,
      "loss": 0.8203,
      "step": 7353
    },
    {
      "epoch": 4.870198675496689,
      "grad_norm": 1.1353353165025122,
      "learning_rate": 4.1186186964402193e-05,
      "loss": 0.7031,
      "step": 7354
    },
    {
      "epoch": 4.870860927152318,
      "grad_norm": 1.3195166399252585,
      "learning_rate": 4.114025523207119e-05,
      "loss": 0.8633,
      "step": 7355
    },
    {
      "epoch": 4.871523178807947,
      "grad_norm": 1.1461496251778847,
      "learning_rate": 4.109434505521769e-05,
      "loss": 0.707,
      "step": 7356
    },
    {
      "epoch": 4.872185430463576,
      "grad_norm": 1.0749130449152144,
      "learning_rate": 4.104845644293247e-05,
      "loss": 0.6992,
      "step": 7357
    },
    {
      "epoch": 4.872847682119206,
      "grad_norm": 1.290555725984743,
      "learning_rate": 4.100258940430192e-05,
      "loss": 0.8125,
      "step": 7358
    },
    {
      "epoch": 4.873509933774835,
      "grad_norm": 1.094416815594876,
      "learning_rate": 4.0956743948408335e-05,
      "loss": 0.7148,
      "step": 7359
    },
    {
      "epoch": 4.874172185430464,
      "grad_norm": 1.1350508059230782,
      "learning_rate": 4.0910920084329546e-05,
      "loss": 0.7422,
      "step": 7360
    },
    {
      "epoch": 4.8748344370860925,
      "grad_norm": 1.0414071647330025,
      "learning_rate": 4.0865117821139234e-05,
      "loss": 0.7031,
      "step": 7361
    },
    {
      "epoch": 4.8754966887417215,
      "grad_norm": 1.3061344760682405,
      "learning_rate": 4.081933716790684e-05,
      "loss": 0.8086,
      "step": 7362
    },
    {
      "epoch": 4.876158940397351,
      "grad_norm": 1.252790970920827,
      "learning_rate": 4.077357813369734e-05,
      "loss": 0.8477,
      "step": 7363
    },
    {
      "epoch": 4.87682119205298,
      "grad_norm": 1.1621631190292105,
      "learning_rate": 4.072784072757163e-05,
      "loss": 0.8047,
      "step": 7364
    },
    {
      "epoch": 4.877483443708609,
      "grad_norm": 1.1359621830483646,
      "learning_rate": 4.068212495858625e-05,
      "loss": 0.8125,
      "step": 7365
    },
    {
      "epoch": 4.878145695364238,
      "grad_norm": 0.9911549000381724,
      "learning_rate": 4.0636430835793435e-05,
      "loss": 0.6992,
      "step": 7366
    },
    {
      "epoch": 4.878807947019867,
      "grad_norm": 1.197650709045364,
      "learning_rate": 4.059075836824106e-05,
      "loss": 0.8242,
      "step": 7367
    },
    {
      "epoch": 4.879470198675497,
      "grad_norm": 1.1239053335877072,
      "learning_rate": 4.054510756497287e-05,
      "loss": 0.8359,
      "step": 7368
    },
    {
      "epoch": 4.880132450331126,
      "grad_norm": 1.2112996895774413,
      "learning_rate": 4.04994784350283e-05,
      "loss": 0.8086,
      "step": 7369
    },
    {
      "epoch": 4.880794701986755,
      "grad_norm": 1.1607143138848424,
      "learning_rate": 4.0453870987442326e-05,
      "loss": 0.8281,
      "step": 7370
    },
    {
      "epoch": 4.881456953642384,
      "grad_norm": 1.2662302782815515,
      "learning_rate": 4.040828523124585e-05,
      "loss": 0.8594,
      "step": 7371
    },
    {
      "epoch": 4.882119205298014,
      "grad_norm": 1.134417942226216,
      "learning_rate": 4.036272117546527e-05,
      "loss": 0.7266,
      "step": 7372
    },
    {
      "epoch": 4.8827814569536425,
      "grad_norm": 1.183900115758477,
      "learning_rate": 4.031717882912288e-05,
      "loss": 0.8242,
      "step": 7373
    },
    {
      "epoch": 4.8834437086092715,
      "grad_norm": 1.1413414710918592,
      "learning_rate": 4.0271658201236564e-05,
      "loss": 0.7188,
      "step": 7374
    },
    {
      "epoch": 4.8841059602649,
      "grad_norm": 1.1853129716550297,
      "learning_rate": 4.0226159300819935e-05,
      "loss": 0.793,
      "step": 7375
    },
    {
      "epoch": 4.88476821192053,
      "grad_norm": 1.2455545688884337,
      "learning_rate": 4.018068213688223e-05,
      "loss": 0.8477,
      "step": 7376
    },
    {
      "epoch": 4.885430463576159,
      "grad_norm": 1.268705041051395,
      "learning_rate": 4.0135226718428503e-05,
      "loss": 0.8633,
      "step": 7377
    },
    {
      "epoch": 4.886092715231788,
      "grad_norm": 1.131993471624313,
      "learning_rate": 4.008979305445949e-05,
      "loss": 0.7773,
      "step": 7378
    },
    {
      "epoch": 4.886754966887417,
      "grad_norm": 1.1500662783551423,
      "learning_rate": 4.0044381153971474e-05,
      "loss": 0.7031,
      "step": 7379
    },
    {
      "epoch": 4.887417218543046,
      "grad_norm": 1.2138783555705959,
      "learning_rate": 3.99989910259566e-05,
      "loss": 0.918,
      "step": 7380
    },
    {
      "epoch": 4.888079470198676,
      "grad_norm": 1.2309992196131647,
      "learning_rate": 3.9953622679402684e-05,
      "loss": 0.8516,
      "step": 7381
    },
    {
      "epoch": 4.888741721854305,
      "grad_norm": 1.1910162264895454,
      "learning_rate": 3.99082761232931e-05,
      "loss": 0.7656,
      "step": 7382
    },
    {
      "epoch": 4.889403973509934,
      "grad_norm": 1.2438151380717715,
      "learning_rate": 3.986295136660696e-05,
      "loss": 0.8242,
      "step": 7383
    },
    {
      "epoch": 4.890066225165563,
      "grad_norm": 1.1880058028807245,
      "learning_rate": 3.981764841831913e-05,
      "loss": 0.8984,
      "step": 7384
    },
    {
      "epoch": 4.890728476821192,
      "grad_norm": 1.1206723809412584,
      "learning_rate": 3.977236728740015e-05,
      "loss": 0.6914,
      "step": 7385
    },
    {
      "epoch": 4.891390728476821,
      "grad_norm": 1.1357520591782848,
      "learning_rate": 3.972710798281614e-05,
      "loss": 0.7812,
      "step": 7386
    },
    {
      "epoch": 4.89205298013245,
      "grad_norm": 1.0862549076758212,
      "learning_rate": 3.968187051352903e-05,
      "loss": 0.6953,
      "step": 7387
    },
    {
      "epoch": 4.892715231788079,
      "grad_norm": 1.151811337553754,
      "learning_rate": 3.963665488849625e-05,
      "loss": 0.7695,
      "step": 7388
    },
    {
      "epoch": 4.893377483443708,
      "grad_norm": 1.1061325935668704,
      "learning_rate": 3.9591461116671086e-05,
      "loss": 0.793,
      "step": 7389
    },
    {
      "epoch": 4.894039735099338,
      "grad_norm": 1.3668979957365022,
      "learning_rate": 3.954628920700246e-05,
      "loss": 0.9766,
      "step": 7390
    },
    {
      "epoch": 4.894701986754967,
      "grad_norm": 1.2967495649681884,
      "learning_rate": 3.9501139168434874e-05,
      "loss": 0.8945,
      "step": 7391
    },
    {
      "epoch": 4.895364238410596,
      "grad_norm": 1.1139410964612915,
      "learning_rate": 3.945601100990854e-05,
      "loss": 0.7422,
      "step": 7392
    },
    {
      "epoch": 4.896026490066225,
      "grad_norm": 1.2826204265690138,
      "learning_rate": 3.941090474035935e-05,
      "loss": 0.9492,
      "step": 7393
    },
    {
      "epoch": 4.896688741721855,
      "grad_norm": 1.2576171826140257,
      "learning_rate": 3.9365820368718934e-05,
      "loss": 0.8477,
      "step": 7394
    },
    {
      "epoch": 4.897350993377484,
      "grad_norm": 1.1058632325962618,
      "learning_rate": 3.932075790391441e-05,
      "loss": 0.7617,
      "step": 7395
    },
    {
      "epoch": 4.898013245033113,
      "grad_norm": 1.1177047076256934,
      "learning_rate": 3.927571735486878e-05,
      "loss": 0.7344,
      "step": 7396
    },
    {
      "epoch": 4.898675496688742,
      "grad_norm": 0.9661482192716608,
      "learning_rate": 3.923069873050046e-05,
      "loss": 0.6016,
      "step": 7397
    },
    {
      "epoch": 4.8993377483443705,
      "grad_norm": 1.1196751148827702,
      "learning_rate": 3.918570203972375e-05,
      "loss": 0.8906,
      "step": 7398
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.2279922519779698,
      "learning_rate": 3.9140727291448496e-05,
      "loss": 0.9023,
      "step": 7399
    },
    {
      "epoch": 4.900662251655629,
      "grad_norm": 1.2380465029829928,
      "learning_rate": 3.909577449458021e-05,
      "loss": 0.9336,
      "step": 7400
    },
    {
      "epoch": 4.901324503311258,
      "grad_norm": 1.0945563577955695,
      "learning_rate": 3.905084365801997e-05,
      "loss": 0.7578,
      "step": 7401
    },
    {
      "epoch": 4.901986754966887,
      "grad_norm": 1.1520975178139263,
      "learning_rate": 3.900593479066475e-05,
      "loss": 0.7578,
      "step": 7402
    },
    {
      "epoch": 4.902649006622516,
      "grad_norm": 1.1320352380759897,
      "learning_rate": 3.896104790140695e-05,
      "loss": 0.7422,
      "step": 7403
    },
    {
      "epoch": 4.903311258278146,
      "grad_norm": 1.0634766131282478,
      "learning_rate": 3.891618299913464e-05,
      "loss": 0.707,
      "step": 7404
    },
    {
      "epoch": 4.903973509933775,
      "grad_norm": 1.1500455703912453,
      "learning_rate": 3.887134009273165e-05,
      "loss": 0.7031,
      "step": 7405
    },
    {
      "epoch": 4.904635761589404,
      "grad_norm": 1.15901713103062,
      "learning_rate": 3.882651919107739e-05,
      "loss": 0.7852,
      "step": 7406
    },
    {
      "epoch": 4.905298013245033,
      "grad_norm": 1.0764623165055063,
      "learning_rate": 3.878172030304686e-05,
      "loss": 0.6641,
      "step": 7407
    },
    {
      "epoch": 4.905960264900663,
      "grad_norm": 1.3904913464729862,
      "learning_rate": 3.873694343751083e-05,
      "loss": 0.8633,
      "step": 7408
    },
    {
      "epoch": 4.906622516556292,
      "grad_norm": 1.196145800648721,
      "learning_rate": 3.869218860333555e-05,
      "loss": 0.8242,
      "step": 7409
    },
    {
      "epoch": 4.9072847682119205,
      "grad_norm": 1.1121733244338483,
      "learning_rate": 3.8647455809383006e-05,
      "loss": 0.7383,
      "step": 7410
    },
    {
      "epoch": 4.907947019867549,
      "grad_norm": 1.1811706188537838,
      "learning_rate": 3.8602745064510895e-05,
      "loss": 0.7461,
      "step": 7411
    },
    {
      "epoch": 4.908609271523179,
      "grad_norm": 1.2397208688265213,
      "learning_rate": 3.8558056377572376e-05,
      "loss": 0.7773,
      "step": 7412
    },
    {
      "epoch": 4.909271523178808,
      "grad_norm": 1.2785836511803408,
      "learning_rate": 3.85133897574163e-05,
      "loss": 0.832,
      "step": 7413
    },
    {
      "epoch": 4.909933774834437,
      "grad_norm": 1.2637534615777861,
      "learning_rate": 3.846874521288719e-05,
      "loss": 0.8945,
      "step": 7414
    },
    {
      "epoch": 4.910596026490066,
      "grad_norm": 1.106875400447583,
      "learning_rate": 3.842412275282524e-05,
      "loss": 0.7266,
      "step": 7415
    },
    {
      "epoch": 4.911258278145695,
      "grad_norm": 1.150522406354129,
      "learning_rate": 3.837952238606613e-05,
      "loss": 0.7656,
      "step": 7416
    },
    {
      "epoch": 4.911920529801325,
      "grad_norm": 1.3162627602555983,
      "learning_rate": 3.833494412144126e-05,
      "loss": 0.957,
      "step": 7417
    },
    {
      "epoch": 4.912582781456954,
      "grad_norm": 1.1418974662127217,
      "learning_rate": 3.829038796777771e-05,
      "loss": 0.6836,
      "step": 7418
    },
    {
      "epoch": 4.913245033112583,
      "grad_norm": 1.1073578960391084,
      "learning_rate": 3.824585393389799e-05,
      "loss": 0.6797,
      "step": 7419
    },
    {
      "epoch": 4.913907284768212,
      "grad_norm": 1.0973551710581024,
      "learning_rate": 3.820134202862047e-05,
      "loss": 0.7539,
      "step": 7420
    },
    {
      "epoch": 4.914569536423841,
      "grad_norm": 1.215744566754476,
      "learning_rate": 3.815685226075891e-05,
      "loss": 0.8398,
      "step": 7421
    },
    {
      "epoch": 4.9152317880794705,
      "grad_norm": 1.1106865172834657,
      "learning_rate": 3.811238463912284e-05,
      "loss": 0.7383,
      "step": 7422
    },
    {
      "epoch": 4.915894039735099,
      "grad_norm": 1.263285886371103,
      "learning_rate": 3.8067939172517405e-05,
      "loss": 0.8633,
      "step": 7423
    },
    {
      "epoch": 4.916556291390728,
      "grad_norm": 1.0446859986871406,
      "learning_rate": 3.802351586974326e-05,
      "loss": 0.6133,
      "step": 7424
    },
    {
      "epoch": 4.917218543046357,
      "grad_norm": 1.1841798161897914,
      "learning_rate": 3.797911473959673e-05,
      "loss": 0.8125,
      "step": 7425
    },
    {
      "epoch": 4.917880794701987,
      "grad_norm": 1.1782557272645842,
      "learning_rate": 3.793473579086973e-05,
      "loss": 0.8125,
      "step": 7426
    },
    {
      "epoch": 4.918543046357616,
      "grad_norm": 1.2110764770630233,
      "learning_rate": 3.789037903234989e-05,
      "loss": 0.7656,
      "step": 7427
    },
    {
      "epoch": 4.919205298013245,
      "grad_norm": 1.27252552564894,
      "learning_rate": 3.784604447282024e-05,
      "loss": 0.8398,
      "step": 7428
    },
    {
      "epoch": 4.919867549668874,
      "grad_norm": 1.1759647720892619,
      "learning_rate": 3.7801732121059645e-05,
      "loss": 0.7617,
      "step": 7429
    },
    {
      "epoch": 4.920529801324504,
      "grad_norm": 1.2524792378718823,
      "learning_rate": 3.775744198584234e-05,
      "loss": 0.9336,
      "step": 7430
    },
    {
      "epoch": 4.921192052980133,
      "grad_norm": 1.3238806590255883,
      "learning_rate": 3.771317407593836e-05,
      "loss": 0.8438,
      "step": 7431
    },
    {
      "epoch": 4.921854304635762,
      "grad_norm": 1.29174805138259,
      "learning_rate": 3.766892840011327e-05,
      "loss": 0.8984,
      "step": 7432
    },
    {
      "epoch": 4.922516556291391,
      "grad_norm": 1.3434236817615206,
      "learning_rate": 3.76247049671282e-05,
      "loss": 0.957,
      "step": 7433
    },
    {
      "epoch": 4.92317880794702,
      "grad_norm": 1.0572505452828616,
      "learning_rate": 3.75805037857398e-05,
      "loss": 0.6602,
      "step": 7434
    },
    {
      "epoch": 4.923841059602649,
      "grad_norm": 1.2583039369837488,
      "learning_rate": 3.753632486470057e-05,
      "loss": 0.7773,
      "step": 7435
    },
    {
      "epoch": 4.924503311258278,
      "grad_norm": 1.2320566017604937,
      "learning_rate": 3.7492168212758385e-05,
      "loss": 0.793,
      "step": 7436
    },
    {
      "epoch": 4.925165562913907,
      "grad_norm": 1.2195255261468456,
      "learning_rate": 3.74480338386567e-05,
      "loss": 0.8242,
      "step": 7437
    },
    {
      "epoch": 4.925827814569536,
      "grad_norm": 1.1563347398383332,
      "learning_rate": 3.740392175113468e-05,
      "loss": 0.8086,
      "step": 7438
    },
    {
      "epoch": 4.926490066225165,
      "grad_norm": 1.2029028194282652,
      "learning_rate": 3.735983195892709e-05,
      "loss": 0.8047,
      "step": 7439
    },
    {
      "epoch": 4.927152317880795,
      "grad_norm": 1.3334485563015717,
      "learning_rate": 3.731576447076413e-05,
      "loss": 0.875,
      "step": 7440
    },
    {
      "epoch": 4.927814569536424,
      "grad_norm": 1.2242311989144472,
      "learning_rate": 3.727171929537165e-05,
      "loss": 0.8281,
      "step": 7441
    },
    {
      "epoch": 4.928476821192053,
      "grad_norm": 1.1574454437553914,
      "learning_rate": 3.722769644147113e-05,
      "loss": 0.7031,
      "step": 7442
    },
    {
      "epoch": 4.929139072847682,
      "grad_norm": 1.314755917685958,
      "learning_rate": 3.7183695917779654e-05,
      "loss": 0.9531,
      "step": 7443
    },
    {
      "epoch": 4.929801324503311,
      "grad_norm": 1.2808965441751783,
      "learning_rate": 3.7139717733009755e-05,
      "loss": 0.875,
      "step": 7444
    },
    {
      "epoch": 4.930463576158941,
      "grad_norm": 1.1608377121142828,
      "learning_rate": 3.709576189586967e-05,
      "loss": 0.75,
      "step": 7445
    },
    {
      "epoch": 4.9311258278145695,
      "grad_norm": 1.0637980299028116,
      "learning_rate": 3.70518284150631e-05,
      "loss": 0.7148,
      "step": 7446
    },
    {
      "epoch": 4.9317880794701985,
      "grad_norm": 1.2466151028087193,
      "learning_rate": 3.700791729928941e-05,
      "loss": 0.8633,
      "step": 7447
    },
    {
      "epoch": 4.932450331125828,
      "grad_norm": 1.1503797462143852,
      "learning_rate": 3.696402855724355e-05,
      "loss": 0.7539,
      "step": 7448
    },
    {
      "epoch": 4.933112582781457,
      "grad_norm": 1.1347038740165059,
      "learning_rate": 3.6920162197615946e-05,
      "loss": 0.7734,
      "step": 7449
    },
    {
      "epoch": 4.933774834437086,
      "grad_norm": 1.1479145721749426,
      "learning_rate": 3.68763182290926e-05,
      "loss": 0.7578,
      "step": 7450
    },
    {
      "epoch": 4.934437086092715,
      "grad_norm": 1.17119451162271,
      "learning_rate": 3.683249666035518e-05,
      "loss": 0.7539,
      "step": 7451
    },
    {
      "epoch": 4.935099337748344,
      "grad_norm": 1.255683675399053,
      "learning_rate": 3.6788697500080873e-05,
      "loss": 0.8203,
      "step": 7452
    },
    {
      "epoch": 4.935761589403974,
      "grad_norm": 1.2166069484502346,
      "learning_rate": 3.674492075694235e-05,
      "loss": 0.7617,
      "step": 7453
    },
    {
      "epoch": 4.936423841059603,
      "grad_norm": 1.293524751942597,
      "learning_rate": 3.6701166439607986e-05,
      "loss": 0.9219,
      "step": 7454
    },
    {
      "epoch": 4.937086092715232,
      "grad_norm": 1.1380733954306739,
      "learning_rate": 3.665743455674156e-05,
      "loss": 0.707,
      "step": 7455
    },
    {
      "epoch": 4.937748344370861,
      "grad_norm": 1.1057824056770715,
      "learning_rate": 3.6613725117002504e-05,
      "loss": 0.6406,
      "step": 7456
    },
    {
      "epoch": 4.93841059602649,
      "grad_norm": 1.3008113397257013,
      "learning_rate": 3.657003812904587e-05,
      "loss": 0.9531,
      "step": 7457
    },
    {
      "epoch": 4.9390728476821195,
      "grad_norm": 1.2433896470831225,
      "learning_rate": 3.652637360152206e-05,
      "loss": 0.9805,
      "step": 7458
    },
    {
      "epoch": 4.9397350993377485,
      "grad_norm": 1.128983515437301,
      "learning_rate": 3.648273154307722e-05,
      "loss": 0.7266,
      "step": 7459
    },
    {
      "epoch": 4.940397350993377,
      "grad_norm": 1.02943786139513,
      "learning_rate": 3.643911196235301e-05,
      "loss": 0.7305,
      "step": 7460
    },
    {
      "epoch": 4.941059602649006,
      "grad_norm": 1.198739269944828,
      "learning_rate": 3.639551486798656e-05,
      "loss": 0.8594,
      "step": 7461
    },
    {
      "epoch": 4.941721854304635,
      "grad_norm": 1.1996968657632952,
      "learning_rate": 3.635194026861055e-05,
      "loss": 0.8008,
      "step": 7462
    },
    {
      "epoch": 4.942384105960265,
      "grad_norm": 1.1052941105463687,
      "learning_rate": 3.630838817285331e-05,
      "loss": 0.75,
      "step": 7463
    },
    {
      "epoch": 4.943046357615894,
      "grad_norm": 1.1270742449877194,
      "learning_rate": 3.626485858933869e-05,
      "loss": 0.6758,
      "step": 7464
    },
    {
      "epoch": 4.943708609271523,
      "grad_norm": 1.1048820578428868,
      "learning_rate": 3.622135152668594e-05,
      "loss": 0.6602,
      "step": 7465
    },
    {
      "epoch": 4.944370860927152,
      "grad_norm": 1.4043697484814968,
      "learning_rate": 3.6177866993510075e-05,
      "loss": 0.9727,
      "step": 7466
    },
    {
      "epoch": 4.945033112582782,
      "grad_norm": 1.2444902663189326,
      "learning_rate": 3.6134404998421427e-05,
      "loss": 0.8672,
      "step": 7467
    },
    {
      "epoch": 4.945695364238411,
      "grad_norm": 1.3427258463420468,
      "learning_rate": 3.6090965550026e-05,
      "loss": 0.8984,
      "step": 7468
    },
    {
      "epoch": 4.94635761589404,
      "grad_norm": 1.1172602365098394,
      "learning_rate": 3.604754865692538e-05,
      "loss": 0.8008,
      "step": 7469
    },
    {
      "epoch": 4.947019867549669,
      "grad_norm": 1.1912974810649115,
      "learning_rate": 3.600415432771654e-05,
      "loss": 0.7266,
      "step": 7470
    },
    {
      "epoch": 4.947682119205298,
      "grad_norm": 1.2521758416805533,
      "learning_rate": 3.596078257099199e-05,
      "loss": 0.8164,
      "step": 7471
    },
    {
      "epoch": 4.948344370860927,
      "grad_norm": 1.048188369889491,
      "learning_rate": 3.591743339533998e-05,
      "loss": 0.668,
      "step": 7472
    },
    {
      "epoch": 4.949006622516556,
      "grad_norm": 1.1878281854425727,
      "learning_rate": 3.587410680934407e-05,
      "loss": 0.8398,
      "step": 7473
    },
    {
      "epoch": 4.949668874172185,
      "grad_norm": 1.1778401486222665,
      "learning_rate": 3.583080282158339e-05,
      "loss": 0.6836,
      "step": 7474
    },
    {
      "epoch": 4.950331125827814,
      "grad_norm": 1.160314496342303,
      "learning_rate": 3.578752144063266e-05,
      "loss": 0.8047,
      "step": 7475
    },
    {
      "epoch": 4.950993377483444,
      "grad_norm": 1.1922544008096467,
      "learning_rate": 3.574426267506212e-05,
      "loss": 0.7461,
      "step": 7476
    },
    {
      "epoch": 4.951655629139073,
      "grad_norm": 1.225946826593522,
      "learning_rate": 3.570102653343743e-05,
      "loss": 0.8477,
      "step": 7477
    },
    {
      "epoch": 4.952317880794702,
      "grad_norm": 1.2179236593698697,
      "learning_rate": 3.5657813024319934e-05,
      "loss": 0.8398,
      "step": 7478
    },
    {
      "epoch": 4.952980132450331,
      "grad_norm": 0.9680398261141759,
      "learning_rate": 3.5614622156266306e-05,
      "loss": 0.6797,
      "step": 7479
    },
    {
      "epoch": 4.95364238410596,
      "grad_norm": 1.252135755973023,
      "learning_rate": 3.5571453937828856e-05,
      "loss": 0.8086,
      "step": 7480
    },
    {
      "epoch": 4.95430463576159,
      "grad_norm": 1.137597017348754,
      "learning_rate": 3.552830837755546e-05,
      "loss": 0.8164,
      "step": 7481
    },
    {
      "epoch": 4.954966887417219,
      "grad_norm": 1.3740265276835795,
      "learning_rate": 3.5485185483989384e-05,
      "loss": 0.9258,
      "step": 7482
    },
    {
      "epoch": 4.9556291390728475,
      "grad_norm": 1.2489682887227498,
      "learning_rate": 3.5442085265669414e-05,
      "loss": 0.8242,
      "step": 7483
    },
    {
      "epoch": 4.9562913907284765,
      "grad_norm": 1.130198087835507,
      "learning_rate": 3.539900773112992e-05,
      "loss": 0.6875,
      "step": 7484
    },
    {
      "epoch": 4.956953642384106,
      "grad_norm": 1.0476198496935811,
      "learning_rate": 3.5355952888900804e-05,
      "loss": 0.6602,
      "step": 7485
    },
    {
      "epoch": 4.957615894039735,
      "grad_norm": 1.1307172134039818,
      "learning_rate": 3.531292074750731e-05,
      "loss": 0.6953,
      "step": 7486
    },
    {
      "epoch": 4.958278145695364,
      "grad_norm": 1.1625804782409128,
      "learning_rate": 3.526991131547039e-05,
      "loss": 0.7773,
      "step": 7487
    },
    {
      "epoch": 4.958940397350993,
      "grad_norm": 1.2319314865879187,
      "learning_rate": 3.522692460130634e-05,
      "loss": 0.8477,
      "step": 7488
    },
    {
      "epoch": 4.959602649006623,
      "grad_norm": 1.2139190941692548,
      "learning_rate": 3.518396061352704e-05,
      "loss": 0.8086,
      "step": 7489
    },
    {
      "epoch": 4.960264900662252,
      "grad_norm": 1.137661099085646,
      "learning_rate": 3.5141019360639923e-05,
      "loss": 0.8203,
      "step": 7490
    },
    {
      "epoch": 4.960927152317881,
      "grad_norm": 1.1122517167542987,
      "learning_rate": 3.509810085114777e-05,
      "loss": 0.6562,
      "step": 7491
    },
    {
      "epoch": 4.96158940397351,
      "grad_norm": 1.2293611851743,
      "learning_rate": 3.505520509354886e-05,
      "loss": 0.8125,
      "step": 7492
    },
    {
      "epoch": 4.962251655629139,
      "grad_norm": 1.3160409703409424,
      "learning_rate": 3.501233209633725e-05,
      "loss": 0.8945,
      "step": 7493
    },
    {
      "epoch": 4.9629139072847686,
      "grad_norm": 1.1951690874547671,
      "learning_rate": 3.496948186800217e-05,
      "loss": 0.7773,
      "step": 7494
    },
    {
      "epoch": 4.9635761589403975,
      "grad_norm": 1.0770505000102806,
      "learning_rate": 3.492665441702845e-05,
      "loss": 0.6602,
      "step": 7495
    },
    {
      "epoch": 4.964238410596026,
      "grad_norm": 1.2562016087542116,
      "learning_rate": 3.4883849751896355e-05,
      "loss": 0.7695,
      "step": 7496
    },
    {
      "epoch": 4.964900662251655,
      "grad_norm": 1.0886596565913125,
      "learning_rate": 3.484106788108185e-05,
      "loss": 0.7031,
      "step": 7497
    },
    {
      "epoch": 4.965562913907284,
      "grad_norm": 1.136453072228562,
      "learning_rate": 3.4798308813056156e-05,
      "loss": 0.6914,
      "step": 7498
    },
    {
      "epoch": 4.966225165562914,
      "grad_norm": 1.3035674731585984,
      "learning_rate": 3.4755572556286035e-05,
      "loss": 0.8828,
      "step": 7499
    },
    {
      "epoch": 4.966887417218543,
      "grad_norm": 1.396453685841964,
      "learning_rate": 3.471285911923377e-05,
      "loss": 0.9492,
      "step": 7500
    },
    {
      "epoch": 4.967549668874172,
      "grad_norm": 1.1879446344275402,
      "learning_rate": 3.467016851035716e-05,
      "loss": 0.9062,
      "step": 7501
    },
    {
      "epoch": 4.968211920529801,
      "grad_norm": 1.039337657449268,
      "learning_rate": 3.462750073810937e-05,
      "loss": 0.6406,
      "step": 7502
    },
    {
      "epoch": 4.968874172185431,
      "grad_norm": 1.141881459745817,
      "learning_rate": 3.458485581093916e-05,
      "loss": 0.7266,
      "step": 7503
    },
    {
      "epoch": 4.96953642384106,
      "grad_norm": 1.0440468679533492,
      "learning_rate": 3.4542233737290674e-05,
      "loss": 0.6602,
      "step": 7504
    },
    {
      "epoch": 4.970198675496689,
      "grad_norm": 1.4654282730806794,
      "learning_rate": 3.4499634525603565e-05,
      "loss": 1.0078,
      "step": 7505
    },
    {
      "epoch": 4.970860927152318,
      "grad_norm": 1.2545131609773548,
      "learning_rate": 3.4457058184313046e-05,
      "loss": 0.8711,
      "step": 7506
    },
    {
      "epoch": 4.9715231788079475,
      "grad_norm": 1.248092534823807,
      "learning_rate": 3.441450472184966e-05,
      "loss": 0.8906,
      "step": 7507
    },
    {
      "epoch": 4.972185430463576,
      "grad_norm": 1.1147056702422578,
      "learning_rate": 3.437197414663944e-05,
      "loss": 0.7188,
      "step": 7508
    },
    {
      "epoch": 4.972847682119205,
      "grad_norm": 1.1999466164983423,
      "learning_rate": 3.432946646710397e-05,
      "loss": 0.8359,
      "step": 7509
    },
    {
      "epoch": 4.973509933774834,
      "grad_norm": 1.2355026689556867,
      "learning_rate": 3.4286981691660305e-05,
      "loss": 0.8711,
      "step": 7510
    },
    {
      "epoch": 4.974172185430463,
      "grad_norm": 1.1408771446297343,
      "learning_rate": 3.424451982872082e-05,
      "loss": 0.7656,
      "step": 7511
    },
    {
      "epoch": 4.974834437086093,
      "grad_norm": 1.265672152514104,
      "learning_rate": 3.42020808866935e-05,
      "loss": 0.8789,
      "step": 7512
    },
    {
      "epoch": 4.975496688741722,
      "grad_norm": 1.1881263466349128,
      "learning_rate": 3.4159664873981794e-05,
      "loss": 0.832,
      "step": 7513
    },
    {
      "epoch": 4.976158940397351,
      "grad_norm": 1.1700901473541707,
      "learning_rate": 3.411727179898447e-05,
      "loss": 0.7617,
      "step": 7514
    },
    {
      "epoch": 4.97682119205298,
      "grad_norm": 1.2488511161550804,
      "learning_rate": 3.407490167009592e-05,
      "loss": 0.8594,
      "step": 7515
    },
    {
      "epoch": 4.977483443708609,
      "grad_norm": 1.168683204277143,
      "learning_rate": 3.4032554495705843e-05,
      "loss": 0.793,
      "step": 7516
    },
    {
      "epoch": 4.978145695364239,
      "grad_norm": 1.1572188069134803,
      "learning_rate": 3.399023028419949e-05,
      "loss": 0.8516,
      "step": 7517
    },
    {
      "epoch": 4.978807947019868,
      "grad_norm": 1.12968111219171,
      "learning_rate": 3.39479290439576e-05,
      "loss": 0.8047,
      "step": 7518
    },
    {
      "epoch": 4.979470198675497,
      "grad_norm": 1.1150208735749265,
      "learning_rate": 3.390565078335623e-05,
      "loss": 0.7109,
      "step": 7519
    },
    {
      "epoch": 4.9801324503311255,
      "grad_norm": 1.079977805472051,
      "learning_rate": 3.386339551076697e-05,
      "loss": 0.7734,
      "step": 7520
    },
    {
      "epoch": 4.980794701986755,
      "grad_norm": 1.1244656952360088,
      "learning_rate": 3.3821163234556846e-05,
      "loss": 0.7109,
      "step": 7521
    },
    {
      "epoch": 4.981456953642384,
      "grad_norm": 1.2696485504524433,
      "learning_rate": 3.37789539630884e-05,
      "loss": 0.8203,
      "step": 7522
    },
    {
      "epoch": 4.982119205298013,
      "grad_norm": 1.2319092806641767,
      "learning_rate": 3.373676770471945e-05,
      "loss": 0.832,
      "step": 7523
    },
    {
      "epoch": 4.982781456953642,
      "grad_norm": 1.075944280524066,
      "learning_rate": 3.3694604467803454e-05,
      "loss": 0.6836,
      "step": 7524
    },
    {
      "epoch": 4.983443708609272,
      "grad_norm": 1.2454103365015539,
      "learning_rate": 3.3652464260689126e-05,
      "loss": 0.9062,
      "step": 7525
    },
    {
      "epoch": 4.984105960264901,
      "grad_norm": 1.1768257560896835,
      "learning_rate": 3.361034709172076e-05,
      "loss": 0.7695,
      "step": 7526
    },
    {
      "epoch": 4.98476821192053,
      "grad_norm": 1.132109065523027,
      "learning_rate": 3.3568252969238074e-05,
      "loss": 0.7578,
      "step": 7527
    },
    {
      "epoch": 4.985430463576159,
      "grad_norm": 1.2659992806552987,
      "learning_rate": 3.352618190157614e-05,
      "loss": 0.7734,
      "step": 7528
    },
    {
      "epoch": 4.986092715231788,
      "grad_norm": 1.2595062399369894,
      "learning_rate": 3.348413389706543e-05,
      "loss": 0.7852,
      "step": 7529
    },
    {
      "epoch": 4.986754966887418,
      "grad_norm": 1.2776243586999054,
      "learning_rate": 3.344210896403212e-05,
      "loss": 0.8711,
      "step": 7530
    },
    {
      "epoch": 4.9874172185430465,
      "grad_norm": 1.1308286578787268,
      "learning_rate": 3.340010711079753e-05,
      "loss": 0.7891,
      "step": 7531
    },
    {
      "epoch": 4.9880794701986755,
      "grad_norm": 1.1807833677725073,
      "learning_rate": 3.335812834567844e-05,
      "loss": 0.7891,
      "step": 7532
    },
    {
      "epoch": 4.988741721854304,
      "grad_norm": 1.28548609694829,
      "learning_rate": 3.331617267698722e-05,
      "loss": 0.8594,
      "step": 7533
    },
    {
      "epoch": 4.989403973509933,
      "grad_norm": 1.1890087576490285,
      "learning_rate": 3.327424011303158e-05,
      "loss": 0.8398,
      "step": 7534
    },
    {
      "epoch": 4.990066225165563,
      "grad_norm": 1.302953743445622,
      "learning_rate": 3.323233066211457e-05,
      "loss": 0.8438,
      "step": 7535
    },
    {
      "epoch": 4.990728476821192,
      "grad_norm": 1.0552955828714154,
      "learning_rate": 3.319044433253482e-05,
      "loss": 0.6992,
      "step": 7536
    },
    {
      "epoch": 4.991390728476821,
      "grad_norm": 1.1417964748147067,
      "learning_rate": 3.3148581132586234e-05,
      "loss": 0.793,
      "step": 7537
    },
    {
      "epoch": 4.99205298013245,
      "grad_norm": 1.1727463109522698,
      "learning_rate": 3.310674107055824e-05,
      "loss": 0.7383,
      "step": 7538
    },
    {
      "epoch": 4.99271523178808,
      "grad_norm": 1.3291269885926662,
      "learning_rate": 3.30649241547357e-05,
      "loss": 0.8086,
      "step": 7539
    },
    {
      "epoch": 4.993377483443709,
      "grad_norm": 1.0818342967622132,
      "learning_rate": 3.302313039339879e-05,
      "loss": 0.6602,
      "step": 7540
    },
    {
      "epoch": 4.994039735099338,
      "grad_norm": 1.2417378260311274,
      "learning_rate": 3.298135979482312e-05,
      "loss": 0.8047,
      "step": 7541
    },
    {
      "epoch": 4.994701986754967,
      "grad_norm": 1.0489255761247729,
      "learning_rate": 3.293961236727978e-05,
      "loss": 0.6797,
      "step": 7542
    },
    {
      "epoch": 4.9953642384105965,
      "grad_norm": 1.1839504977215751,
      "learning_rate": 3.28978881190353e-05,
      "loss": 0.7891,
      "step": 7543
    },
    {
      "epoch": 4.9960264900662255,
      "grad_norm": 1.1474390051561356,
      "learning_rate": 3.285618705835144e-05,
      "loss": 0.7188,
      "step": 7544
    },
    {
      "epoch": 4.996688741721854,
      "grad_norm": 1.2849876109418392,
      "learning_rate": 3.2814509193485615e-05,
      "loss": 0.8125,
      "step": 7545
    },
    {
      "epoch": 4.997350993377483,
      "grad_norm": 1.5099332667838625,
      "learning_rate": 3.277285453269041e-05,
      "loss": 1.1406,
      "step": 7546
    },
    {
      "epoch": 4.998013245033112,
      "grad_norm": 1.286214208187156,
      "learning_rate": 3.2731223084213966e-05,
      "loss": 0.9297,
      "step": 7547
    },
    {
      "epoch": 4.998675496688742,
      "grad_norm": 1.1222759365816126,
      "learning_rate": 3.2689614856299845e-05,
      "loss": 0.7344,
      "step": 7548
    },
    {
      "epoch": 4.999337748344371,
      "grad_norm": 1.2871668091427075,
      "learning_rate": 3.264802985718689e-05,
      "loss": 0.8633,
      "step": 7549
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.183089364256464,
      "learning_rate": 3.260646809510935e-05,
      "loss": 0.7852,
      "step": 7550
    },
    {
      "epoch": 5.0,
      "eval_loss": 2.6689627170562744,
      "eval_runtime": 33.9038,
      "eval_samples_per_second": 9.969,
      "eval_steps_per_second": 9.969,
      "step": 7550
    },
    {
      "epoch": 5.000662251655629,
      "grad_norm": 0.6045593851255149,
      "learning_rate": 3.256492957829707e-05,
      "loss": 0.3105,
      "step": 7551
    },
    {
      "epoch": 5.001324503311258,
      "grad_norm": 0.614453358918784,
      "learning_rate": 3.2523414314975096e-05,
      "loss": 0.3379,
      "step": 7552
    },
    {
      "epoch": 5.001986754966888,
      "grad_norm": 0.5363776568778021,
      "learning_rate": 3.2481922313363855e-05,
      "loss": 0.3418,
      "step": 7553
    },
    {
      "epoch": 5.002649006622517,
      "grad_norm": 0.5436052580347087,
      "learning_rate": 3.2440453581679304e-05,
      "loss": 0.293,
      "step": 7554
    },
    {
      "epoch": 5.003311258278146,
      "grad_norm": 0.6263897891439537,
      "learning_rate": 3.239900812813277e-05,
      "loss": 0.3008,
      "step": 7555
    },
    {
      "epoch": 5.0039735099337745,
      "grad_norm": 0.6139806337730688,
      "learning_rate": 3.235758596093087e-05,
      "loss": 0.3281,
      "step": 7556
    },
    {
      "epoch": 5.004635761589404,
      "grad_norm": 0.6213473034125038,
      "learning_rate": 3.2316187088275644e-05,
      "loss": 0.3105,
      "step": 7557
    },
    {
      "epoch": 5.005298013245033,
      "grad_norm": 0.6081866789009767,
      "learning_rate": 3.2274811518364555e-05,
      "loss": 0.3184,
      "step": 7558
    },
    {
      "epoch": 5.005960264900662,
      "grad_norm": 0.6230614642586979,
      "learning_rate": 3.223345925939053e-05,
      "loss": 0.3398,
      "step": 7559
    },
    {
      "epoch": 5.006622516556291,
      "grad_norm": 0.5760426665791908,
      "learning_rate": 3.2192130319541656e-05,
      "loss": 0.2617,
      "step": 7560
    },
    {
      "epoch": 5.00728476821192,
      "grad_norm": 0.6335750509798314,
      "learning_rate": 3.215082470700166e-05,
      "loss": 0.2852,
      "step": 7561
    },
    {
      "epoch": 5.00794701986755,
      "grad_norm": 0.6597385082153511,
      "learning_rate": 3.210954242994941e-05,
      "loss": 0.291,
      "step": 7562
    },
    {
      "epoch": 5.008609271523179,
      "grad_norm": 0.6751536733643929,
      "learning_rate": 3.206828349655934e-05,
      "loss": 0.3203,
      "step": 7563
    },
    {
      "epoch": 5.009271523178808,
      "grad_norm": 0.7095708921880808,
      "learning_rate": 3.202704791500123e-05,
      "loss": 0.334,
      "step": 7564
    },
    {
      "epoch": 5.009933774834437,
      "grad_norm": 0.6748278147310006,
      "learning_rate": 3.1985835693440135e-05,
      "loss": 0.2773,
      "step": 7565
    },
    {
      "epoch": 5.010596026490067,
      "grad_norm": 0.7253035123265196,
      "learning_rate": 3.194464684003649e-05,
      "loss": 0.2969,
      "step": 7566
    },
    {
      "epoch": 5.011258278145696,
      "grad_norm": 0.6468843304344135,
      "learning_rate": 3.1903481362946284e-05,
      "loss": 0.2793,
      "step": 7567
    },
    {
      "epoch": 5.0119205298013245,
      "grad_norm": 0.6388316129245085,
      "learning_rate": 3.1862339270320734e-05,
      "loss": 0.2305,
      "step": 7568
    },
    {
      "epoch": 5.0125827814569535,
      "grad_norm": 0.736180253215124,
      "learning_rate": 3.182122057030636e-05,
      "loss": 0.3047,
      "step": 7569
    },
    {
      "epoch": 5.013245033112582,
      "grad_norm": 0.7005766682478414,
      "learning_rate": 3.178012527104518e-05,
      "loss": 0.2256,
      "step": 7570
    },
    {
      "epoch": 5.013907284768212,
      "grad_norm": 0.7391476320830732,
      "learning_rate": 3.1739053380674605e-05,
      "loss": 0.2891,
      "step": 7571
    },
    {
      "epoch": 5.014569536423841,
      "grad_norm": 0.6737616588863078,
      "learning_rate": 3.169800490732722e-05,
      "loss": 0.2559,
      "step": 7572
    },
    {
      "epoch": 5.01523178807947,
      "grad_norm": 0.7375107099437852,
      "learning_rate": 3.16569798591312e-05,
      "loss": 0.2812,
      "step": 7573
    },
    {
      "epoch": 5.015894039735099,
      "grad_norm": 0.5479396878081634,
      "learning_rate": 3.161597824420988e-05,
      "loss": 0.1992,
      "step": 7574
    },
    {
      "epoch": 5.016556291390729,
      "grad_norm": 0.6618811652801873,
      "learning_rate": 3.1575000070682096e-05,
      "loss": 0.2246,
      "step": 7575
    },
    {
      "epoch": 5.017218543046358,
      "grad_norm": 0.8143921681845607,
      "learning_rate": 3.153404534666204e-05,
      "loss": 0.3477,
      "step": 7576
    },
    {
      "epoch": 5.017880794701987,
      "grad_norm": 0.7975892096703138,
      "learning_rate": 3.149311408025917e-05,
      "loss": 0.291,
      "step": 7577
    },
    {
      "epoch": 5.018543046357616,
      "grad_norm": 0.7144677117162048,
      "learning_rate": 3.1452206279578315e-05,
      "loss": 0.2656,
      "step": 7578
    },
    {
      "epoch": 5.019205298013245,
      "grad_norm": 0.7442934532373326,
      "learning_rate": 3.141132195271973e-05,
      "loss": 0.3105,
      "step": 7579
    },
    {
      "epoch": 5.0198675496688745,
      "grad_norm": 0.6622531023736544,
      "learning_rate": 3.1370461107779e-05,
      "loss": 0.2578,
      "step": 7580
    },
    {
      "epoch": 5.020529801324503,
      "grad_norm": 0.6697685183960234,
      "learning_rate": 3.132962375284696e-05,
      "loss": 0.2695,
      "step": 7581
    },
    {
      "epoch": 5.021192052980132,
      "grad_norm": 0.6214571286122939,
      "learning_rate": 3.1288809896009994e-05,
      "loss": 0.2295,
      "step": 7582
    },
    {
      "epoch": 5.021854304635761,
      "grad_norm": 0.8289040032620009,
      "learning_rate": 3.1248019545349595e-05,
      "loss": 0.3223,
      "step": 7583
    },
    {
      "epoch": 5.022516556291391,
      "grad_norm": 0.5874854051799077,
      "learning_rate": 3.1207252708942776e-05,
      "loss": 0.2471,
      "step": 7584
    },
    {
      "epoch": 5.02317880794702,
      "grad_norm": 0.688340521126013,
      "learning_rate": 3.1166509394861876e-05,
      "loss": 0.2598,
      "step": 7585
    },
    {
      "epoch": 5.023841059602649,
      "grad_norm": 0.923833078630724,
      "learning_rate": 3.11257896111745e-05,
      "loss": 0.3105,
      "step": 7586
    },
    {
      "epoch": 5.024503311258278,
      "grad_norm": 0.6914107273557181,
      "learning_rate": 3.108509336594355e-05,
      "loss": 0.2715,
      "step": 7587
    },
    {
      "epoch": 5.025165562913907,
      "grad_norm": 0.767632264094937,
      "learning_rate": 3.10444206672275e-05,
      "loss": 0.3203,
      "step": 7588
    },
    {
      "epoch": 5.025827814569537,
      "grad_norm": 0.8290049959724409,
      "learning_rate": 3.1003771523079955e-05,
      "loss": 0.3125,
      "step": 7589
    },
    {
      "epoch": 5.026490066225166,
      "grad_norm": 0.7062920284406934,
      "learning_rate": 3.0963145941549835e-05,
      "loss": 0.3125,
      "step": 7590
    },
    {
      "epoch": 5.027152317880795,
      "grad_norm": 0.7124114501782699,
      "learning_rate": 3.092254393068155e-05,
      "loss": 0.2969,
      "step": 7591
    },
    {
      "epoch": 5.027814569536424,
      "grad_norm": 0.7163962835330602,
      "learning_rate": 3.0881965498514774e-05,
      "loss": 0.2676,
      "step": 7592
    },
    {
      "epoch": 5.028476821192053,
      "grad_norm": 0.8675738561091091,
      "learning_rate": 3.084141065308444e-05,
      "loss": 0.3242,
      "step": 7593
    },
    {
      "epoch": 5.029139072847682,
      "grad_norm": 0.7424484951438328,
      "learning_rate": 3.0800879402420936e-05,
      "loss": 0.3633,
      "step": 7594
    },
    {
      "epoch": 5.029801324503311,
      "grad_norm": 0.6846913241868403,
      "learning_rate": 3.076037175454985e-05,
      "loss": 0.2432,
      "step": 7595
    },
    {
      "epoch": 5.03046357615894,
      "grad_norm": 0.6756363459499994,
      "learning_rate": 3.071988771749219e-05,
      "loss": 0.2402,
      "step": 7596
    },
    {
      "epoch": 5.031125827814569,
      "grad_norm": 0.5954603225645252,
      "learning_rate": 3.067942729926431e-05,
      "loss": 0.2109,
      "step": 7597
    },
    {
      "epoch": 5.031788079470199,
      "grad_norm": 0.6702748916283935,
      "learning_rate": 3.063899050787779e-05,
      "loss": 0.2852,
      "step": 7598
    },
    {
      "epoch": 5.032450331125828,
      "grad_norm": 0.8713949748005583,
      "learning_rate": 3.0598577351339535e-05,
      "loss": 0.3379,
      "step": 7599
    },
    {
      "epoch": 5.033112582781457,
      "grad_norm": 0.6034982603757743,
      "learning_rate": 3.055818783765185e-05,
      "loss": 0.2432,
      "step": 7600
    },
    {
      "epoch": 5.033774834437086,
      "grad_norm": 0.7060689594445229,
      "learning_rate": 3.051782197481238e-05,
      "loss": 0.2676,
      "step": 7601
    },
    {
      "epoch": 5.034437086092716,
      "grad_norm": 0.6941983929403234,
      "learning_rate": 3.0477479770813945e-05,
      "loss": 0.2598,
      "step": 7602
    },
    {
      "epoch": 5.035099337748345,
      "grad_norm": 0.639792293258147,
      "learning_rate": 3.0437161233644823e-05,
      "loss": 0.208,
      "step": 7603
    },
    {
      "epoch": 5.035761589403974,
      "grad_norm": 0.7198302677323313,
      "learning_rate": 3.039686637128848e-05,
      "loss": 0.2773,
      "step": 7604
    },
    {
      "epoch": 5.0364238410596025,
      "grad_norm": 0.6641335071807877,
      "learning_rate": 3.0356595191723796e-05,
      "loss": 0.2441,
      "step": 7605
    },
    {
      "epoch": 5.0370860927152314,
      "grad_norm": 0.8385299960824646,
      "learning_rate": 3.031634770292498e-05,
      "loss": 0.3301,
      "step": 7606
    },
    {
      "epoch": 5.037748344370861,
      "grad_norm": 1.0519939954638169,
      "learning_rate": 3.0276123912861406e-05,
      "loss": 0.4238,
      "step": 7607
    },
    {
      "epoch": 5.03841059602649,
      "grad_norm": 0.7752802139520284,
      "learning_rate": 3.0235923829497897e-05,
      "loss": 0.2617,
      "step": 7608
    },
    {
      "epoch": 5.039072847682119,
      "grad_norm": 0.6767035505422887,
      "learning_rate": 3.0195747460794544e-05,
      "loss": 0.2412,
      "step": 7609
    },
    {
      "epoch": 5.039735099337748,
      "grad_norm": 0.7487400542011913,
      "learning_rate": 3.0155594814706714e-05,
      "loss": 0.3281,
      "step": 7610
    },
    {
      "epoch": 5.040397350993377,
      "grad_norm": 0.7143849430055839,
      "learning_rate": 3.0115465899185044e-05,
      "loss": 0.2871,
      "step": 7611
    },
    {
      "epoch": 5.041059602649007,
      "grad_norm": 0.8067559916070459,
      "learning_rate": 3.0075360722175545e-05,
      "loss": 0.2832,
      "step": 7612
    },
    {
      "epoch": 5.041721854304636,
      "grad_norm": 0.7615298702149778,
      "learning_rate": 3.0035279291619552e-05,
      "loss": 0.2754,
      "step": 7613
    },
    {
      "epoch": 5.042384105960265,
      "grad_norm": 0.6388843296769425,
      "learning_rate": 2.999522161545363e-05,
      "loss": 0.2217,
      "step": 7614
    },
    {
      "epoch": 5.043046357615894,
      "grad_norm": 0.7072617079719921,
      "learning_rate": 2.9955187701609583e-05,
      "loss": 0.2676,
      "step": 7615
    },
    {
      "epoch": 5.0437086092715235,
      "grad_norm": 0.740433366157003,
      "learning_rate": 2.991517755801464e-05,
      "loss": 0.2949,
      "step": 7616
    },
    {
      "epoch": 5.0443708609271525,
      "grad_norm": 0.6572320440827792,
      "learning_rate": 2.9875191192591315e-05,
      "loss": 0.2559,
      "step": 7617
    },
    {
      "epoch": 5.045033112582781,
      "grad_norm": 0.8230001013759648,
      "learning_rate": 2.9835228613257267e-05,
      "loss": 0.2988,
      "step": 7618
    },
    {
      "epoch": 5.04569536423841,
      "grad_norm": 0.67089813454419,
      "learning_rate": 2.9795289827925655e-05,
      "loss": 0.2559,
      "step": 7619
    },
    {
      "epoch": 5.046357615894039,
      "grad_norm": 0.7627826459506433,
      "learning_rate": 2.9755374844504705e-05,
      "loss": 0.3145,
      "step": 7620
    },
    {
      "epoch": 5.047019867549669,
      "grad_norm": 0.9513975737147948,
      "learning_rate": 2.9715483670898105e-05,
      "loss": 0.3164,
      "step": 7621
    },
    {
      "epoch": 5.047682119205298,
      "grad_norm": 0.6849829838889234,
      "learning_rate": 2.9675616315004792e-05,
      "loss": 0.2432,
      "step": 7622
    },
    {
      "epoch": 5.048344370860927,
      "grad_norm": 0.6888569081378938,
      "learning_rate": 2.963577278471893e-05,
      "loss": 0.2969,
      "step": 7623
    },
    {
      "epoch": 5.049006622516556,
      "grad_norm": 0.8288091735587798,
      "learning_rate": 2.9595953087929898e-05,
      "loss": 0.2988,
      "step": 7624
    },
    {
      "epoch": 5.049668874172186,
      "grad_norm": 0.7930344652639415,
      "learning_rate": 2.9556157232522615e-05,
      "loss": 0.3086,
      "step": 7625
    },
    {
      "epoch": 5.050331125827815,
      "grad_norm": 0.6548584610603714,
      "learning_rate": 2.9516385226377066e-05,
      "loss": 0.2578,
      "step": 7626
    },
    {
      "epoch": 5.050993377483444,
      "grad_norm": 0.7796059686270767,
      "learning_rate": 2.947663707736848e-05,
      "loss": 0.2412,
      "step": 7627
    },
    {
      "epoch": 5.051655629139073,
      "grad_norm": 0.7116051506582936,
      "learning_rate": 2.9436912793367517e-05,
      "loss": 0.2441,
      "step": 7628
    },
    {
      "epoch": 5.052317880794702,
      "grad_norm": 0.8213692371199846,
      "learning_rate": 2.9397212382240067e-05,
      "loss": 0.3496,
      "step": 7629
    },
    {
      "epoch": 5.052980132450331,
      "grad_norm": 0.7801315946656635,
      "learning_rate": 2.9357535851847202e-05,
      "loss": 0.3047,
      "step": 7630
    },
    {
      "epoch": 5.05364238410596,
      "grad_norm": 0.7482613323816574,
      "learning_rate": 2.9317883210045397e-05,
      "loss": 0.2695,
      "step": 7631
    },
    {
      "epoch": 5.054304635761589,
      "grad_norm": 0.67907499185157,
      "learning_rate": 2.9278254464686247e-05,
      "loss": 0.2314,
      "step": 7632
    },
    {
      "epoch": 5.054966887417218,
      "grad_norm": 0.7758730145353828,
      "learning_rate": 2.9238649623616744e-05,
      "loss": 0.2832,
      "step": 7633
    },
    {
      "epoch": 5.055629139072848,
      "grad_norm": 0.6587963006206099,
      "learning_rate": 2.9199068694679136e-05,
      "loss": 0.2207,
      "step": 7634
    },
    {
      "epoch": 5.056291390728477,
      "grad_norm": 0.6544764391843848,
      "learning_rate": 2.915951168571086e-05,
      "loss": 0.2451,
      "step": 7635
    },
    {
      "epoch": 5.056953642384106,
      "grad_norm": 0.6520405499517448,
      "learning_rate": 2.9119978604544632e-05,
      "loss": 0.2383,
      "step": 7636
    },
    {
      "epoch": 5.057615894039735,
      "grad_norm": 0.6313472675535934,
      "learning_rate": 2.9080469459008478e-05,
      "loss": 0.2334,
      "step": 7637
    },
    {
      "epoch": 5.058278145695364,
      "grad_norm": 0.6895802592774316,
      "learning_rate": 2.9040984256925708e-05,
      "loss": 0.2441,
      "step": 7638
    },
    {
      "epoch": 5.058940397350994,
      "grad_norm": 0.8088756672108751,
      "learning_rate": 2.9001523006114775e-05,
      "loss": 0.3086,
      "step": 7639
    },
    {
      "epoch": 5.059602649006623,
      "grad_norm": 0.7220205133085944,
      "learning_rate": 2.8962085714389506e-05,
      "loss": 0.2197,
      "step": 7640
    },
    {
      "epoch": 5.0602649006622515,
      "grad_norm": 0.820990773722928,
      "learning_rate": 2.8922672389558903e-05,
      "loss": 0.2656,
      "step": 7641
    },
    {
      "epoch": 5.0609271523178805,
      "grad_norm": 0.786487377711401,
      "learning_rate": 2.888328303942726e-05,
      "loss": 0.2949,
      "step": 7642
    },
    {
      "epoch": 5.06158940397351,
      "grad_norm": 0.7922244188622611,
      "learning_rate": 2.8843917671794177e-05,
      "loss": 0.2832,
      "step": 7643
    },
    {
      "epoch": 5.062251655629139,
      "grad_norm": 0.835564970658743,
      "learning_rate": 2.8804576294454405e-05,
      "loss": 0.3223,
      "step": 7644
    },
    {
      "epoch": 5.062913907284768,
      "grad_norm": 0.7896123519572305,
      "learning_rate": 2.876525891519791e-05,
      "loss": 0.2812,
      "step": 7645
    },
    {
      "epoch": 5.063576158940397,
      "grad_norm": 0.7558894164031373,
      "learning_rate": 2.8725965541810137e-05,
      "loss": 0.2734,
      "step": 7646
    },
    {
      "epoch": 5.064238410596026,
      "grad_norm": 0.7526194985669108,
      "learning_rate": 2.868669618207155e-05,
      "loss": 0.2793,
      "step": 7647
    },
    {
      "epoch": 5.064900662251656,
      "grad_norm": 0.6886377270516222,
      "learning_rate": 2.8647450843757897e-05,
      "loss": 0.2617,
      "step": 7648
    },
    {
      "epoch": 5.065562913907285,
      "grad_norm": 0.8592849933825653,
      "learning_rate": 2.8608229534640225e-05,
      "loss": 0.3281,
      "step": 7649
    },
    {
      "epoch": 5.066225165562914,
      "grad_norm": 0.8247954472251052,
      "learning_rate": 2.856903226248486e-05,
      "loss": 0.2695,
      "step": 7650
    },
    {
      "epoch": 5.066887417218543,
      "grad_norm": 0.6603437357196821,
      "learning_rate": 2.8529859035053236e-05,
      "loss": 0.2539,
      "step": 7651
    },
    {
      "epoch": 5.067549668874173,
      "grad_norm": 0.7145933704465248,
      "learning_rate": 2.8490709860102174e-05,
      "loss": 0.2598,
      "step": 7652
    },
    {
      "epoch": 5.0682119205298015,
      "grad_norm": 0.805700932033964,
      "learning_rate": 2.8451584745383566e-05,
      "loss": 0.2773,
      "step": 7653
    },
    {
      "epoch": 5.0688741721854305,
      "grad_norm": 0.7994612318899332,
      "learning_rate": 2.8412483698644684e-05,
      "loss": 0.2734,
      "step": 7654
    },
    {
      "epoch": 5.069536423841059,
      "grad_norm": 0.6800294021719491,
      "learning_rate": 2.8373406727628036e-05,
      "loss": 0.2617,
      "step": 7655
    },
    {
      "epoch": 5.070198675496688,
      "grad_norm": 0.718540195425804,
      "learning_rate": 2.833435384007126e-05,
      "loss": 0.2432,
      "step": 7656
    },
    {
      "epoch": 5.070860927152318,
      "grad_norm": 0.7048242233506135,
      "learning_rate": 2.829532504370723e-05,
      "loss": 0.2734,
      "step": 7657
    },
    {
      "epoch": 5.071523178807947,
      "grad_norm": 0.829166561652657,
      "learning_rate": 2.825632034626414e-05,
      "loss": 0.3145,
      "step": 7658
    },
    {
      "epoch": 5.072185430463576,
      "grad_norm": 0.8044672132338115,
      "learning_rate": 2.8217339755465394e-05,
      "loss": 0.2812,
      "step": 7659
    },
    {
      "epoch": 5.072847682119205,
      "grad_norm": 0.8604951843726184,
      "learning_rate": 2.817838327902956e-05,
      "loss": 0.3281,
      "step": 7660
    },
    {
      "epoch": 5.073509933774835,
      "grad_norm": 0.7026263856387941,
      "learning_rate": 2.8139450924670443e-05,
      "loss": 0.2676,
      "step": 7661
    },
    {
      "epoch": 5.074172185430464,
      "grad_norm": 0.8829538387056132,
      "learning_rate": 2.8100542700097195e-05,
      "loss": 0.3262,
      "step": 7662
    },
    {
      "epoch": 5.074834437086093,
      "grad_norm": 0.8277355279150733,
      "learning_rate": 2.806165861301397e-05,
      "loss": 0.3398,
      "step": 7663
    },
    {
      "epoch": 5.075496688741722,
      "grad_norm": 0.7517399962052583,
      "learning_rate": 2.8022798671120362e-05,
      "loss": 0.2871,
      "step": 7664
    },
    {
      "epoch": 5.076158940397351,
      "grad_norm": 0.6924743487511614,
      "learning_rate": 2.7983962882110995e-05,
      "loss": 0.2383,
      "step": 7665
    },
    {
      "epoch": 5.07682119205298,
      "grad_norm": 0.7332136228225022,
      "learning_rate": 2.7945151253675885e-05,
      "loss": 0.2852,
      "step": 7666
    },
    {
      "epoch": 5.077483443708609,
      "grad_norm": 0.7569197505053394,
      "learning_rate": 2.7906363793500102e-05,
      "loss": 0.2734,
      "step": 7667
    },
    {
      "epoch": 5.078145695364238,
      "grad_norm": 0.7259861765065343,
      "learning_rate": 2.7867600509264086e-05,
      "loss": 0.2852,
      "step": 7668
    },
    {
      "epoch": 5.078807947019867,
      "grad_norm": 0.9074679048279021,
      "learning_rate": 2.782886140864333e-05,
      "loss": 0.3418,
      "step": 7669
    },
    {
      "epoch": 5.079470198675497,
      "grad_norm": 0.7312402952255912,
      "learning_rate": 2.779014649930867e-05,
      "loss": 0.2539,
      "step": 7670
    },
    {
      "epoch": 5.080132450331126,
      "grad_norm": 0.635122803365682,
      "learning_rate": 2.7751455788926132e-05,
      "loss": 0.2236,
      "step": 7671
    },
    {
      "epoch": 5.080794701986755,
      "grad_norm": 0.7524611601496549,
      "learning_rate": 2.7712789285156874e-05,
      "loss": 0.2656,
      "step": 7672
    },
    {
      "epoch": 5.081456953642384,
      "grad_norm": 0.6782038970456145,
      "learning_rate": 2.7674146995657288e-05,
      "loss": 0.2207,
      "step": 7673
    },
    {
      "epoch": 5.082119205298013,
      "grad_norm": 0.6910893158617771,
      "learning_rate": 2.7635528928079004e-05,
      "loss": 0.2334,
      "step": 7674
    },
    {
      "epoch": 5.082781456953643,
      "grad_norm": 0.7919554516432932,
      "learning_rate": 2.759693509006889e-05,
      "loss": 0.2773,
      "step": 7675
    },
    {
      "epoch": 5.083443708609272,
      "grad_norm": 0.7593597600474896,
      "learning_rate": 2.7558365489268912e-05,
      "loss": 0.249,
      "step": 7676
    },
    {
      "epoch": 5.084105960264901,
      "grad_norm": 0.660311192066355,
      "learning_rate": 2.7519820133316333e-05,
      "loss": 0.2236,
      "step": 7677
    },
    {
      "epoch": 5.0847682119205295,
      "grad_norm": 0.8387875744629516,
      "learning_rate": 2.748129902984353e-05,
      "loss": 0.2812,
      "step": 7678
    },
    {
      "epoch": 5.085430463576159,
      "grad_norm": 0.9545842267535178,
      "learning_rate": 2.7442802186478145e-05,
      "loss": 0.2852,
      "step": 7679
    },
    {
      "epoch": 5.086092715231788,
      "grad_norm": 0.8173956244935504,
      "learning_rate": 2.7404329610843055e-05,
      "loss": 0.3477,
      "step": 7680
    },
    {
      "epoch": 5.086754966887417,
      "grad_norm": 0.6770257090971682,
      "learning_rate": 2.7365881310556197e-05,
      "loss": 0.2559,
      "step": 7681
    },
    {
      "epoch": 5.087417218543046,
      "grad_norm": 0.7069628119785532,
      "learning_rate": 2.7327457293230732e-05,
      "loss": 0.2773,
      "step": 7682
    },
    {
      "epoch": 5.088079470198675,
      "grad_norm": 0.7532216163228226,
      "learning_rate": 2.728905756647517e-05,
      "loss": 0.2285,
      "step": 7683
    },
    {
      "epoch": 5.088741721854305,
      "grad_norm": 0.8830951638311905,
      "learning_rate": 2.7250682137893065e-05,
      "loss": 0.332,
      "step": 7684
    },
    {
      "epoch": 5.089403973509934,
      "grad_norm": 0.7151510775475609,
      "learning_rate": 2.721233101508312e-05,
      "loss": 0.2695,
      "step": 7685
    },
    {
      "epoch": 5.090066225165563,
      "grad_norm": 0.7210162369513717,
      "learning_rate": 2.7174004205639364e-05,
      "loss": 0.2227,
      "step": 7686
    },
    {
      "epoch": 5.090728476821192,
      "grad_norm": 0.7196282053692017,
      "learning_rate": 2.713570171715096e-05,
      "loss": 0.2598,
      "step": 7687
    },
    {
      "epoch": 5.091390728476822,
      "grad_norm": 0.7520314610407249,
      "learning_rate": 2.7097423557202168e-05,
      "loss": 0.2734,
      "step": 7688
    },
    {
      "epoch": 5.092052980132451,
      "grad_norm": 0.789713228110885,
      "learning_rate": 2.70591697333726e-05,
      "loss": 0.2734,
      "step": 7689
    },
    {
      "epoch": 5.0927152317880795,
      "grad_norm": 0.646033166661115,
      "learning_rate": 2.702094025323684e-05,
      "loss": 0.2373,
      "step": 7690
    },
    {
      "epoch": 5.093377483443708,
      "grad_norm": 0.7291339596802344,
      "learning_rate": 2.6982735124364817e-05,
      "loss": 0.2559,
      "step": 7691
    },
    {
      "epoch": 5.094039735099337,
      "grad_norm": 0.6612067237021902,
      "learning_rate": 2.6944554354321634e-05,
      "loss": 0.2441,
      "step": 7692
    },
    {
      "epoch": 5.094701986754967,
      "grad_norm": 0.7376475337643713,
      "learning_rate": 2.6906397950667475e-05,
      "loss": 0.2617,
      "step": 7693
    },
    {
      "epoch": 5.095364238410596,
      "grad_norm": 0.7146897106493991,
      "learning_rate": 2.6868265920957693e-05,
      "loss": 0.2578,
      "step": 7694
    },
    {
      "epoch": 5.096026490066225,
      "grad_norm": 0.6958686201005283,
      "learning_rate": 2.683015827274292e-05,
      "loss": 0.2441,
      "step": 7695
    },
    {
      "epoch": 5.096688741721854,
      "grad_norm": 0.7579195655202206,
      "learning_rate": 2.6792075013568948e-05,
      "loss": 0.2559,
      "step": 7696
    },
    {
      "epoch": 5.097350993377484,
      "grad_norm": 0.7864199275594701,
      "learning_rate": 2.6754016150976615e-05,
      "loss": 0.3086,
      "step": 7697
    },
    {
      "epoch": 5.098013245033113,
      "grad_norm": 0.6233627457073797,
      "learning_rate": 2.671598169250209e-05,
      "loss": 0.2285,
      "step": 7698
    },
    {
      "epoch": 5.098675496688742,
      "grad_norm": 0.6326262813221883,
      "learning_rate": 2.667797164567657e-05,
      "loss": 0.2119,
      "step": 7699
    },
    {
      "epoch": 5.099337748344371,
      "grad_norm": 0.7902654980104751,
      "learning_rate": 2.66399860180265e-05,
      "loss": 0.2598,
      "step": 7700
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.7828906035845621,
      "learning_rate": 2.660202481707351e-05,
      "loss": 0.2578,
      "step": 7701
    },
    {
      "epoch": 5.1006622516556295,
      "grad_norm": 0.6938564471375939,
      "learning_rate": 2.656408805033427e-05,
      "loss": 0.2354,
      "step": 7702
    },
    {
      "epoch": 5.101324503311258,
      "grad_norm": 0.861259806632186,
      "learning_rate": 2.652617572532076e-05,
      "loss": 0.3574,
      "step": 7703
    },
    {
      "epoch": 5.101986754966887,
      "grad_norm": 0.8101499906582765,
      "learning_rate": 2.648828784954007e-05,
      "loss": 0.2871,
      "step": 7704
    },
    {
      "epoch": 5.102649006622516,
      "grad_norm": 0.7571199321679258,
      "learning_rate": 2.6450424430494398e-05,
      "loss": 0.2891,
      "step": 7705
    },
    {
      "epoch": 5.103311258278145,
      "grad_norm": 0.7664986801151913,
      "learning_rate": 2.6412585475681104e-05,
      "loss": 0.2871,
      "step": 7706
    },
    {
      "epoch": 5.103973509933775,
      "grad_norm": 0.6777814653379343,
      "learning_rate": 2.6374770992592752e-05,
      "loss": 0.209,
      "step": 7707
    },
    {
      "epoch": 5.104635761589404,
      "grad_norm": 0.661350490812791,
      "learning_rate": 2.633698098871711e-05,
      "loss": 0.2275,
      "step": 7708
    },
    {
      "epoch": 5.105298013245033,
      "grad_norm": 0.8431602069568898,
      "learning_rate": 2.6299215471536955e-05,
      "loss": 0.3125,
      "step": 7709
    },
    {
      "epoch": 5.105960264900662,
      "grad_norm": 0.7585370379359879,
      "learning_rate": 2.6261474448530358e-05,
      "loss": 0.2578,
      "step": 7710
    },
    {
      "epoch": 5.106622516556292,
      "grad_norm": 0.6627061756432032,
      "learning_rate": 2.6223757927170386e-05,
      "loss": 0.2158,
      "step": 7711
    },
    {
      "epoch": 5.107284768211921,
      "grad_norm": 0.7995715316480778,
      "learning_rate": 2.6186065914925387e-05,
      "loss": 0.3242,
      "step": 7712
    },
    {
      "epoch": 5.10794701986755,
      "grad_norm": 0.6875009407298466,
      "learning_rate": 2.614839841925887e-05,
      "loss": 0.2188,
      "step": 7713
    },
    {
      "epoch": 5.108609271523179,
      "grad_norm": 0.9641227108671551,
      "learning_rate": 2.611075544762937e-05,
      "loss": 0.3105,
      "step": 7714
    },
    {
      "epoch": 5.109271523178808,
      "grad_norm": 0.7429097109738173,
      "learning_rate": 2.6073137007490564e-05,
      "loss": 0.2852,
      "step": 7715
    },
    {
      "epoch": 5.109933774834437,
      "grad_norm": 0.7058542894901473,
      "learning_rate": 2.6035543106291484e-05,
      "loss": 0.252,
      "step": 7716
    },
    {
      "epoch": 5.110596026490066,
      "grad_norm": 0.684963279270005,
      "learning_rate": 2.5997973751476064e-05,
      "loss": 0.248,
      "step": 7717
    },
    {
      "epoch": 5.111258278145695,
      "grad_norm": 0.849185884310407,
      "learning_rate": 2.5960428950483452e-05,
      "loss": 0.2832,
      "step": 7718
    },
    {
      "epoch": 5.111920529801324,
      "grad_norm": 0.8564624585664711,
      "learning_rate": 2.592290871074797e-05,
      "loss": 0.3086,
      "step": 7719
    },
    {
      "epoch": 5.112582781456954,
      "grad_norm": 0.7306409626867655,
      "learning_rate": 2.5885413039699103e-05,
      "loss": 0.2852,
      "step": 7720
    },
    {
      "epoch": 5.113245033112583,
      "grad_norm": 0.7584688151781532,
      "learning_rate": 2.5847941944761334e-05,
      "loss": 0.2402,
      "step": 7721
    },
    {
      "epoch": 5.113907284768212,
      "grad_norm": 0.9400954173095813,
      "learning_rate": 2.5810495433354462e-05,
      "loss": 0.3379,
      "step": 7722
    },
    {
      "epoch": 5.114569536423841,
      "grad_norm": 1.020766213160279,
      "learning_rate": 2.577307351289325e-05,
      "loss": 0.3262,
      "step": 7723
    },
    {
      "epoch": 5.11523178807947,
      "grad_norm": 0.6781216659533541,
      "learning_rate": 2.5735676190787723e-05,
      "loss": 0.2539,
      "step": 7724
    },
    {
      "epoch": 5.1158940397351,
      "grad_norm": 0.7001404215064257,
      "learning_rate": 2.5698303474442926e-05,
      "loss": 0.2598,
      "step": 7725
    },
    {
      "epoch": 5.1165562913907285,
      "grad_norm": 0.7181277288389928,
      "learning_rate": 2.566095537125914e-05,
      "loss": 0.2422,
      "step": 7726
    },
    {
      "epoch": 5.1172185430463575,
      "grad_norm": 0.8244274359938574,
      "learning_rate": 2.562363188863166e-05,
      "loss": 0.3145,
      "step": 7727
    },
    {
      "epoch": 5.117880794701986,
      "grad_norm": 0.8478197279295507,
      "learning_rate": 2.5586333033950984e-05,
      "loss": 0.3066,
      "step": 7728
    },
    {
      "epoch": 5.118543046357616,
      "grad_norm": 0.8727063675105848,
      "learning_rate": 2.554905881460278e-05,
      "loss": 0.2871,
      "step": 7729
    },
    {
      "epoch": 5.119205298013245,
      "grad_norm": 0.6841203186875177,
      "learning_rate": 2.5511809237967696e-05,
      "loss": 0.2217,
      "step": 7730
    },
    {
      "epoch": 5.119867549668874,
      "grad_norm": 0.7770491399861332,
      "learning_rate": 2.5474584311421564e-05,
      "loss": 0.2695,
      "step": 7731
    },
    {
      "epoch": 5.120529801324503,
      "grad_norm": 0.7714153913618808,
      "learning_rate": 2.543738404233539e-05,
      "loss": 0.2676,
      "step": 7732
    },
    {
      "epoch": 5.121192052980132,
      "grad_norm": 0.751606982172731,
      "learning_rate": 2.540020843807525e-05,
      "loss": 0.2871,
      "step": 7733
    },
    {
      "epoch": 5.121854304635762,
      "grad_norm": 0.842153988680616,
      "learning_rate": 2.5363057506002314e-05,
      "loss": 0.3008,
      "step": 7734
    },
    {
      "epoch": 5.122516556291391,
      "grad_norm": 0.7261604214187055,
      "learning_rate": 2.5325931253472925e-05,
      "loss": 0.291,
      "step": 7735
    },
    {
      "epoch": 5.12317880794702,
      "grad_norm": 0.8376212128905876,
      "learning_rate": 2.528882968783846e-05,
      "loss": 0.3477,
      "step": 7736
    },
    {
      "epoch": 5.123841059602649,
      "grad_norm": 0.6914180929555219,
      "learning_rate": 2.5251752816445503e-05,
      "loss": 0.2188,
      "step": 7737
    },
    {
      "epoch": 5.1245033112582785,
      "grad_norm": 0.548928762365887,
      "learning_rate": 2.5214700646635687e-05,
      "loss": 0.1641,
      "step": 7738
    },
    {
      "epoch": 5.1251655629139075,
      "grad_norm": 0.9190817917272078,
      "learning_rate": 2.517767318574575e-05,
      "loss": 0.3496,
      "step": 7739
    },
    {
      "epoch": 5.125827814569536,
      "grad_norm": 0.7729911563879301,
      "learning_rate": 2.514067044110754e-05,
      "loss": 0.2637,
      "step": 7740
    },
    {
      "epoch": 5.126490066225165,
      "grad_norm": 0.7391897330933267,
      "learning_rate": 2.51036924200481e-05,
      "loss": 0.2178,
      "step": 7741
    },
    {
      "epoch": 5.127152317880794,
      "grad_norm": 0.7634435524158038,
      "learning_rate": 2.506673912988945e-05,
      "loss": 0.2559,
      "step": 7742
    },
    {
      "epoch": 5.127814569536424,
      "grad_norm": 0.7699337993007899,
      "learning_rate": 2.5029810577948726e-05,
      "loss": 0.2988,
      "step": 7743
    },
    {
      "epoch": 5.128476821192053,
      "grad_norm": 0.7476224687351686,
      "learning_rate": 2.4992906771538245e-05,
      "loss": 0.2461,
      "step": 7744
    },
    {
      "epoch": 5.129139072847682,
      "grad_norm": 0.678022448321499,
      "learning_rate": 2.495602771796541e-05,
      "loss": 0.2002,
      "step": 7745
    },
    {
      "epoch": 5.129801324503311,
      "grad_norm": 0.7811626858174175,
      "learning_rate": 2.4919173424532645e-05,
      "loss": 0.2676,
      "step": 7746
    },
    {
      "epoch": 5.130463576158941,
      "grad_norm": 0.7871573091954337,
      "learning_rate": 2.4882343898537566e-05,
      "loss": 0.2617,
      "step": 7747
    },
    {
      "epoch": 5.13112582781457,
      "grad_norm": 0.6987962050969169,
      "learning_rate": 2.4845539147272803e-05,
      "loss": 0.2344,
      "step": 7748
    },
    {
      "epoch": 5.131788079470199,
      "grad_norm": 0.6829468511002302,
      "learning_rate": 2.480875917802613e-05,
      "loss": 0.2178,
      "step": 7749
    },
    {
      "epoch": 5.132450331125828,
      "grad_norm": 0.6362573019790945,
      "learning_rate": 2.4772003998080447e-05,
      "loss": 0.2109,
      "step": 7750
    },
    {
      "epoch": 5.1331125827814565,
      "grad_norm": 0.6991382754834095,
      "learning_rate": 2.4735273614713673e-05,
      "loss": 0.2285,
      "step": 7751
    },
    {
      "epoch": 5.133774834437086,
      "grad_norm": 0.6981246927431022,
      "learning_rate": 2.469856803519879e-05,
      "loss": 0.2373,
      "step": 7752
    },
    {
      "epoch": 5.134437086092715,
      "grad_norm": 0.7669323342602807,
      "learning_rate": 2.466188726680398e-05,
      "loss": 0.2969,
      "step": 7753
    },
    {
      "epoch": 5.135099337748344,
      "grad_norm": 0.7018298021103166,
      "learning_rate": 2.462523131679247e-05,
      "loss": 0.2617,
      "step": 7754
    },
    {
      "epoch": 5.135761589403973,
      "grad_norm": 0.74069433283855,
      "learning_rate": 2.4588600192422504e-05,
      "loss": 0.2891,
      "step": 7755
    },
    {
      "epoch": 5.136423841059603,
      "grad_norm": 0.8005129413726052,
      "learning_rate": 2.455199390094749e-05,
      "loss": 0.2695,
      "step": 7756
    },
    {
      "epoch": 5.137086092715232,
      "grad_norm": 0.8596229297044873,
      "learning_rate": 2.4515412449615956e-05,
      "loss": 0.3027,
      "step": 7757
    },
    {
      "epoch": 5.137748344370861,
      "grad_norm": 0.7846535256689628,
      "learning_rate": 2.4478855845671336e-05,
      "loss": 0.2793,
      "step": 7758
    },
    {
      "epoch": 5.13841059602649,
      "grad_norm": 0.7163239492621535,
      "learning_rate": 2.444232409635234e-05,
      "loss": 0.2773,
      "step": 7759
    },
    {
      "epoch": 5.139072847682119,
      "grad_norm": 0.7312078921842875,
      "learning_rate": 2.4405817208892632e-05,
      "loss": 0.248,
      "step": 7760
    },
    {
      "epoch": 5.139735099337749,
      "grad_norm": 0.6990468500562959,
      "learning_rate": 2.436933519052098e-05,
      "loss": 0.2295,
      "step": 7761
    },
    {
      "epoch": 5.140397350993378,
      "grad_norm": 0.7988025120811293,
      "learning_rate": 2.4332878048461318e-05,
      "loss": 0.2754,
      "step": 7762
    },
    {
      "epoch": 5.1410596026490065,
      "grad_norm": 0.8475677177874262,
      "learning_rate": 2.4296445789932528e-05,
      "loss": 0.3066,
      "step": 7763
    },
    {
      "epoch": 5.1417218543046355,
      "grad_norm": 0.7022640182440665,
      "learning_rate": 2.4260038422148585e-05,
      "loss": 0.2207,
      "step": 7764
    },
    {
      "epoch": 5.142384105960265,
      "grad_norm": 0.668390192389239,
      "learning_rate": 2.4223655952318594e-05,
      "loss": 0.2207,
      "step": 7765
    },
    {
      "epoch": 5.143046357615894,
      "grad_norm": 0.7187381016044225,
      "learning_rate": 2.4187298387646735e-05,
      "loss": 0.2617,
      "step": 7766
    },
    {
      "epoch": 5.143708609271523,
      "grad_norm": 0.9069151226966364,
      "learning_rate": 2.4150965735332155e-05,
      "loss": 0.3438,
      "step": 7767
    },
    {
      "epoch": 5.144370860927152,
      "grad_norm": 0.6948619986328017,
      "learning_rate": 2.4114658002569213e-05,
      "loss": 0.2773,
      "step": 7768
    },
    {
      "epoch": 5.145033112582781,
      "grad_norm": 0.6989031727473147,
      "learning_rate": 2.407837519654718e-05,
      "loss": 0.2285,
      "step": 7769
    },
    {
      "epoch": 5.145695364238411,
      "grad_norm": 0.7325190222981751,
      "learning_rate": 2.40421173244505e-05,
      "loss": 0.2441,
      "step": 7770
    },
    {
      "epoch": 5.14635761589404,
      "grad_norm": 0.8138696336805727,
      "learning_rate": 2.400588439345867e-05,
      "loss": 0.2451,
      "step": 7771
    },
    {
      "epoch": 5.147019867549669,
      "grad_norm": 0.7051375699650541,
      "learning_rate": 2.3969676410746232e-05,
      "loss": 0.2324,
      "step": 7772
    },
    {
      "epoch": 5.147682119205298,
      "grad_norm": 0.7429945710207402,
      "learning_rate": 2.3933493383482666e-05,
      "loss": 0.2471,
      "step": 7773
    },
    {
      "epoch": 5.1483443708609276,
      "grad_norm": 0.7714852667435148,
      "learning_rate": 2.3897335318832795e-05,
      "loss": 0.2598,
      "step": 7774
    },
    {
      "epoch": 5.1490066225165565,
      "grad_norm": 0.7888821293211294,
      "learning_rate": 2.386120222395624e-05,
      "loss": 0.2637,
      "step": 7775
    },
    {
      "epoch": 5.149668874172185,
      "grad_norm": 0.8088833872670256,
      "learning_rate": 2.3825094106007742e-05,
      "loss": 0.248,
      "step": 7776
    },
    {
      "epoch": 5.150331125827814,
      "grad_norm": 0.6771132236529286,
      "learning_rate": 2.378901097213717e-05,
      "loss": 0.2275,
      "step": 7777
    },
    {
      "epoch": 5.150993377483443,
      "grad_norm": 0.9340225478813796,
      "learning_rate": 2.3752952829489417e-05,
      "loss": 0.3027,
      "step": 7778
    },
    {
      "epoch": 5.151655629139073,
      "grad_norm": 0.7285149406689923,
      "learning_rate": 2.3716919685204384e-05,
      "loss": 0.2617,
      "step": 7779
    },
    {
      "epoch": 5.152317880794702,
      "grad_norm": 0.809976497911641,
      "learning_rate": 2.3680911546417003e-05,
      "loss": 0.2754,
      "step": 7780
    },
    {
      "epoch": 5.152980132450331,
      "grad_norm": 0.6696806025911933,
      "learning_rate": 2.3644928420257343e-05,
      "loss": 0.2041,
      "step": 7781
    },
    {
      "epoch": 5.15364238410596,
      "grad_norm": 0.6446591335289431,
      "learning_rate": 2.36089703138505e-05,
      "loss": 0.2207,
      "step": 7782
    },
    {
      "epoch": 5.15430463576159,
      "grad_norm": 0.7615194567436615,
      "learning_rate": 2.3573037234316534e-05,
      "loss": 0.2471,
      "step": 7783
    },
    {
      "epoch": 5.154966887417219,
      "grad_norm": 0.7855549258138956,
      "learning_rate": 2.3537129188770648e-05,
      "loss": 0.2969,
      "step": 7784
    },
    {
      "epoch": 5.155629139072848,
      "grad_norm": 0.720816910822466,
      "learning_rate": 2.350124618432301e-05,
      "loss": 0.2617,
      "step": 7785
    },
    {
      "epoch": 5.156291390728477,
      "grad_norm": 0.6743108786856217,
      "learning_rate": 2.3465388228078896e-05,
      "loss": 0.2256,
      "step": 7786
    },
    {
      "epoch": 5.156953642384106,
      "grad_norm": 0.8109927685931756,
      "learning_rate": 2.3429555327138616e-05,
      "loss": 0.3047,
      "step": 7787
    },
    {
      "epoch": 5.157615894039735,
      "grad_norm": 0.7431018804929632,
      "learning_rate": 2.339374748859746e-05,
      "loss": 0.2461,
      "step": 7788
    },
    {
      "epoch": 5.158278145695364,
      "grad_norm": 0.6732401333972741,
      "learning_rate": 2.335796471954575e-05,
      "loss": 0.2334,
      "step": 7789
    },
    {
      "epoch": 5.158940397350993,
      "grad_norm": 0.8462147569929277,
      "learning_rate": 2.3322207027068945e-05,
      "loss": 0.3066,
      "step": 7790
    },
    {
      "epoch": 5.159602649006622,
      "grad_norm": 0.8237260248303043,
      "learning_rate": 2.3286474418247492e-05,
      "loss": 0.2812,
      "step": 7791
    },
    {
      "epoch": 5.160264900662252,
      "grad_norm": 0.7700633334375332,
      "learning_rate": 2.325076690015678e-05,
      "loss": 0.25,
      "step": 7792
    },
    {
      "epoch": 5.160927152317881,
      "grad_norm": 0.7263584361951742,
      "learning_rate": 2.3215084479867407e-05,
      "loss": 0.2676,
      "step": 7793
    },
    {
      "epoch": 5.16158940397351,
      "grad_norm": 0.6986797400725967,
      "learning_rate": 2.3179427164444803e-05,
      "loss": 0.2393,
      "step": 7794
    },
    {
      "epoch": 5.162251655629139,
      "grad_norm": 0.6907789414039427,
      "learning_rate": 2.314379496094957e-05,
      "loss": 0.2363,
      "step": 7795
    },
    {
      "epoch": 5.162913907284768,
      "grad_norm": 0.7675051356638901,
      "learning_rate": 2.3108187876437328e-05,
      "loss": 0.2695,
      "step": 7796
    },
    {
      "epoch": 5.163576158940398,
      "grad_norm": 0.8127241518139511,
      "learning_rate": 2.3072605917958625e-05,
      "loss": 0.3379,
      "step": 7797
    },
    {
      "epoch": 5.164238410596027,
      "grad_norm": 0.82816480949799,
      "learning_rate": 2.3037049092559123e-05,
      "loss": 0.293,
      "step": 7798
    },
    {
      "epoch": 5.164900662251656,
      "grad_norm": 0.7454294303979505,
      "learning_rate": 2.3001517407279523e-05,
      "loss": 0.2871,
      "step": 7799
    },
    {
      "epoch": 5.1655629139072845,
      "grad_norm": 0.73516489463546,
      "learning_rate": 2.2966010869155466e-05,
      "loss": 0.2236,
      "step": 7800
    },
    {
      "epoch": 5.166225165562914,
      "grad_norm": 0.9179353472508677,
      "learning_rate": 2.2930529485217626e-05,
      "loss": 0.3008,
      "step": 7801
    },
    {
      "epoch": 5.166887417218543,
      "grad_norm": 0.7277213196810184,
      "learning_rate": 2.2895073262491732e-05,
      "loss": 0.2617,
      "step": 7802
    },
    {
      "epoch": 5.167549668874172,
      "grad_norm": 0.8339817385295946,
      "learning_rate": 2.2859642207998585e-05,
      "loss": 0.3066,
      "step": 7803
    },
    {
      "epoch": 5.168211920529801,
      "grad_norm": 0.7429993147829709,
      "learning_rate": 2.2824236328753876e-05,
      "loss": 0.2578,
      "step": 7804
    },
    {
      "epoch": 5.16887417218543,
      "grad_norm": 0.703236513271614,
      "learning_rate": 2.278885563176841e-05,
      "loss": 0.2324,
      "step": 7805
    },
    {
      "epoch": 5.16953642384106,
      "grad_norm": 0.7673745687197612,
      "learning_rate": 2.2753500124047936e-05,
      "loss": 0.2246,
      "step": 7806
    },
    {
      "epoch": 5.170198675496689,
      "grad_norm": 0.7541916138461104,
      "learning_rate": 2.271816981259327e-05,
      "loss": 0.2734,
      "step": 7807
    },
    {
      "epoch": 5.170860927152318,
      "grad_norm": 0.6729437800769618,
      "learning_rate": 2.2682864704400243e-05,
      "loss": 0.2334,
      "step": 7808
    },
    {
      "epoch": 5.171523178807947,
      "grad_norm": 0.7576188537186267,
      "learning_rate": 2.2647584806459655e-05,
      "loss": 0.3047,
      "step": 7809
    },
    {
      "epoch": 5.172185430463577,
      "grad_norm": 0.8248779428480717,
      "learning_rate": 2.2612330125757266e-05,
      "loss": 0.248,
      "step": 7810
    },
    {
      "epoch": 5.1728476821192055,
      "grad_norm": 0.7960577424955629,
      "learning_rate": 2.2577100669274024e-05,
      "loss": 0.2754,
      "step": 7811
    },
    {
      "epoch": 5.1735099337748345,
      "grad_norm": 0.6245494287907758,
      "learning_rate": 2.2541896443985697e-05,
      "loss": 0.1924,
      "step": 7812
    },
    {
      "epoch": 5.174172185430463,
      "grad_norm": 0.7473964518000902,
      "learning_rate": 2.250671745686311e-05,
      "loss": 0.2402,
      "step": 7813
    },
    {
      "epoch": 5.174834437086092,
      "grad_norm": 0.7866086139276683,
      "learning_rate": 2.2471563714872127e-05,
      "loss": 0.2812,
      "step": 7814
    },
    {
      "epoch": 5.175496688741722,
      "grad_norm": 0.7868871682517076,
      "learning_rate": 2.2436435224973624e-05,
      "loss": 0.2383,
      "step": 7815
    },
    {
      "epoch": 5.176158940397351,
      "grad_norm": 0.7069579841525414,
      "learning_rate": 2.240133199412338e-05,
      "loss": 0.2412,
      "step": 7816
    },
    {
      "epoch": 5.17682119205298,
      "grad_norm": 0.7741840927510586,
      "learning_rate": 2.2366254029272324e-05,
      "loss": 0.2793,
      "step": 7817
    },
    {
      "epoch": 5.177483443708609,
      "grad_norm": 0.6504363679174688,
      "learning_rate": 2.233120133736619e-05,
      "loss": 0.2012,
      "step": 7818
    },
    {
      "epoch": 5.178145695364238,
      "grad_norm": 0.7472154203345664,
      "learning_rate": 2.2296173925345874e-05,
      "loss": 0.2754,
      "step": 7819
    },
    {
      "epoch": 5.178807947019868,
      "grad_norm": 0.9395590951717676,
      "learning_rate": 2.2261171800147228e-05,
      "loss": 0.3164,
      "step": 7820
    },
    {
      "epoch": 5.179470198675497,
      "grad_norm": 0.7232906213178408,
      "learning_rate": 2.222619496870104e-05,
      "loss": 0.2441,
      "step": 7821
    },
    {
      "epoch": 5.180132450331126,
      "grad_norm": 0.6917111067852199,
      "learning_rate": 2.2191243437933103e-05,
      "loss": 0.2402,
      "step": 7822
    },
    {
      "epoch": 5.180794701986755,
      "grad_norm": 0.801163140962666,
      "learning_rate": 2.2156317214764235e-05,
      "loss": 0.2969,
      "step": 7823
    },
    {
      "epoch": 5.1814569536423845,
      "grad_norm": 0.7272144161344692,
      "learning_rate": 2.2121416306110286e-05,
      "loss": 0.2461,
      "step": 7824
    },
    {
      "epoch": 5.182119205298013,
      "grad_norm": 0.8017637000059644,
      "learning_rate": 2.208654071888194e-05,
      "loss": 0.2871,
      "step": 7825
    },
    {
      "epoch": 5.182781456953642,
      "grad_norm": 0.7292843230729871,
      "learning_rate": 2.2051690459985044e-05,
      "loss": 0.2334,
      "step": 7826
    },
    {
      "epoch": 5.183443708609271,
      "grad_norm": 0.691693179915796,
      "learning_rate": 2.2016865536320293e-05,
      "loss": 0.2578,
      "step": 7827
    },
    {
      "epoch": 5.184105960264901,
      "grad_norm": 0.8590929320266867,
      "learning_rate": 2.198206595478344e-05,
      "loss": 0.2891,
      "step": 7828
    },
    {
      "epoch": 5.18476821192053,
      "grad_norm": 0.8534153344779773,
      "learning_rate": 2.1947291722265236e-05,
      "loss": 0.3047,
      "step": 7829
    },
    {
      "epoch": 5.185430463576159,
      "grad_norm": 0.7570824585480698,
      "learning_rate": 2.191254284565136e-05,
      "loss": 0.2461,
      "step": 7830
    },
    {
      "epoch": 5.186092715231788,
      "grad_norm": 0.64961243562121,
      "learning_rate": 2.1877819331822404e-05,
      "loss": 0.2207,
      "step": 7831
    },
    {
      "epoch": 5.186754966887417,
      "grad_norm": 0.7170010319819818,
      "learning_rate": 2.1843121187654154e-05,
      "loss": 0.2432,
      "step": 7832
    },
    {
      "epoch": 5.187417218543047,
      "grad_norm": 0.7639123302856651,
      "learning_rate": 2.180844842001718e-05,
      "loss": 0.2617,
      "step": 7833
    },
    {
      "epoch": 5.188079470198676,
      "grad_norm": 0.8247615476544231,
      "learning_rate": 2.177380103577706e-05,
      "loss": 0.2715,
      "step": 7834
    },
    {
      "epoch": 5.188741721854305,
      "grad_norm": 0.8239149477043434,
      "learning_rate": 2.1739179041794387e-05,
      "loss": 0.3105,
      "step": 7835
    },
    {
      "epoch": 5.1894039735099335,
      "grad_norm": 0.806295474284952,
      "learning_rate": 2.1704582444924774e-05,
      "loss": 0.2988,
      "step": 7836
    },
    {
      "epoch": 5.1900662251655625,
      "grad_norm": 0.7234980921769271,
      "learning_rate": 2.1670011252018693e-05,
      "loss": 0.2695,
      "step": 7837
    },
    {
      "epoch": 5.190728476821192,
      "grad_norm": 0.7916533289482816,
      "learning_rate": 2.1635465469921605e-05,
      "loss": 0.2441,
      "step": 7838
    },
    {
      "epoch": 5.191390728476821,
      "grad_norm": 0.6991993947714319,
      "learning_rate": 2.1600945105473998e-05,
      "loss": 0.2227,
      "step": 7839
    },
    {
      "epoch": 5.19205298013245,
      "grad_norm": 0.7583537012665814,
      "learning_rate": 2.1566450165511347e-05,
      "loss": 0.252,
      "step": 7840
    },
    {
      "epoch": 5.192715231788079,
      "grad_norm": 0.8020611092768729,
      "learning_rate": 2.153198065686398e-05,
      "loss": 0.3086,
      "step": 7841
    },
    {
      "epoch": 5.193377483443709,
      "grad_norm": 0.7821332148868907,
      "learning_rate": 2.1497536586357305e-05,
      "loss": 0.2852,
      "step": 7842
    },
    {
      "epoch": 5.194039735099338,
      "grad_norm": 0.7134476597583137,
      "learning_rate": 2.1463117960811587e-05,
      "loss": 0.2559,
      "step": 7843
    },
    {
      "epoch": 5.194701986754967,
      "grad_norm": 0.8481653724693978,
      "learning_rate": 2.142872478704214e-05,
      "loss": 0.293,
      "step": 7844
    },
    {
      "epoch": 5.195364238410596,
      "grad_norm": 0.6244441117367191,
      "learning_rate": 2.139435707185923e-05,
      "loss": 0.1953,
      "step": 7845
    },
    {
      "epoch": 5.196026490066225,
      "grad_norm": 0.8216860712768915,
      "learning_rate": 2.136001482206805e-05,
      "loss": 0.2754,
      "step": 7846
    },
    {
      "epoch": 5.196688741721855,
      "grad_norm": 0.9395771798685858,
      "learning_rate": 2.1325698044468707e-05,
      "loss": 0.3184,
      "step": 7847
    },
    {
      "epoch": 5.1973509933774835,
      "grad_norm": 0.7966636242608205,
      "learning_rate": 2.1291406745856353e-05,
      "loss": 0.2715,
      "step": 7848
    },
    {
      "epoch": 5.1980132450331125,
      "grad_norm": 0.7843248582356047,
      "learning_rate": 2.125714093302109e-05,
      "loss": 0.2832,
      "step": 7849
    },
    {
      "epoch": 5.198675496688741,
      "grad_norm": 0.7089643606691242,
      "learning_rate": 2.1222900612747882e-05,
      "loss": 0.2451,
      "step": 7850
    },
    {
      "epoch": 5.199337748344371,
      "grad_norm": 0.7545923243966016,
      "learning_rate": 2.1188685791816734e-05,
      "loss": 0.2539,
      "step": 7851
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.6145839073841828,
      "learning_rate": 2.1154496477002604e-05,
      "loss": 0.1934,
      "step": 7852
    },
    {
      "epoch": 5.200662251655629,
      "grad_norm": 0.8258804921652623,
      "learning_rate": 2.112033267507529e-05,
      "loss": 0.291,
      "step": 7853
    },
    {
      "epoch": 5.201324503311258,
      "grad_norm": 0.6946786096307236,
      "learning_rate": 2.108619439279972e-05,
      "loss": 0.2559,
      "step": 7854
    },
    {
      "epoch": 5.201986754966887,
      "grad_norm": 0.7213162679044548,
      "learning_rate": 2.105208163693557e-05,
      "loss": 0.2422,
      "step": 7855
    },
    {
      "epoch": 5.202649006622517,
      "grad_norm": 0.7822093233457468,
      "learning_rate": 2.101799441423758e-05,
      "loss": 0.2871,
      "step": 7856
    },
    {
      "epoch": 5.203311258278146,
      "grad_norm": 0.8267984183229521,
      "learning_rate": 2.0983932731455476e-05,
      "loss": 0.3203,
      "step": 7857
    },
    {
      "epoch": 5.203973509933775,
      "grad_norm": 0.6524904445273471,
      "learning_rate": 2.094989659533381e-05,
      "loss": 0.2266,
      "step": 7858
    },
    {
      "epoch": 5.204635761589404,
      "grad_norm": 0.6867435681303443,
      "learning_rate": 2.0915886012612086e-05,
      "loss": 0.2354,
      "step": 7859
    },
    {
      "epoch": 5.2052980132450335,
      "grad_norm": 0.7443296483922031,
      "learning_rate": 2.0881900990024842e-05,
      "loss": 0.2344,
      "step": 7860
    },
    {
      "epoch": 5.205960264900662,
      "grad_norm": 0.8009401097429937,
      "learning_rate": 2.084794153430154e-05,
      "loss": 0.2715,
      "step": 7861
    },
    {
      "epoch": 5.206622516556291,
      "grad_norm": 0.6800988859515609,
      "learning_rate": 2.081400765216643e-05,
      "loss": 0.2441,
      "step": 7862
    },
    {
      "epoch": 5.20728476821192,
      "grad_norm": 0.7658912952819359,
      "learning_rate": 2.078009935033893e-05,
      "loss": 0.2773,
      "step": 7863
    },
    {
      "epoch": 5.207947019867549,
      "grad_norm": 0.751982343956833,
      "learning_rate": 2.074621663553318e-05,
      "loss": 0.2598,
      "step": 7864
    },
    {
      "epoch": 5.208609271523179,
      "grad_norm": 0.7639573027285648,
      "learning_rate": 2.0712359514458366e-05,
      "loss": 0.2812,
      "step": 7865
    },
    {
      "epoch": 5.209271523178808,
      "grad_norm": 0.6831280557834392,
      "learning_rate": 2.0678527993818657e-05,
      "loss": 0.2178,
      "step": 7866
    },
    {
      "epoch": 5.209933774834437,
      "grad_norm": 0.8150073574257312,
      "learning_rate": 2.0644722080313007e-05,
      "loss": 0.2754,
      "step": 7867
    },
    {
      "epoch": 5.210596026490066,
      "grad_norm": 0.7671141751603547,
      "learning_rate": 2.0610941780635354e-05,
      "loss": 0.2734,
      "step": 7868
    },
    {
      "epoch": 5.211258278145696,
      "grad_norm": 0.7933255863244286,
      "learning_rate": 2.0577187101474674e-05,
      "loss": 0.252,
      "step": 7869
    },
    {
      "epoch": 5.211920529801325,
      "grad_norm": 0.8636251138367256,
      "learning_rate": 2.054345804951474e-05,
      "loss": 0.2773,
      "step": 7870
    },
    {
      "epoch": 5.212582781456954,
      "grad_norm": 0.7990707397393219,
      "learning_rate": 2.0509754631434238e-05,
      "loss": 0.2988,
      "step": 7871
    },
    {
      "epoch": 5.213245033112583,
      "grad_norm": 0.6783036740295046,
      "learning_rate": 2.0476076853906887e-05,
      "loss": 0.2314,
      "step": 7872
    },
    {
      "epoch": 5.2139072847682115,
      "grad_norm": 0.7531432680149214,
      "learning_rate": 2.04424247236013e-05,
      "loss": 0.2256,
      "step": 7873
    },
    {
      "epoch": 5.214569536423841,
      "grad_norm": 0.7732736746507498,
      "learning_rate": 2.0408798247180892e-05,
      "loss": 0.2656,
      "step": 7874
    },
    {
      "epoch": 5.21523178807947,
      "grad_norm": 0.7470185065825986,
      "learning_rate": 2.0375197431304192e-05,
      "loss": 0.2617,
      "step": 7875
    },
    {
      "epoch": 5.215894039735099,
      "grad_norm": 0.8260755186738208,
      "learning_rate": 2.0341622282624475e-05,
      "loss": 0.2969,
      "step": 7876
    },
    {
      "epoch": 5.216556291390728,
      "grad_norm": 0.9879287370913481,
      "learning_rate": 2.0308072807790026e-05,
      "loss": 0.3203,
      "step": 7877
    },
    {
      "epoch": 5.217218543046358,
      "grad_norm": 0.6561622019317833,
      "learning_rate": 2.0274549013444064e-05,
      "loss": 0.2109,
      "step": 7878
    },
    {
      "epoch": 5.217880794701987,
      "grad_norm": 0.7510663162973034,
      "learning_rate": 2.024105090622464e-05,
      "loss": 0.2559,
      "step": 7879
    },
    {
      "epoch": 5.218543046357616,
      "grad_norm": 0.6889266108870015,
      "learning_rate": 2.020757849276476e-05,
      "loss": 0.2129,
      "step": 7880
    },
    {
      "epoch": 5.219205298013245,
      "grad_norm": 0.7181296877900692,
      "learning_rate": 2.0174131779692343e-05,
      "loss": 0.2373,
      "step": 7881
    },
    {
      "epoch": 5.219867549668874,
      "grad_norm": 0.761794897635098,
      "learning_rate": 2.014071077363028e-05,
      "loss": 0.2373,
      "step": 7882
    },
    {
      "epoch": 5.220529801324504,
      "grad_norm": 0.9157205155805359,
      "learning_rate": 2.010731548119623e-05,
      "loss": 0.2969,
      "step": 7883
    },
    {
      "epoch": 5.221192052980133,
      "grad_norm": 0.769945762165555,
      "learning_rate": 2.0073945909002932e-05,
      "loss": 0.2598,
      "step": 7884
    },
    {
      "epoch": 5.2218543046357615,
      "grad_norm": 0.6875768181166205,
      "learning_rate": 2.0040602063657868e-05,
      "loss": 0.2354,
      "step": 7885
    },
    {
      "epoch": 5.22251655629139,
      "grad_norm": 0.761719545264721,
      "learning_rate": 2.000728395176351e-05,
      "loss": 0.2676,
      "step": 7886
    },
    {
      "epoch": 5.22317880794702,
      "grad_norm": 0.8300738737661454,
      "learning_rate": 1.9973991579917293e-05,
      "loss": 0.2891,
      "step": 7887
    },
    {
      "epoch": 5.223841059602649,
      "grad_norm": 0.7559725731714821,
      "learning_rate": 1.9940724954711414e-05,
      "loss": 0.2275,
      "step": 7888
    },
    {
      "epoch": 5.224503311258278,
      "grad_norm": 0.7890720223389843,
      "learning_rate": 1.9907484082733067e-05,
      "loss": 0.2891,
      "step": 7889
    },
    {
      "epoch": 5.225165562913907,
      "grad_norm": 0.7151515808464772,
      "learning_rate": 1.987426897056437e-05,
      "loss": 0.2598,
      "step": 7890
    },
    {
      "epoch": 5.225827814569536,
      "grad_norm": 0.795159509871878,
      "learning_rate": 1.9841079624782267e-05,
      "loss": 0.2793,
      "step": 7891
    },
    {
      "epoch": 5.226490066225166,
      "grad_norm": 0.6997504839004208,
      "learning_rate": 1.9807916051958584e-05,
      "loss": 0.2334,
      "step": 7892
    },
    {
      "epoch": 5.227152317880795,
      "grad_norm": 0.7878210230088095,
      "learning_rate": 1.9774778258660114e-05,
      "loss": 0.2852,
      "step": 7893
    },
    {
      "epoch": 5.227814569536424,
      "grad_norm": 0.740962289246586,
      "learning_rate": 1.9741666251448572e-05,
      "loss": 0.2344,
      "step": 7894
    },
    {
      "epoch": 5.228476821192053,
      "grad_norm": 0.6534268555726958,
      "learning_rate": 1.970858003688049e-05,
      "loss": 0.1846,
      "step": 7895
    },
    {
      "epoch": 5.2291390728476825,
      "grad_norm": 0.8160893194070954,
      "learning_rate": 1.967551962150725e-05,
      "loss": 0.2891,
      "step": 7896
    },
    {
      "epoch": 5.2298013245033115,
      "grad_norm": 0.857152422118708,
      "learning_rate": 1.964248501187526e-05,
      "loss": 0.2598,
      "step": 7897
    },
    {
      "epoch": 5.23046357615894,
      "grad_norm": 0.737825128083667,
      "learning_rate": 1.960947621452576e-05,
      "loss": 0.2637,
      "step": 7898
    },
    {
      "epoch": 5.231125827814569,
      "grad_norm": 0.8304129359859337,
      "learning_rate": 1.9576493235994826e-05,
      "loss": 0.2891,
      "step": 7899
    },
    {
      "epoch": 5.231788079470198,
      "grad_norm": 0.6782967651452708,
      "learning_rate": 1.9543536082813522e-05,
      "loss": 0.2275,
      "step": 7900
    },
    {
      "epoch": 5.232450331125828,
      "grad_norm": 0.7436280225406535,
      "learning_rate": 1.951060476150767e-05,
      "loss": 0.2598,
      "step": 7901
    },
    {
      "epoch": 5.233112582781457,
      "grad_norm": 0.7740599879493457,
      "learning_rate": 1.9477699278598073e-05,
      "loss": 0.2217,
      "step": 7902
    },
    {
      "epoch": 5.233774834437086,
      "grad_norm": 0.7235351000288407,
      "learning_rate": 1.9444819640600456e-05,
      "loss": 0.2139,
      "step": 7903
    },
    {
      "epoch": 5.234437086092715,
      "grad_norm": 0.9073241661614054,
      "learning_rate": 1.9411965854025324e-05,
      "loss": 0.3223,
      "step": 7904
    },
    {
      "epoch": 5.235099337748345,
      "grad_norm": 0.6191824626793746,
      "learning_rate": 1.9379137925378023e-05,
      "loss": 0.1953,
      "step": 7905
    },
    {
      "epoch": 5.235761589403974,
      "grad_norm": 0.7166782722205333,
      "learning_rate": 1.934633586115899e-05,
      "loss": 0.2451,
      "step": 7906
    },
    {
      "epoch": 5.236423841059603,
      "grad_norm": 0.6879512857012469,
      "learning_rate": 1.931355966786336e-05,
      "loss": 0.2031,
      "step": 7907
    },
    {
      "epoch": 5.237086092715232,
      "grad_norm": 0.8087431839652018,
      "learning_rate": 1.928080935198114e-05,
      "loss": 0.3125,
      "step": 7908
    },
    {
      "epoch": 5.237748344370861,
      "grad_norm": 0.8061488740893928,
      "learning_rate": 1.924808491999733e-05,
      "loss": 0.293,
      "step": 7909
    },
    {
      "epoch": 5.23841059602649,
      "grad_norm": 0.826498377892661,
      "learning_rate": 1.9215386378391763e-05,
      "loss": 0.3047,
      "step": 7910
    },
    {
      "epoch": 5.239072847682119,
      "grad_norm": 0.7654925352219261,
      "learning_rate": 1.9182713733639055e-05,
      "loss": 0.252,
      "step": 7911
    },
    {
      "epoch": 5.239735099337748,
      "grad_norm": 0.8919958068766045,
      "learning_rate": 1.915006699220882e-05,
      "loss": 0.3066,
      "step": 7912
    },
    {
      "epoch": 5.240397350993377,
      "grad_norm": 0.7521641041059619,
      "learning_rate": 1.9117446160565446e-05,
      "loss": 0.2656,
      "step": 7913
    },
    {
      "epoch": 5.241059602649006,
      "grad_norm": 0.7750077729378727,
      "learning_rate": 1.9084851245168242e-05,
      "loss": 0.25,
      "step": 7914
    },
    {
      "epoch": 5.241721854304636,
      "grad_norm": 0.7086573573327722,
      "learning_rate": 1.9052282252471422e-05,
      "loss": 0.252,
      "step": 7915
    },
    {
      "epoch": 5.242384105960265,
      "grad_norm": 0.8003024442763168,
      "learning_rate": 1.9019739188923984e-05,
      "loss": 0.3008,
      "step": 7916
    },
    {
      "epoch": 5.243046357615894,
      "grad_norm": 0.7399107732771303,
      "learning_rate": 1.8987222060969798e-05,
      "loss": 0.2598,
      "step": 7917
    },
    {
      "epoch": 5.243708609271523,
      "grad_norm": 0.6869904443572256,
      "learning_rate": 1.8954730875047653e-05,
      "loss": 0.2393,
      "step": 7918
    },
    {
      "epoch": 5.244370860927153,
      "grad_norm": 0.7365286220424062,
      "learning_rate": 1.89222656375912e-05,
      "loss": 0.2793,
      "step": 7919
    },
    {
      "epoch": 5.245033112582782,
      "grad_norm": 0.7372094423581625,
      "learning_rate": 1.8889826355028897e-05,
      "loss": 0.2559,
      "step": 7920
    },
    {
      "epoch": 5.2456953642384105,
      "grad_norm": 0.7187929352700633,
      "learning_rate": 1.8857413033784125e-05,
      "loss": 0.2266,
      "step": 7921
    },
    {
      "epoch": 5.2463576158940395,
      "grad_norm": 0.8472812920726802,
      "learning_rate": 1.882502568027503e-05,
      "loss": 0.2949,
      "step": 7922
    },
    {
      "epoch": 5.247019867549669,
      "grad_norm": 0.7074531065898034,
      "learning_rate": 1.8792664300914735e-05,
      "loss": 0.2402,
      "step": 7923
    },
    {
      "epoch": 5.247682119205298,
      "grad_norm": 0.61640663806564,
      "learning_rate": 1.8760328902111194e-05,
      "loss": 0.208,
      "step": 7924
    },
    {
      "epoch": 5.248344370860927,
      "grad_norm": 0.776443456901954,
      "learning_rate": 1.872801949026712e-05,
      "loss": 0.2109,
      "step": 7925
    },
    {
      "epoch": 5.249006622516556,
      "grad_norm": 0.7133866616249277,
      "learning_rate": 1.8695736071780133e-05,
      "loss": 0.21,
      "step": 7926
    },
    {
      "epoch": 5.249668874172185,
      "grad_norm": 0.6485833223334139,
      "learning_rate": 1.8663478653042797e-05,
      "loss": 0.2119,
      "step": 7927
    },
    {
      "epoch": 5.250331125827815,
      "grad_norm": 0.7427085708559611,
      "learning_rate": 1.8631247240442436e-05,
      "loss": 0.2129,
      "step": 7928
    },
    {
      "epoch": 5.250993377483444,
      "grad_norm": 0.7499296744343822,
      "learning_rate": 1.859904184036117e-05,
      "loss": 0.252,
      "step": 7929
    },
    {
      "epoch": 5.251655629139073,
      "grad_norm": 0.7985085171721937,
      "learning_rate": 1.8566862459176086e-05,
      "loss": 0.2656,
      "step": 7930
    },
    {
      "epoch": 5.252317880794702,
      "grad_norm": 0.9300195438691514,
      "learning_rate": 1.8534709103259095e-05,
      "loss": 0.3203,
      "step": 7931
    },
    {
      "epoch": 5.252980132450331,
      "grad_norm": 0.7642091787500077,
      "learning_rate": 1.8502581778976865e-05,
      "loss": 0.2559,
      "step": 7932
    },
    {
      "epoch": 5.2536423841059605,
      "grad_norm": 0.6849708734419366,
      "learning_rate": 1.8470480492691046e-05,
      "loss": 0.2119,
      "step": 7933
    },
    {
      "epoch": 5.2543046357615895,
      "grad_norm": 0.7035300854114488,
      "learning_rate": 1.8438405250758005e-05,
      "loss": 0.2334,
      "step": 7934
    },
    {
      "epoch": 5.254966887417218,
      "grad_norm": 0.7333351655402648,
      "learning_rate": 1.840635605952902e-05,
      "loss": 0.2217,
      "step": 7935
    },
    {
      "epoch": 5.255629139072847,
      "grad_norm": 0.9347980105663337,
      "learning_rate": 1.837433292535024e-05,
      "loss": 0.334,
      "step": 7936
    },
    {
      "epoch": 5.256291390728477,
      "grad_norm": 0.7373690683431211,
      "learning_rate": 1.8342335854562584e-05,
      "loss": 0.2227,
      "step": 7937
    },
    {
      "epoch": 5.256953642384106,
      "grad_norm": 0.6773016203022948,
      "learning_rate": 1.831036485350179e-05,
      "loss": 0.2236,
      "step": 7938
    },
    {
      "epoch": 5.257615894039735,
      "grad_norm": 0.8117642600088054,
      "learning_rate": 1.8278419928498543e-05,
      "loss": 0.3008,
      "step": 7939
    },
    {
      "epoch": 5.258278145695364,
      "grad_norm": 0.5921443733774829,
      "learning_rate": 1.82465010858783e-05,
      "loss": 0.2031,
      "step": 7940
    },
    {
      "epoch": 5.258940397350994,
      "grad_norm": 0.79400144697917,
      "learning_rate": 1.8214608331961338e-05,
      "loss": 0.2539,
      "step": 7941
    },
    {
      "epoch": 5.259602649006623,
      "grad_norm": 0.8095376884114691,
      "learning_rate": 1.8182741673062788e-05,
      "loss": 0.2754,
      "step": 7942
    },
    {
      "epoch": 5.260264900662252,
      "grad_norm": 0.6887194395153051,
      "learning_rate": 1.8150901115492656e-05,
      "loss": 0.2061,
      "step": 7943
    },
    {
      "epoch": 5.260927152317881,
      "grad_norm": 0.7188388448071006,
      "learning_rate": 1.811908666555566e-05,
      "loss": 0.2715,
      "step": 7944
    },
    {
      "epoch": 5.26158940397351,
      "grad_norm": 0.6546705021321286,
      "learning_rate": 1.8087298329551514e-05,
      "loss": 0.21,
      "step": 7945
    },
    {
      "epoch": 5.262251655629139,
      "grad_norm": 0.7282351660035895,
      "learning_rate": 1.80555361137746e-05,
      "loss": 0.2451,
      "step": 7946
    },
    {
      "epoch": 5.262913907284768,
      "grad_norm": 0.6548286582061331,
      "learning_rate": 1.8023800024514216e-05,
      "loss": 0.1973,
      "step": 7947
    },
    {
      "epoch": 5.263576158940397,
      "grad_norm": 0.7495955533899803,
      "learning_rate": 1.7992090068054526e-05,
      "loss": 0.2158,
      "step": 7948
    },
    {
      "epoch": 5.264238410596026,
      "grad_norm": 0.6720868251874537,
      "learning_rate": 1.7960406250674418e-05,
      "loss": 0.2344,
      "step": 7949
    },
    {
      "epoch": 5.264900662251655,
      "grad_norm": 0.7270314661402414,
      "learning_rate": 1.792874857864763e-05,
      "loss": 0.2383,
      "step": 7950
    },
    {
      "epoch": 5.265562913907285,
      "grad_norm": 0.6482808754839535,
      "learning_rate": 1.789711705824277e-05,
      "loss": 0.1709,
      "step": 7951
    },
    {
      "epoch": 5.266225165562914,
      "grad_norm": 0.7339027704826079,
      "learning_rate": 1.786551169572327e-05,
      "loss": 0.249,
      "step": 7952
    },
    {
      "epoch": 5.266887417218543,
      "grad_norm": 0.7516874149450281,
      "learning_rate": 1.783393249734734e-05,
      "loss": 0.2695,
      "step": 7953
    },
    {
      "epoch": 5.267549668874172,
      "grad_norm": 0.7135721213880135,
      "learning_rate": 1.780237946936796e-05,
      "loss": 0.2275,
      "step": 7954
    },
    {
      "epoch": 5.268211920529802,
      "grad_norm": 0.7632363839841702,
      "learning_rate": 1.7770852618033048e-05,
      "loss": 0.252,
      "step": 7955
    },
    {
      "epoch": 5.268874172185431,
      "grad_norm": 0.7847594420236709,
      "learning_rate": 1.773935194958531e-05,
      "loss": 0.2539,
      "step": 7956
    },
    {
      "epoch": 5.26953642384106,
      "grad_norm": 0.7970442442093321,
      "learning_rate": 1.7707877470262176e-05,
      "loss": 0.2852,
      "step": 7957
    },
    {
      "epoch": 5.2701986754966885,
      "grad_norm": 0.7830175258830214,
      "learning_rate": 1.7676429186296008e-05,
      "loss": 0.293,
      "step": 7958
    },
    {
      "epoch": 5.2708609271523175,
      "grad_norm": 0.7724537437183625,
      "learning_rate": 1.7645007103913877e-05,
      "loss": 0.2402,
      "step": 7959
    },
    {
      "epoch": 5.271523178807947,
      "grad_norm": 0.6441592542769375,
      "learning_rate": 1.761361122933774e-05,
      "loss": 0.2002,
      "step": 7960
    },
    {
      "epoch": 5.272185430463576,
      "grad_norm": 0.6780491321854711,
      "learning_rate": 1.7582241568784367e-05,
      "loss": 0.1914,
      "step": 7961
    },
    {
      "epoch": 5.272847682119205,
      "grad_norm": 0.7843465604244393,
      "learning_rate": 1.755089812846528e-05,
      "loss": 0.2676,
      "step": 7962
    },
    {
      "epoch": 5.273509933774834,
      "grad_norm": 0.8460037524301623,
      "learning_rate": 1.751958091458678e-05,
      "loss": 0.3066,
      "step": 7963
    },
    {
      "epoch": 5.274172185430464,
      "grad_norm": 0.6787237071126669,
      "learning_rate": 1.7488289933350168e-05,
      "loss": 0.2148,
      "step": 7964
    },
    {
      "epoch": 5.274834437086093,
      "grad_norm": 0.6779377647869225,
      "learning_rate": 1.7457025190951344e-05,
      "loss": 0.2197,
      "step": 7965
    },
    {
      "epoch": 5.275496688741722,
      "grad_norm": 0.9127408928580059,
      "learning_rate": 1.7425786693581057e-05,
      "loss": 0.2949,
      "step": 7966
    },
    {
      "epoch": 5.276158940397351,
      "grad_norm": 0.8119680153441015,
      "learning_rate": 1.739457444742492e-05,
      "loss": 0.2871,
      "step": 7967
    },
    {
      "epoch": 5.27682119205298,
      "grad_norm": 0.7548386737351861,
      "learning_rate": 1.736338845866334e-05,
      "loss": 0.2373,
      "step": 7968
    },
    {
      "epoch": 5.27748344370861,
      "grad_norm": 0.6898523205000069,
      "learning_rate": 1.7332228733471433e-05,
      "loss": 0.2275,
      "step": 7969
    },
    {
      "epoch": 5.2781456953642385,
      "grad_norm": 0.7687371226119528,
      "learning_rate": 1.7301095278019272e-05,
      "loss": 0.2734,
      "step": 7970
    },
    {
      "epoch": 5.278807947019867,
      "grad_norm": 0.8167464633390286,
      "learning_rate": 1.726998809847155e-05,
      "loss": 0.2773,
      "step": 7971
    },
    {
      "epoch": 5.279470198675496,
      "grad_norm": 0.5993317301250232,
      "learning_rate": 1.7238907200987885e-05,
      "loss": 0.1826,
      "step": 7972
    },
    {
      "epoch": 5.280132450331126,
      "grad_norm": 0.7109700283338855,
      "learning_rate": 1.7207852591722686e-05,
      "loss": 0.252,
      "step": 7973
    },
    {
      "epoch": 5.280794701986755,
      "grad_norm": 0.8931880179563709,
      "learning_rate": 1.7176824276825075e-05,
      "loss": 0.3086,
      "step": 7974
    },
    {
      "epoch": 5.281456953642384,
      "grad_norm": 0.779421517158717,
      "learning_rate": 1.7145822262439015e-05,
      "loss": 0.2676,
      "step": 7975
    },
    {
      "epoch": 5.282119205298013,
      "grad_norm": 0.7351259852084753,
      "learning_rate": 1.711484655470326e-05,
      "loss": 0.2236,
      "step": 7976
    },
    {
      "epoch": 5.282781456953642,
      "grad_norm": 0.7217881148605702,
      "learning_rate": 1.7083897159751392e-05,
      "loss": 0.2539,
      "step": 7977
    },
    {
      "epoch": 5.283443708609272,
      "grad_norm": 0.8161987976426147,
      "learning_rate": 1.705297408371169e-05,
      "loss": 0.2598,
      "step": 7978
    },
    {
      "epoch": 5.284105960264901,
      "grad_norm": 0.7419313117716437,
      "learning_rate": 1.702207733270734e-05,
      "loss": 0.2324,
      "step": 7979
    },
    {
      "epoch": 5.28476821192053,
      "grad_norm": 0.7344737137503939,
      "learning_rate": 1.699120691285618e-05,
      "loss": 0.2432,
      "step": 7980
    },
    {
      "epoch": 5.285430463576159,
      "grad_norm": 0.8403808901782538,
      "learning_rate": 1.696036283027096e-05,
      "loss": 0.2773,
      "step": 7981
    },
    {
      "epoch": 5.2860927152317885,
      "grad_norm": 0.7995969822485023,
      "learning_rate": 1.6929545091059184e-05,
      "loss": 0.2598,
      "step": 7982
    },
    {
      "epoch": 5.286754966887417,
      "grad_norm": 0.8638478822797097,
      "learning_rate": 1.6898753701323043e-05,
      "loss": 0.291,
      "step": 7983
    },
    {
      "epoch": 5.287417218543046,
      "grad_norm": 0.7229893962646534,
      "learning_rate": 1.6867988667159625e-05,
      "loss": 0.2227,
      "step": 7984
    },
    {
      "epoch": 5.288079470198675,
      "grad_norm": 0.9362202853223729,
      "learning_rate": 1.6837249994660794e-05,
      "loss": 0.334,
      "step": 7985
    },
    {
      "epoch": 5.288741721854304,
      "grad_norm": 0.788656762071385,
      "learning_rate": 1.6806537689913113e-05,
      "loss": 0.2793,
      "step": 7986
    },
    {
      "epoch": 5.289403973509934,
      "grad_norm": 0.6370836095391719,
      "learning_rate": 1.6775851758997966e-05,
      "loss": 0.1982,
      "step": 7987
    },
    {
      "epoch": 5.290066225165563,
      "grad_norm": 0.6828275624797505,
      "learning_rate": 1.6745192207991527e-05,
      "loss": 0.1982,
      "step": 7988
    },
    {
      "epoch": 5.290728476821192,
      "grad_norm": 0.8294928972465531,
      "learning_rate": 1.6714559042964786e-05,
      "loss": 0.2715,
      "step": 7989
    },
    {
      "epoch": 5.291390728476821,
      "grad_norm": 0.8355373736769015,
      "learning_rate": 1.6683952269983397e-05,
      "loss": 0.2754,
      "step": 7990
    },
    {
      "epoch": 5.292052980132451,
      "grad_norm": 0.748831502763545,
      "learning_rate": 1.665337189510789e-05,
      "loss": 0.2734,
      "step": 7991
    },
    {
      "epoch": 5.29271523178808,
      "grad_norm": 0.8038387893959709,
      "learning_rate": 1.6622817924393496e-05,
      "loss": 0.2832,
      "step": 7992
    },
    {
      "epoch": 5.293377483443709,
      "grad_norm": 0.8518082541684192,
      "learning_rate": 1.659229036389028e-05,
      "loss": 0.3066,
      "step": 7993
    },
    {
      "epoch": 5.294039735099338,
      "grad_norm": 0.8073892977347925,
      "learning_rate": 1.656178921964305e-05,
      "loss": 0.2773,
      "step": 7994
    },
    {
      "epoch": 5.2947019867549665,
      "grad_norm": 0.7115917237251048,
      "learning_rate": 1.653131449769139e-05,
      "loss": 0.2383,
      "step": 7995
    },
    {
      "epoch": 5.295364238410596,
      "grad_norm": 0.7465816932716353,
      "learning_rate": 1.65008662040696e-05,
      "loss": 0.2432,
      "step": 7996
    },
    {
      "epoch": 5.296026490066225,
      "grad_norm": 0.7800854869686216,
      "learning_rate": 1.6470444344806816e-05,
      "loss": 0.25,
      "step": 7997
    },
    {
      "epoch": 5.296688741721854,
      "grad_norm": 0.7087002327778797,
      "learning_rate": 1.6440048925926948e-05,
      "loss": 0.2412,
      "step": 7998
    },
    {
      "epoch": 5.297350993377483,
      "grad_norm": 0.7691620236605988,
      "learning_rate": 1.640967995344858e-05,
      "loss": 0.252,
      "step": 7999
    },
    {
      "epoch": 5.298013245033113,
      "grad_norm": 0.8919828131603305,
      "learning_rate": 1.6379337433385138e-05,
      "loss": 0.3066,
      "step": 8000
    },
    {
      "epoch": 5.298675496688742,
      "grad_norm": 0.7161009282025457,
      "learning_rate": 1.634902137174483e-05,
      "loss": 0.2471,
      "step": 8001
    },
    {
      "epoch": 5.299337748344371,
      "grad_norm": 0.7342675550219723,
      "learning_rate": 1.6318731774530506e-05,
      "loss": 0.248,
      "step": 8002
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.7595454341242521,
      "learning_rate": 1.628846864773994e-05,
      "loss": 0.2471,
      "step": 8003
    },
    {
      "epoch": 5.300662251655629,
      "grad_norm": 0.7521242169778093,
      "learning_rate": 1.625823199736551e-05,
      "loss": 0.2227,
      "step": 8004
    },
    {
      "epoch": 5.301324503311259,
      "grad_norm": 0.7104859619242662,
      "learning_rate": 1.622802182939444e-05,
      "loss": 0.2422,
      "step": 8005
    },
    {
      "epoch": 5.3019867549668875,
      "grad_norm": 0.8335162576467143,
      "learning_rate": 1.6197838149808717e-05,
      "loss": 0.2969,
      "step": 8006
    },
    {
      "epoch": 5.3026490066225165,
      "grad_norm": 0.7381661580603504,
      "learning_rate": 1.6167680964585033e-05,
      "loss": 0.2432,
      "step": 8007
    },
    {
      "epoch": 5.303311258278145,
      "grad_norm": 0.7697856224117433,
      "learning_rate": 1.613755027969483e-05,
      "loss": 0.2754,
      "step": 8008
    },
    {
      "epoch": 5.303973509933774,
      "grad_norm": 0.7617559791722613,
      "learning_rate": 1.6107446101104343e-05,
      "loss": 0.2852,
      "step": 8009
    },
    {
      "epoch": 5.304635761589404,
      "grad_norm": 0.7688447781625987,
      "learning_rate": 1.6077368434774597e-05,
      "loss": 0.2471,
      "step": 8010
    },
    {
      "epoch": 5.305298013245033,
      "grad_norm": 0.7226689335581242,
      "learning_rate": 1.604731728666127e-05,
      "loss": 0.2715,
      "step": 8011
    },
    {
      "epoch": 5.305960264900662,
      "grad_norm": 0.7725143756988609,
      "learning_rate": 1.601729266271481e-05,
      "loss": 0.2363,
      "step": 8012
    },
    {
      "epoch": 5.306622516556291,
      "grad_norm": 0.7207629112468867,
      "learning_rate": 1.5987294568880455e-05,
      "loss": 0.2266,
      "step": 8013
    },
    {
      "epoch": 5.307284768211921,
      "grad_norm": 0.680519637929438,
      "learning_rate": 1.5957323011098222e-05,
      "loss": 0.1914,
      "step": 8014
    },
    {
      "epoch": 5.30794701986755,
      "grad_norm": 0.7686441246641283,
      "learning_rate": 1.5927377995302747e-05,
      "loss": 0.2617,
      "step": 8015
    },
    {
      "epoch": 5.308609271523179,
      "grad_norm": 0.730025125195389,
      "learning_rate": 1.5897459527423556e-05,
      "loss": 0.2246,
      "step": 8016
    },
    {
      "epoch": 5.309271523178808,
      "grad_norm": 0.7095257039566637,
      "learning_rate": 1.5867567613384774e-05,
      "loss": 0.2344,
      "step": 8017
    },
    {
      "epoch": 5.3099337748344375,
      "grad_norm": 0.712177172534391,
      "learning_rate": 1.58377022591054e-05,
      "loss": 0.2363,
      "step": 8018
    },
    {
      "epoch": 5.3105960264900665,
      "grad_norm": 0.7099658537010103,
      "learning_rate": 1.580786347049912e-05,
      "loss": 0.25,
      "step": 8019
    },
    {
      "epoch": 5.311258278145695,
      "grad_norm": 0.7598955477276265,
      "learning_rate": 1.5778051253474328e-05,
      "loss": 0.2793,
      "step": 8020
    },
    {
      "epoch": 5.311920529801324,
      "grad_norm": 0.6599488963677888,
      "learning_rate": 1.5748265613934146e-05,
      "loss": 0.2178,
      "step": 8021
    },
    {
      "epoch": 5.312582781456953,
      "grad_norm": 0.7130910037952829,
      "learning_rate": 1.571850655777656e-05,
      "loss": 0.25,
      "step": 8022
    },
    {
      "epoch": 5.313245033112583,
      "grad_norm": 0.70726632600197,
      "learning_rate": 1.5688774090894174e-05,
      "loss": 0.2432,
      "step": 8023
    },
    {
      "epoch": 5.313907284768212,
      "grad_norm": 0.7505602569981783,
      "learning_rate": 1.5659068219174325e-05,
      "loss": 0.2383,
      "step": 8024
    },
    {
      "epoch": 5.314569536423841,
      "grad_norm": 0.8369938657147894,
      "learning_rate": 1.5629388948499126e-05,
      "loss": 0.293,
      "step": 8025
    },
    {
      "epoch": 5.31523178807947,
      "grad_norm": 0.7837462576587424,
      "learning_rate": 1.5599736284745444e-05,
      "loss": 0.2559,
      "step": 8026
    },
    {
      "epoch": 5.315894039735099,
      "grad_norm": 0.6959392875476804,
      "learning_rate": 1.5570110233784816e-05,
      "loss": 0.2344,
      "step": 8027
    },
    {
      "epoch": 5.316556291390729,
      "grad_norm": 0.7201707803254381,
      "learning_rate": 1.554051080148358e-05,
      "loss": 0.2432,
      "step": 8028
    },
    {
      "epoch": 5.317218543046358,
      "grad_norm": 0.7824244705095718,
      "learning_rate": 1.551093799370268e-05,
      "loss": 0.2773,
      "step": 8029
    },
    {
      "epoch": 5.317880794701987,
      "grad_norm": 0.8530055420298697,
      "learning_rate": 1.5481391816297956e-05,
      "loss": 0.2793,
      "step": 8030
    },
    {
      "epoch": 5.3185430463576155,
      "grad_norm": 0.777585389625231,
      "learning_rate": 1.545187227511987e-05,
      "loss": 0.2754,
      "step": 8031
    },
    {
      "epoch": 5.319205298013245,
      "grad_norm": 0.7889343015717932,
      "learning_rate": 1.5422379376013615e-05,
      "loss": 0.2773,
      "step": 8032
    },
    {
      "epoch": 5.319867549668874,
      "grad_norm": 0.8374539124221668,
      "learning_rate": 1.5392913124819106e-05,
      "loss": 0.3223,
      "step": 8033
    },
    {
      "epoch": 5.320529801324503,
      "grad_norm": 0.692479837902625,
      "learning_rate": 1.5363473527371006e-05,
      "loss": 0.2217,
      "step": 8034
    },
    {
      "epoch": 5.321192052980132,
      "grad_norm": 0.7558923536337373,
      "learning_rate": 1.5334060589498736e-05,
      "loss": 0.2559,
      "step": 8035
    },
    {
      "epoch": 5.321854304635762,
      "grad_norm": 0.7477017975474067,
      "learning_rate": 1.5304674317026338e-05,
      "loss": 0.2334,
      "step": 8036
    },
    {
      "epoch": 5.322516556291391,
      "grad_norm": 0.8381973476256512,
      "learning_rate": 1.5275314715772657e-05,
      "loss": 0.2969,
      "step": 8037
    },
    {
      "epoch": 5.32317880794702,
      "grad_norm": 0.7298714333744627,
      "learning_rate": 1.524598179155126e-05,
      "loss": 0.2344,
      "step": 8038
    },
    {
      "epoch": 5.323841059602649,
      "grad_norm": 0.6565277157097448,
      "learning_rate": 1.5216675550170348e-05,
      "loss": 0.2051,
      "step": 8039
    },
    {
      "epoch": 5.324503311258278,
      "grad_norm": 0.7765764066970834,
      "learning_rate": 1.518739599743295e-05,
      "loss": 0.252,
      "step": 8040
    },
    {
      "epoch": 5.325165562913908,
      "grad_norm": 0.8203342564757837,
      "learning_rate": 1.5158143139136675e-05,
      "loss": 0.2676,
      "step": 8041
    },
    {
      "epoch": 5.325827814569537,
      "grad_norm": 0.8454474062452767,
      "learning_rate": 1.5128916981073974e-05,
      "loss": 0.2812,
      "step": 8042
    },
    {
      "epoch": 5.3264900662251655,
      "grad_norm": 0.8248852089810802,
      "learning_rate": 1.5099717529032002e-05,
      "loss": 0.2988,
      "step": 8043
    },
    {
      "epoch": 5.3271523178807945,
      "grad_norm": 0.769150334674777,
      "learning_rate": 1.5070544788792532e-05,
      "loss": 0.25,
      "step": 8044
    },
    {
      "epoch": 5.327814569536423,
      "grad_norm": 0.8530608835527638,
      "learning_rate": 1.504139876613208e-05,
      "loss": 0.293,
      "step": 8045
    },
    {
      "epoch": 5.328476821192053,
      "grad_norm": 0.7251620538401817,
      "learning_rate": 1.5012279466821908e-05,
      "loss": 0.2441,
      "step": 8046
    },
    {
      "epoch": 5.329139072847682,
      "grad_norm": 0.8570179570353621,
      "learning_rate": 1.4983186896628042e-05,
      "loss": 0.3008,
      "step": 8047
    },
    {
      "epoch": 5.329801324503311,
      "grad_norm": 0.8136615154031157,
      "learning_rate": 1.4954121061311037e-05,
      "loss": 0.2695,
      "step": 8048
    },
    {
      "epoch": 5.33046357615894,
      "grad_norm": 0.6955191136404615,
      "learning_rate": 1.492508196662634e-05,
      "loss": 0.2373,
      "step": 8049
    },
    {
      "epoch": 5.33112582781457,
      "grad_norm": 0.7072012667937185,
      "learning_rate": 1.4896069618323985e-05,
      "loss": 0.2354,
      "step": 8050
    },
    {
      "epoch": 5.331788079470199,
      "grad_norm": 0.779486013021626,
      "learning_rate": 1.4867084022148746e-05,
      "loss": 0.2617,
      "step": 8051
    },
    {
      "epoch": 5.332450331125828,
      "grad_norm": 0.7903633861156499,
      "learning_rate": 1.483812518384016e-05,
      "loss": 0.248,
      "step": 8052
    },
    {
      "epoch": 5.333112582781457,
      "grad_norm": 0.7111625633884585,
      "learning_rate": 1.4809193109132362e-05,
      "loss": 0.2334,
      "step": 8053
    },
    {
      "epoch": 5.3337748344370866,
      "grad_norm": 0.7439465463306729,
      "learning_rate": 1.4780287803754181e-05,
      "loss": 0.2539,
      "step": 8054
    },
    {
      "epoch": 5.3344370860927155,
      "grad_norm": 0.7600225013024784,
      "learning_rate": 1.475140927342931e-05,
      "loss": 0.2578,
      "step": 8055
    },
    {
      "epoch": 5.335099337748344,
      "grad_norm": 0.7584012993543706,
      "learning_rate": 1.472255752387599e-05,
      "loss": 0.2734,
      "step": 8056
    },
    {
      "epoch": 5.335761589403973,
      "grad_norm": 0.6614252367719962,
      "learning_rate": 1.4693732560807147e-05,
      "loss": 0.2129,
      "step": 8057
    },
    {
      "epoch": 5.336423841059602,
      "grad_norm": 0.8407698237852888,
      "learning_rate": 1.4664934389930505e-05,
      "loss": 0.2734,
      "step": 8058
    },
    {
      "epoch": 5.337086092715232,
      "grad_norm": 0.73834019736086,
      "learning_rate": 1.463616301694845e-05,
      "loss": 0.2656,
      "step": 8059
    },
    {
      "epoch": 5.337748344370861,
      "grad_norm": 0.7252586684624422,
      "learning_rate": 1.4607418447557972e-05,
      "loss": 0.249,
      "step": 8060
    },
    {
      "epoch": 5.33841059602649,
      "grad_norm": 0.7757527177894843,
      "learning_rate": 1.457870068745089e-05,
      "loss": 0.3105,
      "step": 8061
    },
    {
      "epoch": 5.339072847682119,
      "grad_norm": 0.744245550807292,
      "learning_rate": 1.4550009742313618e-05,
      "loss": 0.2754,
      "step": 8062
    },
    {
      "epoch": 5.339735099337748,
      "grad_norm": 0.7528920580584396,
      "learning_rate": 1.452134561782728e-05,
      "loss": 0.2354,
      "step": 8063
    },
    {
      "epoch": 5.340397350993378,
      "grad_norm": 0.7742172318109762,
      "learning_rate": 1.4492708319667767e-05,
      "loss": 0.2441,
      "step": 8064
    },
    {
      "epoch": 5.341059602649007,
      "grad_norm": 0.7795519738458873,
      "learning_rate": 1.4464097853505546e-05,
      "loss": 0.249,
      "step": 8065
    },
    {
      "epoch": 5.341721854304636,
      "grad_norm": 0.7795912485267877,
      "learning_rate": 1.4435514225005784e-05,
      "loss": 0.25,
      "step": 8066
    },
    {
      "epoch": 5.342384105960265,
      "grad_norm": 0.7436473879365388,
      "learning_rate": 1.4406957439828387e-05,
      "loss": 0.2637,
      "step": 8067
    },
    {
      "epoch": 5.343046357615894,
      "grad_norm": 0.7662431429565507,
      "learning_rate": 1.4378427503627998e-05,
      "loss": 0.2715,
      "step": 8068
    },
    {
      "epoch": 5.343708609271523,
      "grad_norm": 0.8324831073256408,
      "learning_rate": 1.4349924422053799e-05,
      "loss": 0.2773,
      "step": 8069
    },
    {
      "epoch": 5.344370860927152,
      "grad_norm": 0.6958338353786345,
      "learning_rate": 1.4321448200749708e-05,
      "loss": 0.2129,
      "step": 8070
    },
    {
      "epoch": 5.345033112582781,
      "grad_norm": 0.7677448410136595,
      "learning_rate": 1.429299884535438e-05,
      "loss": 0.2441,
      "step": 8071
    },
    {
      "epoch": 5.34569536423841,
      "grad_norm": 0.7640324615076572,
      "learning_rate": 1.4264576361501146e-05,
      "loss": 0.2734,
      "step": 8072
    },
    {
      "epoch": 5.34635761589404,
      "grad_norm": 0.8297230975019833,
      "learning_rate": 1.4236180754817917e-05,
      "loss": 0.2754,
      "step": 8073
    },
    {
      "epoch": 5.347019867549669,
      "grad_norm": 0.709690343838788,
      "learning_rate": 1.4207812030927396e-05,
      "loss": 0.2578,
      "step": 8074
    },
    {
      "epoch": 5.347682119205298,
      "grad_norm": 0.7516017784484753,
      "learning_rate": 1.4179470195446891e-05,
      "loss": 0.252,
      "step": 8075
    },
    {
      "epoch": 5.348344370860927,
      "grad_norm": 0.6466929509174333,
      "learning_rate": 1.415115525398841e-05,
      "loss": 0.1982,
      "step": 8076
    },
    {
      "epoch": 5.349006622516557,
      "grad_norm": 0.7399392444832139,
      "learning_rate": 1.4122867212158673e-05,
      "loss": 0.2246,
      "step": 8077
    },
    {
      "epoch": 5.349668874172186,
      "grad_norm": 0.6584902328451135,
      "learning_rate": 1.4094606075558978e-05,
      "loss": 0.1934,
      "step": 8078
    },
    {
      "epoch": 5.350331125827815,
      "grad_norm": 0.7382598899854903,
      "learning_rate": 1.4066371849785385e-05,
      "loss": 0.2393,
      "step": 8079
    },
    {
      "epoch": 5.3509933774834435,
      "grad_norm": 0.8728424072823019,
      "learning_rate": 1.4038164540428637e-05,
      "loss": 0.2812,
      "step": 8080
    },
    {
      "epoch": 5.3516556291390724,
      "grad_norm": 0.725310618826158,
      "learning_rate": 1.4009984153074033e-05,
      "loss": 0.2139,
      "step": 8081
    },
    {
      "epoch": 5.352317880794702,
      "grad_norm": 0.6834211838119204,
      "learning_rate": 1.3981830693301644e-05,
      "loss": 0.207,
      "step": 8082
    },
    {
      "epoch": 5.352980132450331,
      "grad_norm": 0.7524304254416659,
      "learning_rate": 1.3953704166686158e-05,
      "loss": 0.252,
      "step": 8083
    },
    {
      "epoch": 5.35364238410596,
      "grad_norm": 0.7588940096867658,
      "learning_rate": 1.392560457879699e-05,
      "loss": 0.2617,
      "step": 8084
    },
    {
      "epoch": 5.354304635761589,
      "grad_norm": 0.701281907391126,
      "learning_rate": 1.3897531935198137e-05,
      "loss": 0.2451,
      "step": 8085
    },
    {
      "epoch": 5.354966887417219,
      "grad_norm": 0.676827253926598,
      "learning_rate": 1.3869486241448353e-05,
      "loss": 0.2305,
      "step": 8086
    },
    {
      "epoch": 5.355629139072848,
      "grad_norm": 0.7721843679891223,
      "learning_rate": 1.3841467503100944e-05,
      "loss": 0.2461,
      "step": 8087
    },
    {
      "epoch": 5.356291390728477,
      "grad_norm": 0.7334716276024691,
      "learning_rate": 1.3813475725703993e-05,
      "loss": 0.2734,
      "step": 8088
    },
    {
      "epoch": 5.356953642384106,
      "grad_norm": 0.7065869744921625,
      "learning_rate": 1.3785510914800197e-05,
      "loss": 0.2334,
      "step": 8089
    },
    {
      "epoch": 5.357615894039735,
      "grad_norm": 0.7793203996817449,
      "learning_rate": 1.3757573075926876e-05,
      "loss": 0.2539,
      "step": 8090
    },
    {
      "epoch": 5.3582781456953645,
      "grad_norm": 0.7959420095873805,
      "learning_rate": 1.3729662214616005e-05,
      "loss": 0.2637,
      "step": 8091
    },
    {
      "epoch": 5.3589403973509935,
      "grad_norm": 0.841821715067985,
      "learning_rate": 1.3701778336394364e-05,
      "loss": 0.2754,
      "step": 8092
    },
    {
      "epoch": 5.359602649006622,
      "grad_norm": 0.7939298880862516,
      "learning_rate": 1.367392144678322e-05,
      "loss": 0.2812,
      "step": 8093
    },
    {
      "epoch": 5.360264900662251,
      "grad_norm": 0.6817358925241885,
      "learning_rate": 1.3646091551298527e-05,
      "loss": 0.2148,
      "step": 8094
    },
    {
      "epoch": 5.360927152317881,
      "grad_norm": 0.6180326333111245,
      "learning_rate": 1.3618288655450942e-05,
      "loss": 0.1943,
      "step": 8095
    },
    {
      "epoch": 5.36158940397351,
      "grad_norm": 0.7149430222148255,
      "learning_rate": 1.3590512764745815e-05,
      "loss": 0.2422,
      "step": 8096
    },
    {
      "epoch": 5.362251655629139,
      "grad_norm": 0.6675901496507601,
      "learning_rate": 1.3562763884683008e-05,
      "loss": 0.2188,
      "step": 8097
    },
    {
      "epoch": 5.362913907284768,
      "grad_norm": 0.7210705471617311,
      "learning_rate": 1.3535042020757181e-05,
      "loss": 0.2383,
      "step": 8098
    },
    {
      "epoch": 5.363576158940397,
      "grad_norm": 0.7384161907382689,
      "learning_rate": 1.350734717845754e-05,
      "loss": 0.2539,
      "step": 8099
    },
    {
      "epoch": 5.364238410596027,
      "grad_norm": 0.7625919990337888,
      "learning_rate": 1.347967936326798e-05,
      "loss": 0.2832,
      "step": 8100
    },
    {
      "epoch": 5.364900662251656,
      "grad_norm": 0.7663930695629941,
      "learning_rate": 1.3452038580667091e-05,
      "loss": 0.252,
      "step": 8101
    },
    {
      "epoch": 5.365562913907285,
      "grad_norm": 0.7950490433041154,
      "learning_rate": 1.3424424836128022e-05,
      "loss": 0.25,
      "step": 8102
    },
    {
      "epoch": 5.366225165562914,
      "grad_norm": 0.7654606075112844,
      "learning_rate": 1.339683813511862e-05,
      "loss": 0.2441,
      "step": 8103
    },
    {
      "epoch": 5.3668874172185435,
      "grad_norm": 0.8597960791885009,
      "learning_rate": 1.3369278483101348e-05,
      "loss": 0.2949,
      "step": 8104
    },
    {
      "epoch": 5.367549668874172,
      "grad_norm": 0.7391333628468434,
      "learning_rate": 1.3341745885533406e-05,
      "loss": 0.2305,
      "step": 8105
    },
    {
      "epoch": 5.368211920529801,
      "grad_norm": 0.6865011956080486,
      "learning_rate": 1.3314240347866467e-05,
      "loss": 0.1992,
      "step": 8106
    },
    {
      "epoch": 5.36887417218543,
      "grad_norm": 0.8469854378389177,
      "learning_rate": 1.3286761875547025e-05,
      "loss": 0.3145,
      "step": 8107
    },
    {
      "epoch": 5.369536423841059,
      "grad_norm": 0.73542903618696,
      "learning_rate": 1.3259310474016077e-05,
      "loss": 0.2422,
      "step": 8108
    },
    {
      "epoch": 5.370198675496689,
      "grad_norm": 0.7121697340215851,
      "learning_rate": 1.3231886148709325e-05,
      "loss": 0.208,
      "step": 8109
    },
    {
      "epoch": 5.370860927152318,
      "grad_norm": 0.7611788184036535,
      "learning_rate": 1.320448890505716e-05,
      "loss": 0.2734,
      "step": 8110
    },
    {
      "epoch": 5.371523178807947,
      "grad_norm": 0.6641738081753786,
      "learning_rate": 1.3177118748484473e-05,
      "loss": 0.2051,
      "step": 8111
    },
    {
      "epoch": 5.372185430463576,
      "grad_norm": 0.7371043570499014,
      "learning_rate": 1.3149775684410862e-05,
      "loss": 0.2217,
      "step": 8112
    },
    {
      "epoch": 5.372847682119206,
      "grad_norm": 0.7235410944515224,
      "learning_rate": 1.312245971825065e-05,
      "loss": 0.2393,
      "step": 8113
    },
    {
      "epoch": 5.373509933774835,
      "grad_norm": 0.7146664462154936,
      "learning_rate": 1.3095170855412657e-05,
      "loss": 0.2441,
      "step": 8114
    },
    {
      "epoch": 5.374172185430464,
      "grad_norm": 0.7118476218223516,
      "learning_rate": 1.3067909101300377e-05,
      "loss": 0.2129,
      "step": 8115
    },
    {
      "epoch": 5.3748344370860925,
      "grad_norm": 0.7776591769575657,
      "learning_rate": 1.3040674461311979e-05,
      "loss": 0.2656,
      "step": 8116
    },
    {
      "epoch": 5.3754966887417215,
      "grad_norm": 0.662157128904447,
      "learning_rate": 1.3013466940840229e-05,
      "loss": 0.2217,
      "step": 8117
    },
    {
      "epoch": 5.376158940397351,
      "grad_norm": 0.8653796429167776,
      "learning_rate": 1.2986286545272518e-05,
      "loss": 0.2949,
      "step": 8118
    },
    {
      "epoch": 5.37682119205298,
      "grad_norm": 0.7816704252931724,
      "learning_rate": 1.295913327999089e-05,
      "loss": 0.3008,
      "step": 8119
    },
    {
      "epoch": 5.377483443708609,
      "grad_norm": 0.7293748306525842,
      "learning_rate": 1.2932007150371982e-05,
      "loss": 0.208,
      "step": 8120
    },
    {
      "epoch": 5.378145695364238,
      "grad_norm": 0.8902552451990786,
      "learning_rate": 1.2904908161787126e-05,
      "loss": 0.3301,
      "step": 8121
    },
    {
      "epoch": 5.378807947019867,
      "grad_norm": 0.7174210207758523,
      "learning_rate": 1.2877836319602164e-05,
      "loss": 0.2441,
      "step": 8122
    },
    {
      "epoch": 5.379470198675497,
      "grad_norm": 0.750952672623268,
      "learning_rate": 1.2850791629177675e-05,
      "loss": 0.2832,
      "step": 8123
    },
    {
      "epoch": 5.380132450331126,
      "grad_norm": 0.6931268188939322,
      "learning_rate": 1.2823774095868795e-05,
      "loss": 0.2275,
      "step": 8124
    },
    {
      "epoch": 5.380794701986755,
      "grad_norm": 0.8765104909900238,
      "learning_rate": 1.2796783725025323e-05,
      "loss": 0.2754,
      "step": 8125
    },
    {
      "epoch": 5.381456953642384,
      "grad_norm": 0.7339528779960038,
      "learning_rate": 1.2769820521991686e-05,
      "loss": 0.2295,
      "step": 8126
    },
    {
      "epoch": 5.382119205298014,
      "grad_norm": 0.7796513420193587,
      "learning_rate": 1.2742884492106865e-05,
      "loss": 0.2236,
      "step": 8127
    },
    {
      "epoch": 5.3827814569536425,
      "grad_norm": 0.8350439485368039,
      "learning_rate": 1.271597564070449e-05,
      "loss": 0.2891,
      "step": 8128
    },
    {
      "epoch": 5.3834437086092715,
      "grad_norm": 0.7954422380292928,
      "learning_rate": 1.2689093973112852e-05,
      "loss": 0.2227,
      "step": 8129
    },
    {
      "epoch": 5.3841059602649,
      "grad_norm": 0.8668144498166878,
      "learning_rate": 1.2662239494654858e-05,
      "loss": 0.2578,
      "step": 8130
    },
    {
      "epoch": 5.38476821192053,
      "grad_norm": 0.5885514608931014,
      "learning_rate": 1.2635412210647938e-05,
      "loss": 0.1689,
      "step": 8131
    },
    {
      "epoch": 5.385430463576159,
      "grad_norm": 0.9297808001075322,
      "learning_rate": 1.2608612126404244e-05,
      "loss": 0.3105,
      "step": 8132
    },
    {
      "epoch": 5.386092715231788,
      "grad_norm": 0.7160444303368904,
      "learning_rate": 1.2581839247230496e-05,
      "loss": 0.2393,
      "step": 8133
    },
    {
      "epoch": 5.386754966887417,
      "grad_norm": 0.77735727113849,
      "learning_rate": 1.2555093578428022e-05,
      "loss": 0.2451,
      "step": 8134
    },
    {
      "epoch": 5.387417218543046,
      "grad_norm": 0.7941862836734495,
      "learning_rate": 1.2528375125292783e-05,
      "loss": 0.2695,
      "step": 8135
    },
    {
      "epoch": 5.388079470198676,
      "grad_norm": 0.9282686334098263,
      "learning_rate": 1.2501683893115317e-05,
      "loss": 0.3184,
      "step": 8136
    },
    {
      "epoch": 5.388741721854305,
      "grad_norm": 0.7680855118564253,
      "learning_rate": 1.2475019887180793e-05,
      "loss": 0.2412,
      "step": 8137
    },
    {
      "epoch": 5.389403973509934,
      "grad_norm": 0.7090166065141378,
      "learning_rate": 1.244838311276904e-05,
      "loss": 0.2148,
      "step": 8138
    },
    {
      "epoch": 5.390066225165563,
      "grad_norm": 0.7411006042308812,
      "learning_rate": 1.2421773575154404e-05,
      "loss": 0.249,
      "step": 8139
    },
    {
      "epoch": 5.390728476821192,
      "grad_norm": 0.705245131197477,
      "learning_rate": 1.2395191279605854e-05,
      "loss": 0.2314,
      "step": 8140
    },
    {
      "epoch": 5.391390728476821,
      "grad_norm": 0.8648090928323934,
      "learning_rate": 1.2368636231387013e-05,
      "loss": 0.2891,
      "step": 8141
    },
    {
      "epoch": 5.39205298013245,
      "grad_norm": 0.9133575883761696,
      "learning_rate": 1.2342108435756127e-05,
      "loss": 0.3066,
      "step": 8142
    },
    {
      "epoch": 5.392715231788079,
      "grad_norm": 0.8290083753182598,
      "learning_rate": 1.231560789796594e-05,
      "loss": 0.293,
      "step": 8143
    },
    {
      "epoch": 5.393377483443708,
      "grad_norm": 0.7248293484836095,
      "learning_rate": 1.2289134623263907e-05,
      "loss": 0.249,
      "step": 8144
    },
    {
      "epoch": 5.394039735099338,
      "grad_norm": 0.7134182883441372,
      "learning_rate": 1.226268861689202e-05,
      "loss": 0.2188,
      "step": 8145
    },
    {
      "epoch": 5.394701986754967,
      "grad_norm": 0.7024244159127842,
      "learning_rate": 1.223626988408687e-05,
      "loss": 0.2236,
      "step": 8146
    },
    {
      "epoch": 5.395364238410596,
      "grad_norm": 0.7117994170427495,
      "learning_rate": 1.2209878430079723e-05,
      "loss": 0.2266,
      "step": 8147
    },
    {
      "epoch": 5.396026490066225,
      "grad_norm": 0.9054738899100201,
      "learning_rate": 1.2183514260096367e-05,
      "loss": 0.2891,
      "step": 8148
    },
    {
      "epoch": 5.396688741721855,
      "grad_norm": 0.7010688716628896,
      "learning_rate": 1.2157177379357141e-05,
      "loss": 0.2217,
      "step": 8149
    },
    {
      "epoch": 5.397350993377484,
      "grad_norm": 0.752388869843788,
      "learning_rate": 1.213086779307716e-05,
      "loss": 0.2773,
      "step": 8150
    },
    {
      "epoch": 5.398013245033113,
      "grad_norm": 0.8115645998945585,
      "learning_rate": 1.2104585506465987e-05,
      "loss": 0.2734,
      "step": 8151
    },
    {
      "epoch": 5.398675496688742,
      "grad_norm": 0.802017631925854,
      "learning_rate": 1.2078330524727774e-05,
      "loss": 0.3105,
      "step": 8152
    },
    {
      "epoch": 5.3993377483443705,
      "grad_norm": 0.8211867609746558,
      "learning_rate": 1.2052102853061318e-05,
      "loss": 0.2695,
      "step": 8153
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.7795316870922536,
      "learning_rate": 1.202590249666006e-05,
      "loss": 0.2793,
      "step": 8154
    },
    {
      "epoch": 5.400662251655629,
      "grad_norm": 0.7850957469530484,
      "learning_rate": 1.1999729460711905e-05,
      "loss": 0.2695,
      "step": 8155
    },
    {
      "epoch": 5.401324503311258,
      "grad_norm": 0.8452925272713329,
      "learning_rate": 1.1973583750399468e-05,
      "loss": 0.3125,
      "step": 8156
    },
    {
      "epoch": 5.401986754966887,
      "grad_norm": 0.7576852496345284,
      "learning_rate": 1.1947465370899828e-05,
      "loss": 0.2578,
      "step": 8157
    },
    {
      "epoch": 5.402649006622516,
      "grad_norm": 0.7589360996698207,
      "learning_rate": 1.1921374327384781e-05,
      "loss": 0.25,
      "step": 8158
    },
    {
      "epoch": 5.403311258278146,
      "grad_norm": 0.7372719760609229,
      "learning_rate": 1.189531062502066e-05,
      "loss": 0.2197,
      "step": 8159
    },
    {
      "epoch": 5.403973509933775,
      "grad_norm": 0.7028243003358164,
      "learning_rate": 1.1869274268968354e-05,
      "loss": 0.2256,
      "step": 8160
    },
    {
      "epoch": 5.404635761589404,
      "grad_norm": 0.7227228402664468,
      "learning_rate": 1.1843265264383323e-05,
      "loss": 0.2031,
      "step": 8161
    },
    {
      "epoch": 5.405298013245033,
      "grad_norm": 0.7494359889569733,
      "learning_rate": 1.1817283616415696e-05,
      "loss": 0.2637,
      "step": 8162
    },
    {
      "epoch": 5.405960264900663,
      "grad_norm": 0.6085543777252429,
      "learning_rate": 1.1791329330210142e-05,
      "loss": 0.1943,
      "step": 8163
    },
    {
      "epoch": 5.406622516556292,
      "grad_norm": 0.7947274207666705,
      "learning_rate": 1.1765402410905882e-05,
      "loss": 0.2383,
      "step": 8164
    },
    {
      "epoch": 5.4072847682119205,
      "grad_norm": 0.7348735191153554,
      "learning_rate": 1.1739502863636762e-05,
      "loss": 0.2246,
      "step": 8165
    },
    {
      "epoch": 5.407947019867549,
      "grad_norm": 0.7447796731198978,
      "learning_rate": 1.1713630693531163e-05,
      "loss": 0.2578,
      "step": 8166
    },
    {
      "epoch": 5.408609271523179,
      "grad_norm": 0.6633076058683766,
      "learning_rate": 1.1687785905712082e-05,
      "loss": 0.2021,
      "step": 8167
    },
    {
      "epoch": 5.409271523178808,
      "grad_norm": 0.7755868988314782,
      "learning_rate": 1.1661968505297115e-05,
      "loss": 0.2334,
      "step": 8168
    },
    {
      "epoch": 5.409933774834437,
      "grad_norm": 0.7979223686511705,
      "learning_rate": 1.1636178497398386e-05,
      "loss": 0.2539,
      "step": 8169
    },
    {
      "epoch": 5.410596026490066,
      "grad_norm": 0.6340012719774182,
      "learning_rate": 1.1610415887122543e-05,
      "loss": 0.1914,
      "step": 8170
    },
    {
      "epoch": 5.411258278145695,
      "grad_norm": 0.7629744640947688,
      "learning_rate": 1.1584680679571006e-05,
      "loss": 0.2354,
      "step": 8171
    },
    {
      "epoch": 5.411920529801325,
      "grad_norm": 0.6710205472309697,
      "learning_rate": 1.155897287983955e-05,
      "loss": 0.207,
      "step": 8172
    },
    {
      "epoch": 5.412582781456954,
      "grad_norm": 0.6640371554223963,
      "learning_rate": 1.1533292493018637e-05,
      "loss": 0.2051,
      "step": 8173
    },
    {
      "epoch": 5.413245033112583,
      "grad_norm": 0.7383189576535907,
      "learning_rate": 1.1507639524193263e-05,
      "loss": 0.2422,
      "step": 8174
    },
    {
      "epoch": 5.413907284768212,
      "grad_norm": 0.7961543215425335,
      "learning_rate": 1.1482013978443067e-05,
      "loss": 0.2832,
      "step": 8175
    },
    {
      "epoch": 5.414569536423841,
      "grad_norm": 0.822293768376884,
      "learning_rate": 1.1456415860842122e-05,
      "loss": 0.2734,
      "step": 8176
    },
    {
      "epoch": 5.4152317880794705,
      "grad_norm": 0.8304980819465055,
      "learning_rate": 1.1430845176459208e-05,
      "loss": 0.2471,
      "step": 8177
    },
    {
      "epoch": 5.415894039735099,
      "grad_norm": 0.6869328437647758,
      "learning_rate": 1.1405301930357574e-05,
      "loss": 0.1953,
      "step": 8178
    },
    {
      "epoch": 5.416556291390728,
      "grad_norm": 0.8053336974629604,
      "learning_rate": 1.1379786127595125e-05,
      "loss": 0.2812,
      "step": 8179
    },
    {
      "epoch": 5.417218543046357,
      "grad_norm": 0.7645932450540927,
      "learning_rate": 1.1354297773224218e-05,
      "loss": 0.249,
      "step": 8180
    },
    {
      "epoch": 5.417880794701987,
      "grad_norm": 0.789999939534861,
      "learning_rate": 1.1328836872291903e-05,
      "loss": 0.2559,
      "step": 8181
    },
    {
      "epoch": 5.418543046357616,
      "grad_norm": 0.780082624630594,
      "learning_rate": 1.1303403429839675e-05,
      "loss": 0.2539,
      "step": 8182
    },
    {
      "epoch": 5.419205298013245,
      "grad_norm": 0.7400716013208438,
      "learning_rate": 1.1277997450903676e-05,
      "loss": 0.2236,
      "step": 8183
    },
    {
      "epoch": 5.419867549668874,
      "grad_norm": 0.8470551630512101,
      "learning_rate": 1.1252618940514596e-05,
      "loss": 0.3008,
      "step": 8184
    },
    {
      "epoch": 5.420529801324503,
      "grad_norm": 0.8115084483960443,
      "learning_rate": 1.122726790369765e-05,
      "loss": 0.25,
      "step": 8185
    },
    {
      "epoch": 5.421192052980133,
      "grad_norm": 0.8175666513983704,
      "learning_rate": 1.1201944345472608e-05,
      "loss": 0.2852,
      "step": 8186
    },
    {
      "epoch": 5.421854304635762,
      "grad_norm": 0.7363014576660762,
      "learning_rate": 1.1176648270853873e-05,
      "loss": 0.2314,
      "step": 8187
    },
    {
      "epoch": 5.422516556291391,
      "grad_norm": 0.6597580004001037,
      "learning_rate": 1.1151379684850353e-05,
      "loss": 0.2393,
      "step": 8188
    },
    {
      "epoch": 5.42317880794702,
      "grad_norm": 0.6665124518440448,
      "learning_rate": 1.1126138592465483e-05,
      "loss": 0.2236,
      "step": 8189
    },
    {
      "epoch": 5.423841059602649,
      "grad_norm": 0.7395893595479944,
      "learning_rate": 1.1100924998697296e-05,
      "loss": 0.2148,
      "step": 8190
    },
    {
      "epoch": 5.424503311258278,
      "grad_norm": 0.8222924520485109,
      "learning_rate": 1.1075738908538417e-05,
      "loss": 0.291,
      "step": 8191
    },
    {
      "epoch": 5.425165562913907,
      "grad_norm": 0.8495959298384754,
      "learning_rate": 1.1050580326975921e-05,
      "loss": 0.2373,
      "step": 8192
    },
    {
      "epoch": 5.425827814569536,
      "grad_norm": 0.8384429757339605,
      "learning_rate": 1.1025449258991559e-05,
      "loss": 0.2578,
      "step": 8193
    },
    {
      "epoch": 5.426490066225165,
      "grad_norm": 0.7562773819474549,
      "learning_rate": 1.1000345709561497e-05,
      "loss": 0.2031,
      "step": 8194
    },
    {
      "epoch": 5.427152317880795,
      "grad_norm": 0.6754742488972901,
      "learning_rate": 1.0975269683656578e-05,
      "loss": 0.2061,
      "step": 8195
    },
    {
      "epoch": 5.427814569536424,
      "grad_norm": 0.7575133685163968,
      "learning_rate": 1.0950221186242147e-05,
      "loss": 0.2354,
      "step": 8196
    },
    {
      "epoch": 5.428476821192053,
      "grad_norm": 0.7135852000437264,
      "learning_rate": 1.092520022227807e-05,
      "loss": 0.2305,
      "step": 8197
    },
    {
      "epoch": 5.429139072847682,
      "grad_norm": 0.7543585159134615,
      "learning_rate": 1.0900206796718764e-05,
      "loss": 0.2324,
      "step": 8198
    },
    {
      "epoch": 5.429801324503312,
      "grad_norm": 0.8129481972683335,
      "learning_rate": 1.087524091451324e-05,
      "loss": 0.25,
      "step": 8199
    },
    {
      "epoch": 5.430463576158941,
      "grad_norm": 0.8077685259747593,
      "learning_rate": 1.0850302580605058e-05,
      "loss": 0.2295,
      "step": 8200
    },
    {
      "epoch": 5.4311258278145695,
      "grad_norm": 0.7457859115731689,
      "learning_rate": 1.0825391799932237e-05,
      "loss": 0.2539,
      "step": 8201
    },
    {
      "epoch": 5.4317880794701985,
      "grad_norm": 0.7500849211638266,
      "learning_rate": 1.0800508577427447e-05,
      "loss": 0.25,
      "step": 8202
    },
    {
      "epoch": 5.432450331125827,
      "grad_norm": 0.6969300558331686,
      "learning_rate": 1.0775652918017796e-05,
      "loss": 0.208,
      "step": 8203
    },
    {
      "epoch": 5.433112582781457,
      "grad_norm": 0.7941004332415325,
      "learning_rate": 1.0750824826625031e-05,
      "loss": 0.2695,
      "step": 8204
    },
    {
      "epoch": 5.433774834437086,
      "grad_norm": 0.7423193525808635,
      "learning_rate": 1.072602430816542e-05,
      "loss": 0.2412,
      "step": 8205
    },
    {
      "epoch": 5.434437086092715,
      "grad_norm": 0.7283960495069702,
      "learning_rate": 1.0701251367549718e-05,
      "loss": 0.2461,
      "step": 8206
    },
    {
      "epoch": 5.435099337748344,
      "grad_norm": 0.8535558556368544,
      "learning_rate": 1.06765060096832e-05,
      "loss": 0.2734,
      "step": 8207
    },
    {
      "epoch": 5.435761589403974,
      "grad_norm": 0.7525491434304608,
      "learning_rate": 1.065178823946583e-05,
      "loss": 0.2422,
      "step": 8208
    },
    {
      "epoch": 5.436423841059603,
      "grad_norm": 0.7630745567166811,
      "learning_rate": 1.0627098061791978e-05,
      "loss": 0.2559,
      "step": 8209
    },
    {
      "epoch": 5.437086092715232,
      "grad_norm": 0.8634963924755517,
      "learning_rate": 1.060243548155053e-05,
      "loss": 0.2773,
      "step": 8210
    },
    {
      "epoch": 5.437748344370861,
      "grad_norm": 0.7700201159252429,
      "learning_rate": 1.0577800503625e-05,
      "loss": 0.2695,
      "step": 8211
    },
    {
      "epoch": 5.43841059602649,
      "grad_norm": 0.782024996803962,
      "learning_rate": 1.0553193132893417e-05,
      "loss": 0.2695,
      "step": 8212
    },
    {
      "epoch": 5.4390728476821195,
      "grad_norm": 0.837352402992951,
      "learning_rate": 1.0528613374228284e-05,
      "loss": 0.2871,
      "step": 8213
    },
    {
      "epoch": 5.4397350993377485,
      "grad_norm": 0.8052086983769423,
      "learning_rate": 1.0504061232496709e-05,
      "loss": 0.2832,
      "step": 8214
    },
    {
      "epoch": 5.440397350993377,
      "grad_norm": 0.910522628102496,
      "learning_rate": 1.047953671256025e-05,
      "loss": 0.3027,
      "step": 8215
    },
    {
      "epoch": 5.441059602649006,
      "grad_norm": 0.6726072249008708,
      "learning_rate": 1.0455039819275074e-05,
      "loss": 0.2197,
      "step": 8216
    },
    {
      "epoch": 5.441721854304636,
      "grad_norm": 0.6823027367476258,
      "learning_rate": 1.0430570557491869e-05,
      "loss": 0.249,
      "step": 8217
    },
    {
      "epoch": 5.442384105960265,
      "grad_norm": 0.7412709353727522,
      "learning_rate": 1.0406128932055807e-05,
      "loss": 0.2246,
      "step": 8218
    },
    {
      "epoch": 5.443046357615894,
      "grad_norm": 0.7663270401544181,
      "learning_rate": 1.0381714947806586e-05,
      "loss": 0.2539,
      "step": 8219
    },
    {
      "epoch": 5.443708609271523,
      "grad_norm": 0.8462294760811713,
      "learning_rate": 1.0357328609578468e-05,
      "loss": 0.2812,
      "step": 8220
    },
    {
      "epoch": 5.444370860927152,
      "grad_norm": 0.7292670055148073,
      "learning_rate": 1.0332969922200263e-05,
      "loss": 0.2227,
      "step": 8221
    },
    {
      "epoch": 5.445033112582782,
      "grad_norm": 0.8815330380510904,
      "learning_rate": 1.0308638890495225e-05,
      "loss": 0.3145,
      "step": 8222
    },
    {
      "epoch": 5.445695364238411,
      "grad_norm": 0.8436743622481963,
      "learning_rate": 1.0284335519281217e-05,
      "loss": 0.2637,
      "step": 8223
    },
    {
      "epoch": 5.44635761589404,
      "grad_norm": 0.6737165412135997,
      "learning_rate": 1.0260059813370541e-05,
      "loss": 0.2344,
      "step": 8224
    },
    {
      "epoch": 5.447019867549669,
      "grad_norm": 0.7995762922951847,
      "learning_rate": 1.0235811777570086e-05,
      "loss": 0.2754,
      "step": 8225
    },
    {
      "epoch": 5.447682119205298,
      "grad_norm": 0.8482090119796771,
      "learning_rate": 1.021159141668126e-05,
      "loss": 0.3027,
      "step": 8226
    },
    {
      "epoch": 5.448344370860927,
      "grad_norm": 0.7992928107368851,
      "learning_rate": 1.0187398735499941e-05,
      "loss": 0.2637,
      "step": 8227
    },
    {
      "epoch": 5.449006622516556,
      "grad_norm": 0.8022607478377197,
      "learning_rate": 1.0163233738816568e-05,
      "loss": 0.2832,
      "step": 8228
    },
    {
      "epoch": 5.449668874172185,
      "grad_norm": 0.7048177652434733,
      "learning_rate": 1.0139096431416126e-05,
      "loss": 0.2207,
      "step": 8229
    },
    {
      "epoch": 5.450331125827814,
      "grad_norm": 0.7424175924202886,
      "learning_rate": 1.011498681807803e-05,
      "loss": 0.2256,
      "step": 8230
    },
    {
      "epoch": 5.450993377483444,
      "grad_norm": 0.8678018402148715,
      "learning_rate": 1.0090904903576258e-05,
      "loss": 0.2715,
      "step": 8231
    },
    {
      "epoch": 5.451655629139073,
      "grad_norm": 0.705198379446596,
      "learning_rate": 1.0066850692679312e-05,
      "loss": 0.2295,
      "step": 8232
    },
    {
      "epoch": 5.452317880794702,
      "grad_norm": 0.724380239751693,
      "learning_rate": 1.0042824190150234e-05,
      "loss": 0.249,
      "step": 8233
    },
    {
      "epoch": 5.452980132450331,
      "grad_norm": 0.76509104105022,
      "learning_rate": 1.0018825400746517e-05,
      "loss": 0.2393,
      "step": 8234
    },
    {
      "epoch": 5.45364238410596,
      "grad_norm": 0.6892756107253529,
      "learning_rate": 9.99485432922021e-06,
      "loss": 0.2334,
      "step": 8235
    },
    {
      "epoch": 5.45430463576159,
      "grad_norm": 0.8040562507015419,
      "learning_rate": 9.97091098031783e-06,
      "loss": 0.249,
      "step": 8236
    },
    {
      "epoch": 5.454966887417219,
      "grad_norm": 0.7575597080649262,
      "learning_rate": 9.946995358780474e-06,
      "loss": 0.2314,
      "step": 8237
    },
    {
      "epoch": 5.4556291390728475,
      "grad_norm": 0.718341691814952,
      "learning_rate": 9.923107469343683e-06,
      "loss": 0.249,
      "step": 8238
    },
    {
      "epoch": 5.4562913907284765,
      "grad_norm": 0.6796712610276961,
      "learning_rate": 9.899247316737557e-06,
      "loss": 0.2197,
      "step": 8239
    },
    {
      "epoch": 5.456953642384106,
      "grad_norm": 0.8364442650831285,
      "learning_rate": 9.875414905686652e-06,
      "loss": 0.2969,
      "step": 8240
    },
    {
      "epoch": 5.457615894039735,
      "grad_norm": 0.6750954248636425,
      "learning_rate": 9.851610240910058e-06,
      "loss": 0.207,
      "step": 8241
    },
    {
      "epoch": 5.458278145695364,
      "grad_norm": 0.6982925630966095,
      "learning_rate": 9.82783332712142e-06,
      "loss": 0.2324,
      "step": 8242
    },
    {
      "epoch": 5.458940397350993,
      "grad_norm": 0.7567891967997111,
      "learning_rate": 9.804084169028792e-06,
      "loss": 0.2617,
      "step": 8243
    },
    {
      "epoch": 5.459602649006623,
      "grad_norm": 0.9282080768713186,
      "learning_rate": 9.78036277133476e-06,
      "loss": 0.3555,
      "step": 8244
    },
    {
      "epoch": 5.460264900662252,
      "grad_norm": 0.698834967713629,
      "learning_rate": 9.7566691387365e-06,
      "loss": 0.2148,
      "step": 8245
    },
    {
      "epoch": 5.460927152317881,
      "grad_norm": 0.7101774292543698,
      "learning_rate": 9.733003275925594e-06,
      "loss": 0.2275,
      "step": 8246
    },
    {
      "epoch": 5.46158940397351,
      "grad_norm": 0.7805355081769234,
      "learning_rate": 9.709365187588124e-06,
      "loss": 0.252,
      "step": 8247
    },
    {
      "epoch": 5.462251655629139,
      "grad_norm": 0.8274963832791492,
      "learning_rate": 9.685754878404717e-06,
      "loss": 0.2832,
      "step": 8248
    },
    {
      "epoch": 5.4629139072847686,
      "grad_norm": 0.764600385943334,
      "learning_rate": 9.66217235305053e-06,
      "loss": 0.2676,
      "step": 8249
    },
    {
      "epoch": 5.4635761589403975,
      "grad_norm": 0.7636334168787373,
      "learning_rate": 9.63861761619511e-06,
      "loss": 0.2236,
      "step": 8250
    },
    {
      "epoch": 5.464238410596026,
      "grad_norm": 0.9644969940582598,
      "learning_rate": 9.615090672502613e-06,
      "loss": 0.3262,
      "step": 8251
    },
    {
      "epoch": 5.464900662251655,
      "grad_norm": 0.6995941733866757,
      "learning_rate": 9.591591526631608e-06,
      "loss": 0.2236,
      "step": 8252
    },
    {
      "epoch": 5.465562913907284,
      "grad_norm": 0.7917754617393823,
      "learning_rate": 9.568120183235211e-06,
      "loss": 0.3047,
      "step": 8253
    },
    {
      "epoch": 5.466225165562914,
      "grad_norm": 0.6729855738696939,
      "learning_rate": 9.544676646961036e-06,
      "loss": 0.208,
      "step": 8254
    },
    {
      "epoch": 5.466887417218543,
      "grad_norm": 0.7366912523378247,
      "learning_rate": 9.521260922451152e-06,
      "loss": 0.2197,
      "step": 8255
    },
    {
      "epoch": 5.467549668874172,
      "grad_norm": 0.7243111905019703,
      "learning_rate": 9.497873014342116e-06,
      "loss": 0.21,
      "step": 8256
    },
    {
      "epoch": 5.468211920529801,
      "grad_norm": 0.7604047882262971,
      "learning_rate": 9.474512927265026e-06,
      "loss": 0.2754,
      "step": 8257
    },
    {
      "epoch": 5.468874172185431,
      "grad_norm": 0.7472783503399074,
      "learning_rate": 9.451180665845482e-06,
      "loss": 0.252,
      "step": 8258
    },
    {
      "epoch": 5.46953642384106,
      "grad_norm": 0.6986079764339138,
      "learning_rate": 9.42787623470347e-06,
      "loss": 0.2295,
      "step": 8259
    },
    {
      "epoch": 5.470198675496689,
      "grad_norm": 0.7558842667525061,
      "learning_rate": 9.404599638453597e-06,
      "loss": 0.2305,
      "step": 8260
    },
    {
      "epoch": 5.470860927152318,
      "grad_norm": 0.8514022457802254,
      "learning_rate": 9.381350881704863e-06,
      "loss": 0.3281,
      "step": 8261
    },
    {
      "epoch": 5.4715231788079475,
      "grad_norm": 0.8199458953577055,
      "learning_rate": 9.358129969060768e-06,
      "loss": 0.2559,
      "step": 8262
    },
    {
      "epoch": 5.472185430463576,
      "grad_norm": 0.7727945978518875,
      "learning_rate": 9.334936905119383e-06,
      "loss": 0.248,
      "step": 8263
    },
    {
      "epoch": 5.472847682119205,
      "grad_norm": 0.6898482664036545,
      "learning_rate": 9.311771694473169e-06,
      "loss": 0.2285,
      "step": 8264
    },
    {
      "epoch": 5.473509933774834,
      "grad_norm": 0.7728471369233042,
      "learning_rate": 9.28863434170904e-06,
      "loss": 0.2324,
      "step": 8265
    },
    {
      "epoch": 5.474172185430463,
      "grad_norm": 0.6800552107777947,
      "learning_rate": 9.265524851408562e-06,
      "loss": 0.2236,
      "step": 8266
    },
    {
      "epoch": 5.474834437086093,
      "grad_norm": 0.6416980743440165,
      "learning_rate": 9.242443228147627e-06,
      "loss": 0.1855,
      "step": 8267
    },
    {
      "epoch": 5.475496688741722,
      "grad_norm": 0.7395458056646712,
      "learning_rate": 9.219389476496663e-06,
      "loss": 0.2656,
      "step": 8268
    },
    {
      "epoch": 5.476158940397351,
      "grad_norm": 0.7252764856930516,
      "learning_rate": 9.196363601020567e-06,
      "loss": 0.2354,
      "step": 8269
    },
    {
      "epoch": 5.47682119205298,
      "grad_norm": 0.7981120496073488,
      "learning_rate": 9.17336560627876e-06,
      "loss": 0.2275,
      "step": 8270
    },
    {
      "epoch": 5.477483443708609,
      "grad_norm": 0.6849213180202682,
      "learning_rate": 9.150395496825064e-06,
      "loss": 0.2178,
      "step": 8271
    },
    {
      "epoch": 5.478145695364239,
      "grad_norm": 0.7742025613850498,
      "learning_rate": 9.127453277207875e-06,
      "loss": 0.25,
      "step": 8272
    },
    {
      "epoch": 5.478807947019868,
      "grad_norm": 0.8014910574622097,
      "learning_rate": 9.104538951969958e-06,
      "loss": 0.2852,
      "step": 8273
    },
    {
      "epoch": 5.479470198675497,
      "grad_norm": 0.8294948101863496,
      "learning_rate": 9.081652525648652e-06,
      "loss": 0.2832,
      "step": 8274
    },
    {
      "epoch": 5.4801324503311255,
      "grad_norm": 0.8209309893961193,
      "learning_rate": 9.058794002775733e-06,
      "loss": 0.3027,
      "step": 8275
    },
    {
      "epoch": 5.480794701986755,
      "grad_norm": 1.1535190935514115,
      "learning_rate": 9.035963387877442e-06,
      "loss": 0.4082,
      "step": 8276
    },
    {
      "epoch": 5.481456953642384,
      "grad_norm": 0.7032787783811938,
      "learning_rate": 9.013160685474485e-06,
      "loss": 0.2285,
      "step": 8277
    },
    {
      "epoch": 5.482119205298013,
      "grad_norm": 0.7839174397945688,
      "learning_rate": 8.990385900082064e-06,
      "loss": 0.2598,
      "step": 8278
    },
    {
      "epoch": 5.482781456953642,
      "grad_norm": 0.706368724884191,
      "learning_rate": 8.96763903620989e-06,
      "loss": 0.2363,
      "step": 8279
    },
    {
      "epoch": 5.483443708609272,
      "grad_norm": 0.7127085294222835,
      "learning_rate": 8.944920098362041e-06,
      "loss": 0.2354,
      "step": 8280
    },
    {
      "epoch": 5.484105960264901,
      "grad_norm": 0.6614521180400365,
      "learning_rate": 8.92222909103717e-06,
      "loss": 0.2129,
      "step": 8281
    },
    {
      "epoch": 5.48476821192053,
      "grad_norm": 0.781839744938468,
      "learning_rate": 8.89956601872835e-06,
      "loss": 0.2891,
      "step": 8282
    },
    {
      "epoch": 5.485430463576159,
      "grad_norm": 0.7650277271382648,
      "learning_rate": 8.876930885923106e-06,
      "loss": 0.2422,
      "step": 8283
    },
    {
      "epoch": 5.486092715231788,
      "grad_norm": 0.759510725817291,
      "learning_rate": 8.854323697103488e-06,
      "loss": 0.2451,
      "step": 8284
    },
    {
      "epoch": 5.486754966887418,
      "grad_norm": 0.8120853996743709,
      "learning_rate": 8.831744456745948e-06,
      "loss": 0.2539,
      "step": 8285
    },
    {
      "epoch": 5.4874172185430465,
      "grad_norm": 0.8713494720344912,
      "learning_rate": 8.809193169321427e-06,
      "loss": 0.2578,
      "step": 8286
    },
    {
      "epoch": 5.4880794701986755,
      "grad_norm": 0.7227020010770573,
      "learning_rate": 8.786669839295402e-06,
      "loss": 0.2178,
      "step": 8287
    },
    {
      "epoch": 5.488741721854304,
      "grad_norm": 0.7542463220051343,
      "learning_rate": 8.764174471127688e-06,
      "loss": 0.2334,
      "step": 8288
    },
    {
      "epoch": 5.489403973509933,
      "grad_norm": 0.7999487423582928,
      "learning_rate": 8.74170706927264e-06,
      "loss": 0.2354,
      "step": 8289
    },
    {
      "epoch": 5.490066225165563,
      "grad_norm": 0.7383005138056044,
      "learning_rate": 8.71926763817905e-06,
      "loss": 0.2402,
      "step": 8290
    },
    {
      "epoch": 5.490728476821192,
      "grad_norm": 0.7455341099804799,
      "learning_rate": 8.69685618229023e-06,
      "loss": 0.2559,
      "step": 8291
    },
    {
      "epoch": 5.491390728476821,
      "grad_norm": 0.7420239301405779,
      "learning_rate": 8.674472706043878e-06,
      "loss": 0.2178,
      "step": 8292
    },
    {
      "epoch": 5.49205298013245,
      "grad_norm": 0.7148609906004662,
      "learning_rate": 8.652117213872151e-06,
      "loss": 0.2109,
      "step": 8293
    },
    {
      "epoch": 5.49271523178808,
      "grad_norm": 0.7844358463326505,
      "learning_rate": 8.629789710201707e-06,
      "loss": 0.249,
      "step": 8294
    },
    {
      "epoch": 5.493377483443709,
      "grad_norm": 0.6962563409073436,
      "learning_rate": 8.607490199453677e-06,
      "loss": 0.2461,
      "step": 8295
    },
    {
      "epoch": 5.494039735099338,
      "grad_norm": 0.806374167167584,
      "learning_rate": 8.585218686043593e-06,
      "loss": 0.2637,
      "step": 8296
    },
    {
      "epoch": 5.494701986754967,
      "grad_norm": 0.749256973432726,
      "learning_rate": 8.56297517438148e-06,
      "loss": 0.2617,
      "step": 8297
    },
    {
      "epoch": 5.495364238410596,
      "grad_norm": 0.7876949175835746,
      "learning_rate": 8.540759668871799e-06,
      "loss": 0.252,
      "step": 8298
    },
    {
      "epoch": 5.4960264900662255,
      "grad_norm": 0.7087124789169329,
      "learning_rate": 8.518572173913479e-06,
      "loss": 0.207,
      "step": 8299
    },
    {
      "epoch": 5.496688741721854,
      "grad_norm": 0.7974959260157949,
      "learning_rate": 8.496412693899907e-06,
      "loss": 0.2949,
      "step": 8300
    },
    {
      "epoch": 5.497350993377483,
      "grad_norm": 0.6467321618308874,
      "learning_rate": 8.474281233218922e-06,
      "loss": 0.1748,
      "step": 8301
    },
    {
      "epoch": 5.498013245033112,
      "grad_norm": 0.7707040606716175,
      "learning_rate": 8.452177796252752e-06,
      "loss": 0.2715,
      "step": 8302
    },
    {
      "epoch": 5.498675496688742,
      "grad_norm": 0.7919071808346391,
      "learning_rate": 8.430102387378195e-06,
      "loss": 0.2812,
      "step": 8303
    },
    {
      "epoch": 5.499337748344371,
      "grad_norm": 0.8710522382639634,
      "learning_rate": 8.408055010966424e-06,
      "loss": 0.3008,
      "step": 8304
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.7726653714638038,
      "learning_rate": 8.386035671383045e-06,
      "loss": 0.2207,
      "step": 8305
    },
    {
      "epoch": 5.500662251655629,
      "grad_norm": 0.8368796364761532,
      "learning_rate": 8.364044372988155e-06,
      "loss": 0.2471,
      "step": 8306
    },
    {
      "epoch": 5.501324503311258,
      "grad_norm": 0.6616917380141908,
      "learning_rate": 8.342081120136284e-06,
      "loss": 0.1865,
      "step": 8307
    },
    {
      "epoch": 5.501986754966888,
      "grad_norm": 0.7452066452415348,
      "learning_rate": 8.320145917176408e-06,
      "loss": 0.2266,
      "step": 8308
    },
    {
      "epoch": 5.502649006622517,
      "grad_norm": 0.7257236661562294,
      "learning_rate": 8.298238768451948e-06,
      "loss": 0.2441,
      "step": 8309
    },
    {
      "epoch": 5.503311258278146,
      "grad_norm": 0.8034540985697016,
      "learning_rate": 8.276359678300754e-06,
      "loss": 0.2559,
      "step": 8310
    },
    {
      "epoch": 5.5039735099337745,
      "grad_norm": 0.6775550839967807,
      "learning_rate": 8.254508651055158e-06,
      "loss": 0.2129,
      "step": 8311
    },
    {
      "epoch": 5.5046357615894035,
      "grad_norm": 0.825451786072652,
      "learning_rate": 8.232685691041936e-06,
      "loss": 0.2793,
      "step": 8312
    },
    {
      "epoch": 5.505298013245033,
      "grad_norm": 0.7215712011001146,
      "learning_rate": 8.210890802582243e-06,
      "loss": 0.2354,
      "step": 8313
    },
    {
      "epoch": 5.505960264900662,
      "grad_norm": 0.7269564301420719,
      "learning_rate": 8.189123989991714e-06,
      "loss": 0.2227,
      "step": 8314
    },
    {
      "epoch": 5.506622516556291,
      "grad_norm": 0.7623588933700701,
      "learning_rate": 8.167385257580434e-06,
      "loss": 0.2422,
      "step": 8315
    },
    {
      "epoch": 5.50728476821192,
      "grad_norm": 0.8612237448598782,
      "learning_rate": 8.145674609652975e-06,
      "loss": 0.2773,
      "step": 8316
    },
    {
      "epoch": 5.50794701986755,
      "grad_norm": 0.8531345885322131,
      "learning_rate": 8.123992050508216e-06,
      "loss": 0.3262,
      "step": 8317
    },
    {
      "epoch": 5.508609271523179,
      "grad_norm": 0.6212988854206485,
      "learning_rate": 8.102337584439606e-06,
      "loss": 0.1855,
      "step": 8318
    },
    {
      "epoch": 5.509271523178808,
      "grad_norm": 0.7727103218675102,
      "learning_rate": 8.08071121573493e-06,
      "loss": 0.249,
      "step": 8319
    },
    {
      "epoch": 5.509933774834437,
      "grad_norm": 0.7302727176373264,
      "learning_rate": 8.059112948676482e-06,
      "loss": 0.2598,
      "step": 8320
    },
    {
      "epoch": 5.510596026490067,
      "grad_norm": 0.8979670971514583,
      "learning_rate": 8.037542787541008e-06,
      "loss": 0.3535,
      "step": 8321
    },
    {
      "epoch": 5.511258278145696,
      "grad_norm": 0.7491866413513846,
      "learning_rate": 8.016000736599588e-06,
      "loss": 0.2246,
      "step": 8322
    },
    {
      "epoch": 5.5119205298013245,
      "grad_norm": 0.8398765613626522,
      "learning_rate": 7.994486800117794e-06,
      "loss": 0.2988,
      "step": 8323
    },
    {
      "epoch": 5.5125827814569535,
      "grad_norm": 0.853552413315536,
      "learning_rate": 7.973000982355688e-06,
      "loss": 0.3047,
      "step": 8324
    },
    {
      "epoch": 5.513245033112582,
      "grad_norm": 0.7051913954437126,
      "learning_rate": 7.95154328756768e-06,
      "loss": 0.2227,
      "step": 8325
    },
    {
      "epoch": 5.513907284768212,
      "grad_norm": 0.6407479685054761,
      "learning_rate": 7.930113720002606e-06,
      "loss": 0.2021,
      "step": 8326
    },
    {
      "epoch": 5.514569536423841,
      "grad_norm": 0.773330342520804,
      "learning_rate": 7.90871228390379e-06,
      "loss": 0.252,
      "step": 8327
    },
    {
      "epoch": 5.51523178807947,
      "grad_norm": 0.8133827483817562,
      "learning_rate": 7.88733898350899e-06,
      "loss": 0.3086,
      "step": 8328
    },
    {
      "epoch": 5.515894039735099,
      "grad_norm": 0.8515957072000154,
      "learning_rate": 7.86599382305032e-06,
      "loss": 0.2832,
      "step": 8329
    },
    {
      "epoch": 5.516556291390728,
      "grad_norm": 0.8330364555146641,
      "learning_rate": 7.844676806754419e-06,
      "loss": 0.3125,
      "step": 8330
    },
    {
      "epoch": 5.517218543046358,
      "grad_norm": 0.8028675040663321,
      "learning_rate": 7.823387938842223e-06,
      "loss": 0.2715,
      "step": 8331
    },
    {
      "epoch": 5.517880794701987,
      "grad_norm": 0.6961876344455217,
      "learning_rate": 7.802127223529232e-06,
      "loss": 0.2041,
      "step": 8332
    },
    {
      "epoch": 5.518543046357616,
      "grad_norm": 0.7365859303159131,
      "learning_rate": 7.780894665025305e-06,
      "loss": 0.2168,
      "step": 8333
    },
    {
      "epoch": 5.519205298013245,
      "grad_norm": 0.7241149163549856,
      "learning_rate": 7.759690267534735e-06,
      "loss": 0.2373,
      "step": 8334
    },
    {
      "epoch": 5.5198675496688745,
      "grad_norm": 0.7949174588562877,
      "learning_rate": 7.738514035256177e-06,
      "loss": 0.2373,
      "step": 8335
    },
    {
      "epoch": 5.520529801324503,
      "grad_norm": 0.7526697155621349,
      "learning_rate": 7.71736597238286e-06,
      "loss": 0.249,
      "step": 8336
    },
    {
      "epoch": 5.521192052980132,
      "grad_norm": 0.648639742627535,
      "learning_rate": 7.696246083102303e-06,
      "loss": 0.2002,
      "step": 8337
    },
    {
      "epoch": 5.521854304635761,
      "grad_norm": 0.8102700019392715,
      "learning_rate": 7.675154371596442e-06,
      "loss": 0.2598,
      "step": 8338
    },
    {
      "epoch": 5.522516556291391,
      "grad_norm": 0.6620507102770588,
      "learning_rate": 7.654090842041737e-06,
      "loss": 0.1963,
      "step": 8339
    },
    {
      "epoch": 5.52317880794702,
      "grad_norm": 0.7161798959685081,
      "learning_rate": 7.633055498609004e-06,
      "loss": 0.2158,
      "step": 8340
    },
    {
      "epoch": 5.523841059602649,
      "grad_norm": 0.8038250774594444,
      "learning_rate": 7.612048345463439e-06,
      "loss": 0.2715,
      "step": 8341
    },
    {
      "epoch": 5.524503311258278,
      "grad_norm": 0.869101589851218,
      "learning_rate": 7.591069386764737e-06,
      "loss": 0.2734,
      "step": 8342
    },
    {
      "epoch": 5.525165562913907,
      "grad_norm": 0.6908105298538103,
      "learning_rate": 7.570118626666955e-06,
      "loss": 0.2158,
      "step": 8343
    },
    {
      "epoch": 5.525827814569537,
      "grad_norm": 0.8165604776930756,
      "learning_rate": 7.549196069318592e-06,
      "loss": 0.2734,
      "step": 8344
    },
    {
      "epoch": 5.526490066225166,
      "grad_norm": 0.7421269720817115,
      "learning_rate": 7.528301718862567e-06,
      "loss": 0.252,
      "step": 8345
    },
    {
      "epoch": 5.527152317880795,
      "grad_norm": 0.8052111551035224,
      "learning_rate": 7.507435579436189e-06,
      "loss": 0.252,
      "step": 8346
    },
    {
      "epoch": 5.527814569536424,
      "grad_norm": 0.6982969874923907,
      "learning_rate": 7.486597655171167e-06,
      "loss": 0.2275,
      "step": 8347
    },
    {
      "epoch": 5.5284768211920525,
      "grad_norm": 0.797358972529865,
      "learning_rate": 7.465787950193669e-06,
      "loss": 0.25,
      "step": 8348
    },
    {
      "epoch": 5.529139072847682,
      "grad_norm": 0.8572471956592532,
      "learning_rate": 7.445006468624282e-06,
      "loss": 0.3047,
      "step": 8349
    },
    {
      "epoch": 5.529801324503311,
      "grad_norm": 0.7365344931306134,
      "learning_rate": 7.424253214577963e-06,
      "loss": 0.2471,
      "step": 8350
    },
    {
      "epoch": 5.53046357615894,
      "grad_norm": 0.7437903744234947,
      "learning_rate": 7.40352819216406e-06,
      "loss": 0.2383,
      "step": 8351
    },
    {
      "epoch": 5.531125827814569,
      "grad_norm": 0.7695939616483454,
      "learning_rate": 7.382831405486389e-06,
      "loss": 0.2715,
      "step": 8352
    },
    {
      "epoch": 5.531788079470199,
      "grad_norm": 0.7950244074580723,
      "learning_rate": 7.362162858643156e-06,
      "loss": 0.2656,
      "step": 8353
    },
    {
      "epoch": 5.532450331125828,
      "grad_norm": 0.6624313644233896,
      "learning_rate": 7.34152255572697e-06,
      "loss": 0.1934,
      "step": 8354
    },
    {
      "epoch": 5.533112582781457,
      "grad_norm": 0.780890017609156,
      "learning_rate": 7.320910500824861e-06,
      "loss": 0.2305,
      "step": 8355
    },
    {
      "epoch": 5.533774834437086,
      "grad_norm": 0.6447238574903741,
      "learning_rate": 7.3003266980182e-06,
      "loss": 0.1992,
      "step": 8356
    },
    {
      "epoch": 5.534437086092716,
      "grad_norm": 0.814351305047172,
      "learning_rate": 7.2797711513828574e-06,
      "loss": 0.2402,
      "step": 8357
    },
    {
      "epoch": 5.535099337748345,
      "grad_norm": 0.7968039095791204,
      "learning_rate": 7.2592438649890945e-06,
      "loss": 0.2754,
      "step": 8358
    },
    {
      "epoch": 5.535761589403974,
      "grad_norm": 0.9427827728803346,
      "learning_rate": 7.238744842901478e-06,
      "loss": 0.2617,
      "step": 8359
    },
    {
      "epoch": 5.5364238410596025,
      "grad_norm": 0.787436588649743,
      "learning_rate": 7.218274089179094e-06,
      "loss": 0.2617,
      "step": 8360
    },
    {
      "epoch": 5.5370860927152314,
      "grad_norm": 0.8051572475828139,
      "learning_rate": 7.197831607875398e-06,
      "loss": 0.248,
      "step": 8361
    },
    {
      "epoch": 5.537748344370861,
      "grad_norm": 0.7091971882525859,
      "learning_rate": 7.177417403038221e-06,
      "loss": 0.2207,
      "step": 8362
    },
    {
      "epoch": 5.53841059602649,
      "grad_norm": 0.6777760101457878,
      "learning_rate": 7.157031478709779e-06,
      "loss": 0.2207,
      "step": 8363
    },
    {
      "epoch": 5.539072847682119,
      "grad_norm": 0.6780213178088493,
      "learning_rate": 7.13667383892676e-06,
      "loss": 0.2031,
      "step": 8364
    },
    {
      "epoch": 5.539735099337748,
      "grad_norm": 0.8566820225082039,
      "learning_rate": 7.116344487720205e-06,
      "loss": 0.3086,
      "step": 8365
    },
    {
      "epoch": 5.540397350993377,
      "grad_norm": 0.7164370790299112,
      "learning_rate": 7.096043429115528e-06,
      "loss": 0.2148,
      "step": 8366
    },
    {
      "epoch": 5.541059602649007,
      "grad_norm": 0.7029028451645084,
      "learning_rate": 7.075770667132613e-06,
      "loss": 0.2217,
      "step": 8367
    },
    {
      "epoch": 5.541721854304636,
      "grad_norm": 0.719559595144086,
      "learning_rate": 7.05552620578565e-06,
      "loss": 0.2285,
      "step": 8368
    },
    {
      "epoch": 5.542384105960265,
      "grad_norm": 0.6602198377859929,
      "learning_rate": 7.035310049083315e-06,
      "loss": 0.2188,
      "step": 8369
    },
    {
      "epoch": 5.543046357615894,
      "grad_norm": 0.748166217083447,
      "learning_rate": 7.0151222010286555e-06,
      "loss": 0.2734,
      "step": 8370
    },
    {
      "epoch": 5.5437086092715235,
      "grad_norm": 0.6793001286682699,
      "learning_rate": 6.994962665619059e-06,
      "loss": 0.2129,
      "step": 8371
    },
    {
      "epoch": 5.5443708609271525,
      "grad_norm": 0.7337748312306621,
      "learning_rate": 6.9748314468463295e-06,
      "loss": 0.2354,
      "step": 8372
    },
    {
      "epoch": 5.545033112582781,
      "grad_norm": 0.8885860920839272,
      "learning_rate": 6.954728548696698e-06,
      "loss": 0.3262,
      "step": 8373
    },
    {
      "epoch": 5.54569536423841,
      "grad_norm": 0.7650343173612224,
      "learning_rate": 6.934653975150811e-06,
      "loss": 0.25,
      "step": 8374
    },
    {
      "epoch": 5.54635761589404,
      "grad_norm": 0.7312496296099791,
      "learning_rate": 6.9146077301835916e-06,
      "loss": 0.2451,
      "step": 8375
    },
    {
      "epoch": 5.547019867549669,
      "grad_norm": 0.7947847552281614,
      "learning_rate": 6.894589817764478e-06,
      "loss": 0.2451,
      "step": 8376
    },
    {
      "epoch": 5.547682119205298,
      "grad_norm": 0.7900757998657023,
      "learning_rate": 6.874600241857237e-06,
      "loss": 0.252,
      "step": 8377
    },
    {
      "epoch": 5.548344370860927,
      "grad_norm": 0.684882118600078,
      "learning_rate": 6.8546390064200155e-06,
      "loss": 0.2256,
      "step": 8378
    },
    {
      "epoch": 5.549006622516556,
      "grad_norm": 0.8190141657866639,
      "learning_rate": 6.834706115405386e-06,
      "loss": 0.2754,
      "step": 8379
    },
    {
      "epoch": 5.549668874172186,
      "grad_norm": 0.7459821985111394,
      "learning_rate": 6.814801572760275e-06,
      "loss": 0.2314,
      "step": 8380
    },
    {
      "epoch": 5.550331125827815,
      "grad_norm": 0.789873779128214,
      "learning_rate": 6.794925382426014e-06,
      "loss": 0.2285,
      "step": 8381
    },
    {
      "epoch": 5.550993377483444,
      "grad_norm": 0.8127378337742824,
      "learning_rate": 6.775077548338353e-06,
      "loss": 0.25,
      "step": 8382
    },
    {
      "epoch": 5.551655629139073,
      "grad_norm": 0.6492678219688907,
      "learning_rate": 6.755258074427333e-06,
      "loss": 0.1943,
      "step": 8383
    },
    {
      "epoch": 5.552317880794702,
      "grad_norm": 0.7293835109097302,
      "learning_rate": 6.735466964617464e-06,
      "loss": 0.2285,
      "step": 8384
    },
    {
      "epoch": 5.552980132450331,
      "grad_norm": 0.7363112026655881,
      "learning_rate": 6.715704222827595e-06,
      "loss": 0.2539,
      "step": 8385
    },
    {
      "epoch": 5.55364238410596,
      "grad_norm": 0.8300986547597354,
      "learning_rate": 6.695969852971012e-06,
      "loss": 0.252,
      "step": 8386
    },
    {
      "epoch": 5.554304635761589,
      "grad_norm": 0.829444671322198,
      "learning_rate": 6.676263858955305e-06,
      "loss": 0.2432,
      "step": 8387
    },
    {
      "epoch": 5.554966887417218,
      "grad_norm": 0.778051796223688,
      "learning_rate": 6.656586244682538e-06,
      "loss": 0.2598,
      "step": 8388
    },
    {
      "epoch": 5.555629139072848,
      "grad_norm": 0.7601144399086871,
      "learning_rate": 6.636937014049043e-06,
      "loss": 0.2539,
      "step": 8389
    },
    {
      "epoch": 5.556291390728477,
      "grad_norm": 0.806625963453341,
      "learning_rate": 6.617316170945624e-06,
      "loss": 0.2832,
      "step": 8390
    },
    {
      "epoch": 5.556953642384106,
      "grad_norm": 0.8344953961753315,
      "learning_rate": 6.597723719257459e-06,
      "loss": 0.2969,
      "step": 8391
    },
    {
      "epoch": 5.557615894039735,
      "grad_norm": 0.7445752009619905,
      "learning_rate": 6.578159662864041e-06,
      "loss": 0.252,
      "step": 8392
    },
    {
      "epoch": 5.558278145695365,
      "grad_norm": 0.6890255025252918,
      "learning_rate": 6.5586240056392565e-06,
      "loss": 0.2256,
      "step": 8393
    },
    {
      "epoch": 5.558940397350994,
      "grad_norm": 0.8308510965175706,
      "learning_rate": 6.539116751451462e-06,
      "loss": 0.2676,
      "step": 8394
    },
    {
      "epoch": 5.559602649006623,
      "grad_norm": 0.8203633569264653,
      "learning_rate": 6.519637904163266e-06,
      "loss": 0.2676,
      "step": 8395
    },
    {
      "epoch": 5.5602649006622515,
      "grad_norm": 0.8050032967942341,
      "learning_rate": 6.5001874676317015e-06,
      "loss": 0.252,
      "step": 8396
    },
    {
      "epoch": 5.5609271523178805,
      "grad_norm": 0.7448345502268655,
      "learning_rate": 6.480765445708186e-06,
      "loss": 0.2334,
      "step": 8397
    },
    {
      "epoch": 5.56158940397351,
      "grad_norm": 0.7576869613315572,
      "learning_rate": 6.461371842238528e-06,
      "loss": 0.2832,
      "step": 8398
    },
    {
      "epoch": 5.562251655629139,
      "grad_norm": 0.8997034901971405,
      "learning_rate": 6.442006661062837e-06,
      "loss": 0.2852,
      "step": 8399
    },
    {
      "epoch": 5.562913907284768,
      "grad_norm": 0.7859713816157559,
      "learning_rate": 6.422669906015698e-06,
      "loss": 0.252,
      "step": 8400
    },
    {
      "epoch": 5.563576158940397,
      "grad_norm": 0.7912667782919214,
      "learning_rate": 6.403361580925947e-06,
      "loss": 0.2578,
      "step": 8401
    },
    {
      "epoch": 5.564238410596026,
      "grad_norm": 0.6832060697119714,
      "learning_rate": 6.384081689616894e-06,
      "loss": 0.2217,
      "step": 8402
    },
    {
      "epoch": 5.564900662251656,
      "grad_norm": 0.6991137397027128,
      "learning_rate": 6.364830235906182e-06,
      "loss": 0.2412,
      "step": 8403
    },
    {
      "epoch": 5.565562913907285,
      "grad_norm": 0.761468316267935,
      "learning_rate": 6.345607223605814e-06,
      "loss": 0.2305,
      "step": 8404
    },
    {
      "epoch": 5.566225165562914,
      "grad_norm": 0.6916545574548287,
      "learning_rate": 6.326412656522128e-06,
      "loss": 0.2158,
      "step": 8405
    },
    {
      "epoch": 5.566887417218543,
      "grad_norm": 0.74579428230351,
      "learning_rate": 6.3072465384559015e-06,
      "loss": 0.252,
      "step": 8406
    },
    {
      "epoch": 5.567549668874172,
      "grad_norm": 0.7823247587881895,
      "learning_rate": 6.28810887320228e-06,
      "loss": 0.2695,
      "step": 8407
    },
    {
      "epoch": 5.5682119205298015,
      "grad_norm": 0.7198752351127756,
      "learning_rate": 6.268999664550683e-06,
      "loss": 0.21,
      "step": 8408
    },
    {
      "epoch": 5.5688741721854305,
      "grad_norm": 0.7547697319321347,
      "learning_rate": 6.249918916284969e-06,
      "loss": 0.2305,
      "step": 8409
    },
    {
      "epoch": 5.569536423841059,
      "grad_norm": 0.71998029381535,
      "learning_rate": 6.230866632183345e-06,
      "loss": 0.2275,
      "step": 8410
    },
    {
      "epoch": 5.570198675496689,
      "grad_norm": 0.633533752953935,
      "learning_rate": 6.211842816018414e-06,
      "loss": 0.1982,
      "step": 8411
    },
    {
      "epoch": 5.570860927152318,
      "grad_norm": 0.7704261637844638,
      "learning_rate": 6.192847471557045e-06,
      "loss": 0.2344,
      "step": 8412
    },
    {
      "epoch": 5.571523178807947,
      "grad_norm": 0.7463619405142556,
      "learning_rate": 6.1738806025606105e-06,
      "loss": 0.2559,
      "step": 8413
    },
    {
      "epoch": 5.572185430463576,
      "grad_norm": 0.8024549232017715,
      "learning_rate": 6.154942212784708e-06,
      "loss": 0.2314,
      "step": 8414
    },
    {
      "epoch": 5.572847682119205,
      "grad_norm": 0.6514501930914766,
      "learning_rate": 6.136032305979372e-06,
      "loss": 0.2236,
      "step": 8415
    },
    {
      "epoch": 5.573509933774835,
      "grad_norm": 0.8206122133660599,
      "learning_rate": 6.117150885889005e-06,
      "loss": 0.2393,
      "step": 8416
    },
    {
      "epoch": 5.574172185430464,
      "grad_norm": 0.6286922233077189,
      "learning_rate": 6.098297956252302e-06,
      "loss": 0.1943,
      "step": 8417
    },
    {
      "epoch": 5.574834437086093,
      "grad_norm": 0.8595431519278207,
      "learning_rate": 6.079473520802391e-06,
      "loss": 0.3008,
      "step": 8418
    },
    {
      "epoch": 5.575496688741722,
      "grad_norm": 0.7776277998643802,
      "learning_rate": 6.060677583266726e-06,
      "loss": 0.25,
      "step": 8419
    },
    {
      "epoch": 5.576158940397351,
      "grad_norm": 0.7919388692558922,
      "learning_rate": 6.04191014736713e-06,
      "loss": 0.2432,
      "step": 8420
    },
    {
      "epoch": 5.57682119205298,
      "grad_norm": 0.6910393287704936,
      "learning_rate": 6.023171216819711e-06,
      "loss": 0.2354,
      "step": 8421
    },
    {
      "epoch": 5.577483443708609,
      "grad_norm": 0.7811169638029272,
      "learning_rate": 6.004460795335036e-06,
      "loss": 0.2383,
      "step": 8422
    },
    {
      "epoch": 5.578145695364238,
      "grad_norm": 0.7506580018110335,
      "learning_rate": 5.985778886618009e-06,
      "loss": 0.2148,
      "step": 8423
    },
    {
      "epoch": 5.578807947019867,
      "grad_norm": 0.7897003599137093,
      "learning_rate": 5.967125494367802e-06,
      "loss": 0.25,
      "step": 8424
    },
    {
      "epoch": 5.579470198675496,
      "grad_norm": 0.7291403893480034,
      "learning_rate": 5.948500622278046e-06,
      "loss": 0.2188,
      "step": 8425
    },
    {
      "epoch": 5.580132450331126,
      "grad_norm": 0.8168084017981146,
      "learning_rate": 5.929904274036657e-06,
      "loss": 0.2852,
      "step": 8426
    },
    {
      "epoch": 5.580794701986755,
      "grad_norm": 0.6928197176456917,
      "learning_rate": 5.911336453325921e-06,
      "loss": 0.2119,
      "step": 8427
    },
    {
      "epoch": 5.581456953642384,
      "grad_norm": 0.7117914697721293,
      "learning_rate": 5.892797163822516e-06,
      "loss": 0.2422,
      "step": 8428
    },
    {
      "epoch": 5.582119205298013,
      "grad_norm": 0.7752433603498773,
      "learning_rate": 5.874286409197404e-06,
      "loss": 0.2275,
      "step": 8429
    },
    {
      "epoch": 5.582781456953643,
      "grad_norm": 0.8223441062466899,
      "learning_rate": 5.855804193115904e-06,
      "loss": 0.2832,
      "step": 8430
    },
    {
      "epoch": 5.583443708609272,
      "grad_norm": 0.6980149777548678,
      "learning_rate": 5.8373505192377545e-06,
      "loss": 0.2354,
      "step": 8431
    },
    {
      "epoch": 5.584105960264901,
      "grad_norm": 0.7548319089763472,
      "learning_rate": 5.818925391216983e-06,
      "loss": 0.2539,
      "step": 8432
    },
    {
      "epoch": 5.5847682119205295,
      "grad_norm": 0.7703726783696607,
      "learning_rate": 5.800528812701938e-06,
      "loss": 0.2373,
      "step": 8433
    },
    {
      "epoch": 5.585430463576159,
      "grad_norm": 0.7254880123857514,
      "learning_rate": 5.782160787335388e-06,
      "loss": 0.2188,
      "step": 8434
    },
    {
      "epoch": 5.586092715231788,
      "grad_norm": 0.7705445269541704,
      "learning_rate": 5.7638213187544245e-06,
      "loss": 0.2539,
      "step": 8435
    },
    {
      "epoch": 5.586754966887417,
      "grad_norm": 0.5955668623920266,
      "learning_rate": 5.745510410590426e-06,
      "loss": 0.1748,
      "step": 8436
    },
    {
      "epoch": 5.587417218543046,
      "grad_norm": 0.7736918690466908,
      "learning_rate": 5.727228066469225e-06,
      "loss": 0.2559,
      "step": 8437
    },
    {
      "epoch": 5.588079470198675,
      "grad_norm": 0.7678887157863181,
      "learning_rate": 5.708974290010859e-06,
      "loss": 0.2754,
      "step": 8438
    },
    {
      "epoch": 5.588741721854305,
      "grad_norm": 0.6595611925196745,
      "learning_rate": 5.6907490848298544e-06,
      "loss": 0.1924,
      "step": 8439
    },
    {
      "epoch": 5.589403973509934,
      "grad_norm": 0.7199444960157584,
      "learning_rate": 5.6725524545349735e-06,
      "loss": 0.2061,
      "step": 8440
    },
    {
      "epoch": 5.590066225165563,
      "grad_norm": 0.8106593922681078,
      "learning_rate": 5.654384402729384e-06,
      "loss": 0.2969,
      "step": 8441
    },
    {
      "epoch": 5.590728476821192,
      "grad_norm": 0.8554274419406235,
      "learning_rate": 5.636244933010525e-06,
      "loss": 0.2949,
      "step": 8442
    },
    {
      "epoch": 5.591390728476821,
      "grad_norm": 0.7750320447759975,
      "learning_rate": 5.618134048970258e-06,
      "loss": 0.2451,
      "step": 8443
    },
    {
      "epoch": 5.592052980132451,
      "grad_norm": 0.7579028643633943,
      "learning_rate": 5.600051754194729e-06,
      "loss": 0.2383,
      "step": 8444
    },
    {
      "epoch": 5.5927152317880795,
      "grad_norm": 0.7843461395641105,
      "learning_rate": 5.581998052264441e-06,
      "loss": 0.2451,
      "step": 8445
    },
    {
      "epoch": 5.593377483443708,
      "grad_norm": 0.6375469834454234,
      "learning_rate": 5.563972946754253e-06,
      "loss": 0.1963,
      "step": 8446
    },
    {
      "epoch": 5.594039735099337,
      "grad_norm": 0.8267859142640496,
      "learning_rate": 5.545976441233324e-06,
      "loss": 0.248,
      "step": 8447
    },
    {
      "epoch": 5.594701986754967,
      "grad_norm": 0.742734296315138,
      "learning_rate": 5.528008539265155e-06,
      "loss": 0.2295,
      "step": 8448
    },
    {
      "epoch": 5.595364238410596,
      "grad_norm": 0.744074110937815,
      "learning_rate": 5.5100692444076336e-06,
      "loss": 0.2373,
      "step": 8449
    },
    {
      "epoch": 5.596026490066225,
      "grad_norm": 0.8012950600698244,
      "learning_rate": 5.492158560212917e-06,
      "loss": 0.2432,
      "step": 8450
    },
    {
      "epoch": 5.596688741721854,
      "grad_norm": 0.7367631048301474,
      "learning_rate": 5.474276490227502e-06,
      "loss": 0.2432,
      "step": 8451
    },
    {
      "epoch": 5.597350993377484,
      "grad_norm": 0.7721930189762625,
      "learning_rate": 5.456423037992308e-06,
      "loss": 0.2441,
      "step": 8452
    },
    {
      "epoch": 5.598013245033113,
      "grad_norm": 0.7238026352825647,
      "learning_rate": 5.438598207042505e-06,
      "loss": 0.2334,
      "step": 8453
    },
    {
      "epoch": 5.598675496688742,
      "grad_norm": 0.7310900653353348,
      "learning_rate": 5.420802000907571e-06,
      "loss": 0.2129,
      "step": 8454
    },
    {
      "epoch": 5.599337748344371,
      "grad_norm": 0.675129106333715,
      "learning_rate": 5.4030344231113855e-06,
      "loss": 0.1836,
      "step": 8455
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.725645843175105,
      "learning_rate": 5.385295477172152e-06,
      "loss": 0.25,
      "step": 8456
    },
    {
      "epoch": 5.6006622516556295,
      "grad_norm": 0.7635799158133673,
      "learning_rate": 5.367585166602361e-06,
      "loss": 0.2598,
      "step": 8457
    },
    {
      "epoch": 5.601324503311258,
      "grad_norm": 0.8167069447447451,
      "learning_rate": 5.349903494908875e-06,
      "loss": 0.2773,
      "step": 8458
    },
    {
      "epoch": 5.601986754966887,
      "grad_norm": 0.7453381338680775,
      "learning_rate": 5.332250465592825e-06,
      "loss": 0.2285,
      "step": 8459
    },
    {
      "epoch": 5.602649006622516,
      "grad_norm": 0.7679852408327323,
      "learning_rate": 5.314626082149764e-06,
      "loss": 0.2695,
      "step": 8460
    },
    {
      "epoch": 5.603311258278145,
      "grad_norm": 0.7437396224587862,
      "learning_rate": 5.297030348069503e-06,
      "loss": 0.2363,
      "step": 8461
    },
    {
      "epoch": 5.603973509933775,
      "grad_norm": 0.7561477499614427,
      "learning_rate": 5.27946326683622e-06,
      "loss": 0.2363,
      "step": 8462
    },
    {
      "epoch": 5.604635761589404,
      "grad_norm": 0.8040407439067624,
      "learning_rate": 5.261924841928333e-06,
      "loss": 0.2451,
      "step": 8463
    },
    {
      "epoch": 5.605298013245033,
      "grad_norm": 0.8287852295913165,
      "learning_rate": 5.244415076818714e-06,
      "loss": 0.2539,
      "step": 8464
    },
    {
      "epoch": 5.605960264900662,
      "grad_norm": 0.7714540217243087,
      "learning_rate": 5.22693397497449e-06,
      "loss": 0.2354,
      "step": 8465
    },
    {
      "epoch": 5.606622516556292,
      "grad_norm": 0.6396068428681412,
      "learning_rate": 5.2094815398571075e-06,
      "loss": 0.1924,
      "step": 8466
    },
    {
      "epoch": 5.607284768211921,
      "grad_norm": 0.7822072028577506,
      "learning_rate": 5.192057774922337e-06,
      "loss": 0.2637,
      "step": 8467
    },
    {
      "epoch": 5.60794701986755,
      "grad_norm": 0.9030796734585833,
      "learning_rate": 5.174662683620301e-06,
      "loss": 0.2871,
      "step": 8468
    },
    {
      "epoch": 5.608609271523179,
      "grad_norm": 0.7698454819279575,
      "learning_rate": 5.157296269395428e-06,
      "loss": 0.2539,
      "step": 8469
    },
    {
      "epoch": 5.609271523178808,
      "grad_norm": 0.7003535530530464,
      "learning_rate": 5.139958535686451e-06,
      "loss": 0.2217,
      "step": 8470
    },
    {
      "epoch": 5.609933774834437,
      "grad_norm": 0.7217079424270505,
      "learning_rate": 5.122649485926439e-06,
      "loss": 0.209,
      "step": 8471
    },
    {
      "epoch": 5.610596026490066,
      "grad_norm": 0.8664892909776489,
      "learning_rate": 5.105369123542819e-06,
      "loss": 0.3105,
      "step": 8472
    },
    {
      "epoch": 5.611258278145695,
      "grad_norm": 0.7708588323659494,
      "learning_rate": 5.088117451957252e-06,
      "loss": 0.2539,
      "step": 8473
    },
    {
      "epoch": 5.611920529801324,
      "grad_norm": 0.7651796170686557,
      "learning_rate": 5.070894474585807e-06,
      "loss": 0.252,
      "step": 8474
    },
    {
      "epoch": 5.612582781456954,
      "grad_norm": 0.8030373171086901,
      "learning_rate": 5.053700194838806e-06,
      "loss": 0.2451,
      "step": 8475
    },
    {
      "epoch": 5.613245033112583,
      "grad_norm": 0.7502180844733405,
      "learning_rate": 5.036534616120924e-06,
      "loss": 0.2236,
      "step": 8476
    },
    {
      "epoch": 5.613907284768212,
      "grad_norm": 0.8619030455109354,
      "learning_rate": 5.019397741831143e-06,
      "loss": 0.3184,
      "step": 8477
    },
    {
      "epoch": 5.614569536423841,
      "grad_norm": 0.7205090823633746,
      "learning_rate": 5.002289575362767e-06,
      "loss": 0.25,
      "step": 8478
    },
    {
      "epoch": 5.61523178807947,
      "grad_norm": 0.8249374689913206,
      "learning_rate": 4.985210120103383e-06,
      "loss": 0.2852,
      "step": 8479
    },
    {
      "epoch": 5.6158940397351,
      "grad_norm": 0.7382911410621639,
      "learning_rate": 4.968159379434938e-06,
      "loss": 0.2363,
      "step": 8480
    },
    {
      "epoch": 5.6165562913907285,
      "grad_norm": 0.6641485301840342,
      "learning_rate": 4.951137356733681e-06,
      "loss": 0.2275,
      "step": 8481
    },
    {
      "epoch": 5.6172185430463575,
      "grad_norm": 0.724756162719566,
      "learning_rate": 4.9341440553701315e-06,
      "loss": 0.248,
      "step": 8482
    },
    {
      "epoch": 5.617880794701986,
      "grad_norm": 0.7266456200452215,
      "learning_rate": 4.917179478709215e-06,
      "loss": 0.2598,
      "step": 8483
    },
    {
      "epoch": 5.618543046357616,
      "grad_norm": 1.0474044201359225,
      "learning_rate": 4.900243630110062e-06,
      "loss": 0.3965,
      "step": 8484
    },
    {
      "epoch": 5.619205298013245,
      "grad_norm": 0.8426537751290345,
      "learning_rate": 4.883336512926189e-06,
      "loss": 0.3203,
      "step": 8485
    },
    {
      "epoch": 5.619867549668874,
      "grad_norm": 0.785220401367622,
      "learning_rate": 4.866458130505418e-06,
      "loss": 0.2715,
      "step": 8486
    },
    {
      "epoch": 5.620529801324503,
      "grad_norm": 0.8425988365176172,
      "learning_rate": 4.849608486189827e-06,
      "loss": 0.2754,
      "step": 8487
    },
    {
      "epoch": 5.621192052980133,
      "grad_norm": 0.7369887467602914,
      "learning_rate": 4.832787583315828e-06,
      "loss": 0.2246,
      "step": 8488
    },
    {
      "epoch": 5.621854304635762,
      "grad_norm": 0.7361218707803845,
      "learning_rate": 4.815995425214192e-06,
      "loss": 0.2217,
      "step": 8489
    },
    {
      "epoch": 5.622516556291391,
      "grad_norm": 0.7985520742746308,
      "learning_rate": 4.799232015209959e-06,
      "loss": 0.2578,
      "step": 8490
    },
    {
      "epoch": 5.62317880794702,
      "grad_norm": 0.6454044058097419,
      "learning_rate": 4.782497356622422e-06,
      "loss": 0.1904,
      "step": 8491
    },
    {
      "epoch": 5.623841059602649,
      "grad_norm": 0.854726650031557,
      "learning_rate": 4.7657914527652994e-06,
      "loss": 0.2949,
      "step": 8492
    },
    {
      "epoch": 5.6245033112582785,
      "grad_norm": 0.8426656528155526,
      "learning_rate": 4.749114306946528e-06,
      "loss": 0.2969,
      "step": 8493
    },
    {
      "epoch": 5.6251655629139075,
      "grad_norm": 0.7167154625595982,
      "learning_rate": 4.732465922468348e-06,
      "loss": 0.2441,
      "step": 8494
    },
    {
      "epoch": 5.625827814569536,
      "grad_norm": 0.8681019410389452,
      "learning_rate": 4.715846302627374e-06,
      "loss": 0.3027,
      "step": 8495
    },
    {
      "epoch": 5.626490066225165,
      "grad_norm": 0.6992059233735132,
      "learning_rate": 4.699255450714456e-06,
      "loss": 0.2178,
      "step": 8496
    },
    {
      "epoch": 5.627152317880794,
      "grad_norm": 0.6661296447732384,
      "learning_rate": 4.6826933700147676e-06,
      "loss": 0.2031,
      "step": 8497
    },
    {
      "epoch": 5.627814569536424,
      "grad_norm": 0.7338539459661653,
      "learning_rate": 4.666160063807817e-06,
      "loss": 0.2539,
      "step": 8498
    },
    {
      "epoch": 5.628476821192053,
      "grad_norm": 0.6622866212414646,
      "learning_rate": 4.649655535367369e-06,
      "loss": 0.2178,
      "step": 8499
    },
    {
      "epoch": 5.629139072847682,
      "grad_norm": 0.7854816111874673,
      "learning_rate": 4.633179787961494e-06,
      "loss": 0.2754,
      "step": 8500
    },
    {
      "epoch": 5.629801324503311,
      "grad_norm": 0.6961236917235227,
      "learning_rate": 4.616732824852598e-06,
      "loss": 0.2041,
      "step": 8501
    },
    {
      "epoch": 5.630463576158941,
      "grad_norm": 0.7444002390163725,
      "learning_rate": 4.600314649297393e-06,
      "loss": 0.2412,
      "step": 8502
    },
    {
      "epoch": 5.63112582781457,
      "grad_norm": 0.7075861272820734,
      "learning_rate": 4.583925264546828e-06,
      "loss": 0.2285,
      "step": 8503
    },
    {
      "epoch": 5.631788079470199,
      "grad_norm": 0.7448360725956312,
      "learning_rate": 4.567564673846208e-06,
      "loss": 0.2393,
      "step": 8504
    },
    {
      "epoch": 5.632450331125828,
      "grad_norm": 0.774574677500248,
      "learning_rate": 4.551232880435091e-06,
      "loss": 0.2314,
      "step": 8505
    },
    {
      "epoch": 5.633112582781457,
      "grad_norm": 0.6721241266165442,
      "learning_rate": 4.534929887547389e-06,
      "loss": 0.1992,
      "step": 8506
    },
    {
      "epoch": 5.633774834437086,
      "grad_norm": 0.6955758003291067,
      "learning_rate": 4.518655698411272e-06,
      "loss": 0.2236,
      "step": 8507
    },
    {
      "epoch": 5.634437086092715,
      "grad_norm": 0.7259023429007827,
      "learning_rate": 4.502410316249211e-06,
      "loss": 0.2236,
      "step": 8508
    },
    {
      "epoch": 5.635099337748344,
      "grad_norm": 0.6501745976506197,
      "learning_rate": 4.486193744277982e-06,
      "loss": 0.2197,
      "step": 8509
    },
    {
      "epoch": 5.635761589403973,
      "grad_norm": 0.650200124014585,
      "learning_rate": 4.470005985708669e-06,
      "loss": 0.1992,
      "step": 8510
    },
    {
      "epoch": 5.636423841059603,
      "grad_norm": 0.6653012577069665,
      "learning_rate": 4.453847043746622e-06,
      "loss": 0.2236,
      "step": 8511
    },
    {
      "epoch": 5.637086092715232,
      "grad_norm": 0.6974592809883344,
      "learning_rate": 4.437716921591483e-06,
      "loss": 0.2178,
      "step": 8512
    },
    {
      "epoch": 5.637748344370861,
      "grad_norm": 0.8825311385695657,
      "learning_rate": 4.421615622437197e-06,
      "loss": 0.2969,
      "step": 8513
    },
    {
      "epoch": 5.63841059602649,
      "grad_norm": 0.7311989974103863,
      "learning_rate": 4.40554314947203e-06,
      "loss": 0.2305,
      "step": 8514
    },
    {
      "epoch": 5.639072847682119,
      "grad_norm": 0.7106977707789908,
      "learning_rate": 4.389499505878502e-06,
      "loss": 0.2207,
      "step": 8515
    },
    {
      "epoch": 5.639735099337749,
      "grad_norm": 0.8321591582209643,
      "learning_rate": 4.373484694833457e-06,
      "loss": 0.3047,
      "step": 8516
    },
    {
      "epoch": 5.640397350993378,
      "grad_norm": 0.8722504798142211,
      "learning_rate": 4.357498719507974e-06,
      "loss": 0.2832,
      "step": 8517
    },
    {
      "epoch": 5.6410596026490065,
      "grad_norm": 0.8112295318165914,
      "learning_rate": 4.341541583067487e-06,
      "loss": 0.2812,
      "step": 8518
    },
    {
      "epoch": 5.6417218543046355,
      "grad_norm": 0.8015420691657599,
      "learning_rate": 4.3256132886717035e-06,
      "loss": 0.252,
      "step": 8519
    },
    {
      "epoch": 5.642384105960264,
      "grad_norm": 0.8264326679469394,
      "learning_rate": 4.309713839474582e-06,
      "loss": 0.25,
      "step": 8520
    },
    {
      "epoch": 5.643046357615894,
      "grad_norm": 0.7123104355496577,
      "learning_rate": 4.293843238624406e-06,
      "loss": 0.2402,
      "step": 8521
    },
    {
      "epoch": 5.643708609271523,
      "grad_norm": 0.6518236349832517,
      "learning_rate": 4.27800148926371e-06,
      "loss": 0.2061,
      "step": 8522
    },
    {
      "epoch": 5.644370860927152,
      "grad_norm": 0.6630715092768414,
      "learning_rate": 4.262188594529404e-06,
      "loss": 0.1934,
      "step": 8523
    },
    {
      "epoch": 5.645033112582782,
      "grad_norm": 0.7916764999954217,
      "learning_rate": 4.24640455755258e-06,
      "loss": 0.2578,
      "step": 8524
    },
    {
      "epoch": 5.645695364238411,
      "grad_norm": 0.8036782363264839,
      "learning_rate": 4.230649381458623e-06,
      "loss": 0.2734,
      "step": 8525
    },
    {
      "epoch": 5.64635761589404,
      "grad_norm": 0.7253599321378659,
      "learning_rate": 4.214923069367321e-06,
      "loss": 0.1982,
      "step": 8526
    },
    {
      "epoch": 5.647019867549669,
      "grad_norm": 0.7133757881977699,
      "learning_rate": 4.199225624392633e-06,
      "loss": 0.2148,
      "step": 8527
    },
    {
      "epoch": 5.647682119205298,
      "grad_norm": 0.7704658825528609,
      "learning_rate": 4.183557049642788e-06,
      "loss": 0.2539,
      "step": 8528
    },
    {
      "epoch": 5.6483443708609276,
      "grad_norm": 0.7623762474808625,
      "learning_rate": 4.16791734822039e-06,
      "loss": 0.249,
      "step": 8529
    },
    {
      "epoch": 5.6490066225165565,
      "grad_norm": 0.7185483067537491,
      "learning_rate": 4.152306523222276e-06,
      "loss": 0.209,
      "step": 8530
    },
    {
      "epoch": 5.649668874172185,
      "grad_norm": 0.6744706952726712,
      "learning_rate": 4.136724577739542e-06,
      "loss": 0.2188,
      "step": 8531
    },
    {
      "epoch": 5.650331125827814,
      "grad_norm": 0.8612375882948342,
      "learning_rate": 4.1211715148576355e-06,
      "loss": 0.2949,
      "step": 8532
    },
    {
      "epoch": 5.650993377483443,
      "grad_norm": 0.7524673215856749,
      "learning_rate": 4.105647337656192e-06,
      "loss": 0.2344,
      "step": 8533
    },
    {
      "epoch": 5.651655629139073,
      "grad_norm": 0.8624627487494877,
      "learning_rate": 4.0901520492092055e-06,
      "loss": 0.2793,
      "step": 8534
    },
    {
      "epoch": 5.652317880794702,
      "grad_norm": 0.7331302662725838,
      "learning_rate": 4.074685652584919e-06,
      "loss": 0.2402,
      "step": 8535
    },
    {
      "epoch": 5.652980132450331,
      "grad_norm": 0.7830063282058954,
      "learning_rate": 4.059248150845867e-06,
      "loss": 0.2734,
      "step": 8536
    },
    {
      "epoch": 5.65364238410596,
      "grad_norm": 0.6801219166697693,
      "learning_rate": 4.043839547048805e-06,
      "loss": 0.2051,
      "step": 8537
    },
    {
      "epoch": 5.654304635761589,
      "grad_norm": 0.675234335847937,
      "learning_rate": 4.028459844244858e-06,
      "loss": 0.2207,
      "step": 8538
    },
    {
      "epoch": 5.654966887417219,
      "grad_norm": 0.6881030706520367,
      "learning_rate": 4.01310904547939e-06,
      "loss": 0.2129,
      "step": 8539
    },
    {
      "epoch": 5.655629139072848,
      "grad_norm": 0.7552724632652142,
      "learning_rate": 3.997787153791987e-06,
      "loss": 0.2373,
      "step": 8540
    },
    {
      "epoch": 5.656291390728477,
      "grad_norm": 0.7139538729361956,
      "learning_rate": 3.982494172216621e-06,
      "loss": 0.2178,
      "step": 8541
    },
    {
      "epoch": 5.656953642384106,
      "grad_norm": 0.7041626426663392,
      "learning_rate": 3.967230103781421e-06,
      "loss": 0.1963,
      "step": 8542
    },
    {
      "epoch": 5.657615894039735,
      "grad_norm": 0.7219860831987656,
      "learning_rate": 3.951994951508869e-06,
      "loss": 0.2129,
      "step": 8543
    },
    {
      "epoch": 5.658278145695364,
      "grad_norm": 0.8409502608875353,
      "learning_rate": 3.936788718415718e-06,
      "loss": 0.2773,
      "step": 8544
    },
    {
      "epoch": 5.658940397350993,
      "grad_norm": 0.8220322646406694,
      "learning_rate": 3.921611407512959e-06,
      "loss": 0.2676,
      "step": 8545
    },
    {
      "epoch": 5.659602649006622,
      "grad_norm": 0.7175441090841517,
      "learning_rate": 3.906463021805839e-06,
      "loss": 0.2129,
      "step": 8546
    },
    {
      "epoch": 5.660264900662252,
      "grad_norm": 0.6993501135172824,
      "learning_rate": 3.891343564293992e-06,
      "loss": 0.2119,
      "step": 8547
    },
    {
      "epoch": 5.660927152317881,
      "grad_norm": 0.7328783589762904,
      "learning_rate": 3.876253037971206e-06,
      "loss": 0.2246,
      "step": 8548
    },
    {
      "epoch": 5.66158940397351,
      "grad_norm": 0.7656772656708439,
      "learning_rate": 3.86119144582554e-06,
      "loss": 0.2363,
      "step": 8549
    },
    {
      "epoch": 5.662251655629139,
      "grad_norm": 0.9025197899645491,
      "learning_rate": 3.84615879083941e-06,
      "loss": 0.3223,
      "step": 8550
    },
    {
      "epoch": 5.662913907284768,
      "grad_norm": 0.853666487162792,
      "learning_rate": 3.831155075989434e-06,
      "loss": 0.2617,
      "step": 8551
    },
    {
      "epoch": 5.663576158940398,
      "grad_norm": 0.6642960617220305,
      "learning_rate": 3.8161803042465185e-06,
      "loss": 0.2217,
      "step": 8552
    },
    {
      "epoch": 5.664238410596027,
      "grad_norm": 0.6970475076943218,
      "learning_rate": 3.8012344785758596e-06,
      "loss": 0.2002,
      "step": 8553
    },
    {
      "epoch": 5.664900662251656,
      "grad_norm": 0.7323323643977634,
      "learning_rate": 3.786317601936889e-06,
      "loss": 0.2305,
      "step": 8554
    },
    {
      "epoch": 5.6655629139072845,
      "grad_norm": 0.9448954263774241,
      "learning_rate": 3.771429677283294e-06,
      "loss": 0.3125,
      "step": 8555
    },
    {
      "epoch": 5.6662251655629134,
      "grad_norm": 0.6869703212333618,
      "learning_rate": 3.7565707075631167e-06,
      "loss": 0.2119,
      "step": 8556
    },
    {
      "epoch": 5.666887417218543,
      "grad_norm": 0.7468329210827919,
      "learning_rate": 3.7417406957185537e-06,
      "loss": 0.2275,
      "step": 8557
    },
    {
      "epoch": 5.667549668874172,
      "grad_norm": 0.7178786799515048,
      "learning_rate": 3.7269396446861057e-06,
      "loss": 0.1992,
      "step": 8558
    },
    {
      "epoch": 5.668211920529801,
      "grad_norm": 0.7268704655523025,
      "learning_rate": 3.712167557396595e-06,
      "loss": 0.2383,
      "step": 8559
    },
    {
      "epoch": 5.66887417218543,
      "grad_norm": 0.6662134359771252,
      "learning_rate": 3.697424436775032e-06,
      "loss": 0.2197,
      "step": 8560
    },
    {
      "epoch": 5.66953642384106,
      "grad_norm": 0.7576907119054107,
      "learning_rate": 3.6827102857407475e-06,
      "loss": 0.2461,
      "step": 8561
    },
    {
      "epoch": 5.670198675496689,
      "grad_norm": 0.75228012118287,
      "learning_rate": 3.6680251072072787e-06,
      "loss": 0.2334,
      "step": 8562
    },
    {
      "epoch": 5.670860927152318,
      "grad_norm": 0.833643875365092,
      "learning_rate": 3.6533689040824988e-06,
      "loss": 0.2617,
      "step": 8563
    },
    {
      "epoch": 5.671523178807947,
      "grad_norm": 0.7704374423066492,
      "learning_rate": 3.6387416792684697e-06,
      "loss": 0.2471,
      "step": 8564
    },
    {
      "epoch": 5.672185430463577,
      "grad_norm": 0.8086400627370286,
      "learning_rate": 3.624143435661592e-06,
      "loss": 0.2432,
      "step": 8565
    },
    {
      "epoch": 5.6728476821192055,
      "grad_norm": 0.648369241314559,
      "learning_rate": 3.6095741761524355e-06,
      "loss": 0.2119,
      "step": 8566
    },
    {
      "epoch": 5.6735099337748345,
      "grad_norm": 0.7035289876089487,
      "learning_rate": 3.59503390362591e-06,
      "loss": 0.2266,
      "step": 8567
    },
    {
      "epoch": 5.674172185430463,
      "grad_norm": 0.7910925795454296,
      "learning_rate": 3.580522620961146e-06,
      "loss": 0.2246,
      "step": 8568
    },
    {
      "epoch": 5.674834437086092,
      "grad_norm": 0.6874882487466664,
      "learning_rate": 3.566040331031561e-06,
      "loss": 0.1992,
      "step": 8569
    },
    {
      "epoch": 5.675496688741722,
      "grad_norm": 0.7049272783022219,
      "learning_rate": 3.551587036704795e-06,
      "loss": 0.2236,
      "step": 8570
    },
    {
      "epoch": 5.676158940397351,
      "grad_norm": 0.6846851855216668,
      "learning_rate": 3.537162740842758e-06,
      "loss": 0.2207,
      "step": 8571
    },
    {
      "epoch": 5.67682119205298,
      "grad_norm": 0.6900816179124258,
      "learning_rate": 3.5227674463016664e-06,
      "loss": 0.2031,
      "step": 8572
    },
    {
      "epoch": 5.677483443708609,
      "grad_norm": 0.6858039868723763,
      "learning_rate": 3.508401155931906e-06,
      "loss": 0.209,
      "step": 8573
    },
    {
      "epoch": 5.678145695364238,
      "grad_norm": 0.808526551510992,
      "learning_rate": 3.494063872578201e-06,
      "loss": 0.248,
      "step": 8574
    },
    {
      "epoch": 5.678807947019868,
      "grad_norm": 0.7473907055470922,
      "learning_rate": 3.4797555990794645e-06,
      "loss": 0.2295,
      "step": 8575
    },
    {
      "epoch": 5.679470198675497,
      "grad_norm": 0.7926581910556464,
      "learning_rate": 3.46547633826893e-06,
      "loss": 0.2793,
      "step": 8576
    },
    {
      "epoch": 5.680132450331126,
      "grad_norm": 0.7996822227126037,
      "learning_rate": 3.451226092974052e-06,
      "loss": 0.2812,
      "step": 8577
    },
    {
      "epoch": 5.680794701986755,
      "grad_norm": 0.6676539453929504,
      "learning_rate": 3.4370048660165404e-06,
      "loss": 0.2168,
      "step": 8578
    },
    {
      "epoch": 5.6814569536423845,
      "grad_norm": 0.7683045936438035,
      "learning_rate": 3.4228126602123264e-06,
      "loss": 0.252,
      "step": 8579
    },
    {
      "epoch": 5.682119205298013,
      "grad_norm": 0.7442792110756081,
      "learning_rate": 3.4086494783716776e-06,
      "loss": 0.2402,
      "step": 8580
    },
    {
      "epoch": 5.682781456953642,
      "grad_norm": 0.8850624310786916,
      "learning_rate": 3.394515323299052e-06,
      "loss": 0.2988,
      "step": 8581
    },
    {
      "epoch": 5.683443708609271,
      "grad_norm": 0.8546360882550629,
      "learning_rate": 3.38041019779316e-06,
      "loss": 0.2891,
      "step": 8582
    },
    {
      "epoch": 5.684105960264901,
      "grad_norm": 0.7725394639362063,
      "learning_rate": 3.366334104646984e-06,
      "loss": 0.2256,
      "step": 8583
    },
    {
      "epoch": 5.68476821192053,
      "grad_norm": 0.799132548546653,
      "learning_rate": 3.3522870466477616e-06,
      "loss": 0.2598,
      "step": 8584
    },
    {
      "epoch": 5.685430463576159,
      "grad_norm": 0.643589630239686,
      "learning_rate": 3.3382690265769675e-06,
      "loss": 0.1836,
      "step": 8585
    },
    {
      "epoch": 5.686092715231788,
      "grad_norm": 0.7947782396089556,
      "learning_rate": 3.3242800472103316e-06,
      "loss": 0.2441,
      "step": 8586
    },
    {
      "epoch": 5.686754966887417,
      "grad_norm": 0.8028651906906636,
      "learning_rate": 3.3103201113178047e-06,
      "loss": 0.2598,
      "step": 8587
    },
    {
      "epoch": 5.687417218543047,
      "grad_norm": 0.7922659986714103,
      "learning_rate": 3.2963892216636754e-06,
      "loss": 0.2441,
      "step": 8588
    },
    {
      "epoch": 5.688079470198676,
      "grad_norm": 0.8025376847540597,
      "learning_rate": 3.2824873810063546e-06,
      "loss": 0.2363,
      "step": 8589
    },
    {
      "epoch": 5.688741721854305,
      "grad_norm": 0.6783142847557733,
      "learning_rate": 3.2686145920986064e-06,
      "loss": 0.2227,
      "step": 8590
    },
    {
      "epoch": 5.6894039735099335,
      "grad_norm": 0.7800951401510476,
      "learning_rate": 3.254770857687383e-06,
      "loss": 0.2715,
      "step": 8591
    },
    {
      "epoch": 5.6900662251655625,
      "grad_norm": 0.7376079901427984,
      "learning_rate": 3.240956180513893e-06,
      "loss": 0.2393,
      "step": 8592
    },
    {
      "epoch": 5.690728476821192,
      "grad_norm": 0.7675613267619977,
      "learning_rate": 3.227170563313647e-06,
      "loss": 0.2471,
      "step": 8593
    },
    {
      "epoch": 5.691390728476821,
      "grad_norm": 0.6444713595340121,
      "learning_rate": 3.2134140088163284e-06,
      "loss": 0.2021,
      "step": 8594
    },
    {
      "epoch": 5.69205298013245,
      "grad_norm": 0.7421009342083975,
      "learning_rate": 3.1996865197458586e-06,
      "loss": 0.2422,
      "step": 8595
    },
    {
      "epoch": 5.692715231788079,
      "grad_norm": 0.7727205023404466,
      "learning_rate": 3.18598809882048e-06,
      "loss": 0.2344,
      "step": 8596
    },
    {
      "epoch": 5.693377483443709,
      "grad_norm": 0.8790695565464431,
      "learning_rate": 3.172318748752639e-06,
      "loss": 0.291,
      "step": 8597
    },
    {
      "epoch": 5.694039735099338,
      "grad_norm": 0.7217626471124348,
      "learning_rate": 3.1586784722489876e-06,
      "loss": 0.2441,
      "step": 8598
    },
    {
      "epoch": 5.694701986754967,
      "grad_norm": 0.7007544142337919,
      "learning_rate": 3.1450672720104986e-06,
      "loss": 0.1797,
      "step": 8599
    },
    {
      "epoch": 5.695364238410596,
      "grad_norm": 0.8626877288751443,
      "learning_rate": 3.1314851507323156e-06,
      "loss": 0.2617,
      "step": 8600
    },
    {
      "epoch": 5.696026490066226,
      "grad_norm": 0.7587759161443056,
      "learning_rate": 3.117932111103871e-06,
      "loss": 0.208,
      "step": 8601
    },
    {
      "epoch": 5.696688741721855,
      "grad_norm": 0.7655784543850515,
      "learning_rate": 3.1044081558088185e-06,
      "loss": 0.2324,
      "step": 8602
    },
    {
      "epoch": 5.6973509933774835,
      "grad_norm": 0.7106988096305012,
      "learning_rate": 3.0909132875250486e-06,
      "loss": 0.209,
      "step": 8603
    },
    {
      "epoch": 5.6980132450331125,
      "grad_norm": 0.6833138393880326,
      "learning_rate": 3.0774475089247075e-06,
      "loss": 0.2422,
      "step": 8604
    },
    {
      "epoch": 5.698675496688741,
      "grad_norm": 0.7430705986006035,
      "learning_rate": 3.064010822674179e-06,
      "loss": 0.2324,
      "step": 8605
    },
    {
      "epoch": 5.699337748344371,
      "grad_norm": 0.8602248738876542,
      "learning_rate": 3.050603231434068e-06,
      "loss": 0.2754,
      "step": 8606
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.7380158473752596,
      "learning_rate": 3.037224737859234e-06,
      "loss": 0.2314,
      "step": 8607
    },
    {
      "epoch": 5.700662251655629,
      "grad_norm": 0.6632044511143497,
      "learning_rate": 3.023875344598775e-06,
      "loss": 0.2188,
      "step": 8608
    },
    {
      "epoch": 5.701324503311258,
      "grad_norm": 0.6206776248097631,
      "learning_rate": 3.0105550542960264e-06,
      "loss": 0.1953,
      "step": 8609
    },
    {
      "epoch": 5.701986754966887,
      "grad_norm": 0.7805705457006857,
      "learning_rate": 2.9972638695885454e-06,
      "loss": 0.2451,
      "step": 8610
    },
    {
      "epoch": 5.702649006622517,
      "grad_norm": 0.7637203865877168,
      "learning_rate": 2.984001793108176e-06,
      "loss": 0.2676,
      "step": 8611
    },
    {
      "epoch": 5.703311258278146,
      "grad_norm": 0.8546547458816739,
      "learning_rate": 2.9707688274809183e-06,
      "loss": 0.291,
      "step": 8612
    },
    {
      "epoch": 5.703973509933775,
      "grad_norm": 0.7315628751428759,
      "learning_rate": 2.9575649753270593e-06,
      "loss": 0.2324,
      "step": 8613
    },
    {
      "epoch": 5.704635761589404,
      "grad_norm": 0.8296784239040956,
      "learning_rate": 2.9443902392611407e-06,
      "loss": 0.2773,
      "step": 8614
    },
    {
      "epoch": 5.7052980132450335,
      "grad_norm": 0.8341940809898911,
      "learning_rate": 2.9312446218918926e-06,
      "loss": 0.2637,
      "step": 8615
    },
    {
      "epoch": 5.705960264900662,
      "grad_norm": 0.6827283667465832,
      "learning_rate": 2.9181281258222665e-06,
      "loss": 0.2158,
      "step": 8616
    },
    {
      "epoch": 5.706622516556291,
      "grad_norm": 0.7238618702923731,
      "learning_rate": 2.9050407536495346e-06,
      "loss": 0.2412,
      "step": 8617
    },
    {
      "epoch": 5.70728476821192,
      "grad_norm": 1.0761811260950445,
      "learning_rate": 2.891982507965124e-06,
      "loss": 0.3789,
      "step": 8618
    },
    {
      "epoch": 5.70794701986755,
      "grad_norm": 0.7594725598999803,
      "learning_rate": 2.878953391354699e-06,
      "loss": 0.2559,
      "step": 8619
    },
    {
      "epoch": 5.708609271523179,
      "grad_norm": 0.740918104653288,
      "learning_rate": 2.865953406398197e-06,
      "loss": 0.2832,
      "step": 8620
    },
    {
      "epoch": 5.709271523178808,
      "grad_norm": 0.8158432659720243,
      "learning_rate": 2.8529825556697748e-06,
      "loss": 0.2715,
      "step": 8621
    },
    {
      "epoch": 5.709933774834437,
      "grad_norm": 0.7074515762835106,
      "learning_rate": 2.8400408417377786e-06,
      "loss": 0.2461,
      "step": 8622
    },
    {
      "epoch": 5.710596026490066,
      "grad_norm": 0.7652949200933105,
      "learning_rate": 2.827128267164841e-06,
      "loss": 0.2715,
      "step": 8623
    },
    {
      "epoch": 5.711258278145696,
      "grad_norm": 0.7474207508758464,
      "learning_rate": 2.8142448345077673e-06,
      "loss": 0.2734,
      "step": 8624
    },
    {
      "epoch": 5.711920529801325,
      "grad_norm": 0.7188052119138011,
      "learning_rate": 2.80139054631765e-06,
      "loss": 0.2109,
      "step": 8625
    },
    {
      "epoch": 5.712582781456954,
      "grad_norm": 0.7147471376432075,
      "learning_rate": 2.7885654051397867e-06,
      "loss": 0.2197,
      "step": 8626
    },
    {
      "epoch": 5.713245033112583,
      "grad_norm": 0.818025315003347,
      "learning_rate": 2.775769413513712e-06,
      "loss": 0.2793,
      "step": 8627
    },
    {
      "epoch": 5.7139072847682115,
      "grad_norm": 0.7737542097602708,
      "learning_rate": 2.7630025739731498e-06,
      "loss": 0.2432,
      "step": 8628
    },
    {
      "epoch": 5.714569536423841,
      "grad_norm": 0.7151855972234946,
      "learning_rate": 2.7502648890460775e-06,
      "loss": 0.2188,
      "step": 8629
    },
    {
      "epoch": 5.71523178807947,
      "grad_norm": 0.6979875731903878,
      "learning_rate": 2.7375563612547446e-06,
      "loss": 0.2188,
      "step": 8630
    },
    {
      "epoch": 5.715894039735099,
      "grad_norm": 0.7892823754864899,
      "learning_rate": 2.7248769931155546e-06,
      "loss": 0.2578,
      "step": 8631
    },
    {
      "epoch": 5.716556291390728,
      "grad_norm": 0.7057219418411872,
      "learning_rate": 2.7122267871391823e-06,
      "loss": 0.2139,
      "step": 8632
    },
    {
      "epoch": 5.717218543046357,
      "grad_norm": 0.723895318119924,
      "learning_rate": 2.6996057458304908e-06,
      "loss": 0.2148,
      "step": 8633
    },
    {
      "epoch": 5.717880794701987,
      "grad_norm": 0.7368163756512313,
      "learning_rate": 2.6870138716886313e-06,
      "loss": 0.2041,
      "step": 8634
    },
    {
      "epoch": 5.718543046357616,
      "grad_norm": 0.7286834079608066,
      "learning_rate": 2.674451167206876e-06,
      "loss": 0.2246,
      "step": 8635
    },
    {
      "epoch": 5.719205298013245,
      "grad_norm": 0.816493453494043,
      "learning_rate": 2.661917634872851e-06,
      "loss": 0.293,
      "step": 8636
    },
    {
      "epoch": 5.719867549668874,
      "grad_norm": 1.0768925360474695,
      "learning_rate": 2.6494132771682887e-06,
      "loss": 0.4219,
      "step": 8637
    },
    {
      "epoch": 5.720529801324504,
      "grad_norm": 0.7362226662124302,
      "learning_rate": 2.6369380965692246e-06,
      "loss": 0.2412,
      "step": 8638
    },
    {
      "epoch": 5.721192052980133,
      "grad_norm": 0.6634310156524303,
      "learning_rate": 2.6244920955458994e-06,
      "loss": 0.2021,
      "step": 8639
    },
    {
      "epoch": 5.7218543046357615,
      "grad_norm": 0.7388836698531333,
      "learning_rate": 2.612075276562725e-06,
      "loss": 0.2344,
      "step": 8640
    },
    {
      "epoch": 5.72251655629139,
      "grad_norm": 0.7647830281345201,
      "learning_rate": 2.5996876420783854e-06,
      "loss": 0.249,
      "step": 8641
    },
    {
      "epoch": 5.72317880794702,
      "grad_norm": 0.742924859890923,
      "learning_rate": 2.587329194545801e-06,
      "loss": 0.2598,
      "step": 8642
    },
    {
      "epoch": 5.723841059602649,
      "grad_norm": 0.871409100066138,
      "learning_rate": 2.5749999364120643e-06,
      "loss": 0.2969,
      "step": 8643
    },
    {
      "epoch": 5.724503311258278,
      "grad_norm": 0.7323304545548751,
      "learning_rate": 2.5626998701185065e-06,
      "loss": 0.2539,
      "step": 8644
    },
    {
      "epoch": 5.725165562913907,
      "grad_norm": 0.7208838531302878,
      "learning_rate": 2.550428998100679e-06,
      "loss": 0.2412,
      "step": 8645
    },
    {
      "epoch": 5.725827814569536,
      "grad_norm": 0.6668759913903309,
      "learning_rate": 2.5381873227883874e-06,
      "loss": 0.1934,
      "step": 8646
    },
    {
      "epoch": 5.726490066225166,
      "grad_norm": 0.5976642479280211,
      "learning_rate": 2.5259748466055774e-06,
      "loss": 0.1846,
      "step": 8647
    },
    {
      "epoch": 5.727152317880795,
      "grad_norm": 0.810121359011573,
      "learning_rate": 2.513791571970497e-06,
      "loss": 0.2471,
      "step": 8648
    },
    {
      "epoch": 5.727814569536424,
      "grad_norm": 0.7210981807843834,
      "learning_rate": 2.5016375012955504e-06,
      "loss": 0.1885,
      "step": 8649
    },
    {
      "epoch": 5.728476821192053,
      "grad_norm": 0.7833789979075667,
      "learning_rate": 2.4895126369873953e-06,
      "loss": 0.2383,
      "step": 8650
    },
    {
      "epoch": 5.729139072847682,
      "grad_norm": 0.8151711431201897,
      "learning_rate": 2.477416981446895e-06,
      "loss": 0.2656,
      "step": 8651
    },
    {
      "epoch": 5.7298013245033115,
      "grad_norm": 0.8924368760128949,
      "learning_rate": 2.465350537069133e-06,
      "loss": 0.332,
      "step": 8652
    },
    {
      "epoch": 5.73046357615894,
      "grad_norm": 0.693292124364326,
      "learning_rate": 2.4533133062433653e-06,
      "loss": 0.2227,
      "step": 8653
    },
    {
      "epoch": 5.731125827814569,
      "grad_norm": 0.8739236686153801,
      "learning_rate": 2.441305291353152e-06,
      "loss": 0.3008,
      "step": 8654
    },
    {
      "epoch": 5.731788079470198,
      "grad_norm": 0.935125555202239,
      "learning_rate": 2.4293264947761904e-06,
      "loss": 0.3418,
      "step": 8655
    },
    {
      "epoch": 5.732450331125828,
      "grad_norm": 0.7218634894708744,
      "learning_rate": 2.417376918884417e-06,
      "loss": 0.2246,
      "step": 8656
    },
    {
      "epoch": 5.733112582781457,
      "grad_norm": 0.8350880091907508,
      "learning_rate": 2.405456566044006e-06,
      "loss": 0.2832,
      "step": 8657
    },
    {
      "epoch": 5.733774834437086,
      "grad_norm": 0.7428693814787666,
      "learning_rate": 2.3935654386153026e-06,
      "loss": 0.2305,
      "step": 8658
    },
    {
      "epoch": 5.734437086092715,
      "grad_norm": 0.8429084936738148,
      "learning_rate": 2.381703538952906e-06,
      "loss": 0.2773,
      "step": 8659
    },
    {
      "epoch": 5.735099337748345,
      "grad_norm": 0.7807614098197941,
      "learning_rate": 2.369870869405588e-06,
      "loss": 0.2539,
      "step": 8660
    },
    {
      "epoch": 5.735761589403974,
      "grad_norm": 0.8037381157085843,
      "learning_rate": 2.3580674323163574e-06,
      "loss": 0.2617,
      "step": 8661
    },
    {
      "epoch": 5.736423841059603,
      "grad_norm": 0.6683369236054126,
      "learning_rate": 2.3462932300224448e-06,
      "loss": 0.2012,
      "step": 8662
    },
    {
      "epoch": 5.737086092715232,
      "grad_norm": 0.7468594377186247,
      "learning_rate": 2.3345482648552682e-06,
      "loss": 0.2471,
      "step": 8663
    },
    {
      "epoch": 5.737748344370861,
      "grad_norm": 0.7753412220266124,
      "learning_rate": 2.3228325391404677e-06,
      "loss": 0.2393,
      "step": 8664
    },
    {
      "epoch": 5.73841059602649,
      "grad_norm": 0.7195722412527152,
      "learning_rate": 2.3111460551978877e-06,
      "loss": 0.2061,
      "step": 8665
    },
    {
      "epoch": 5.739072847682119,
      "grad_norm": 0.7373996863934508,
      "learning_rate": 2.2994888153415605e-06,
      "loss": 0.2363,
      "step": 8666
    },
    {
      "epoch": 5.739735099337748,
      "grad_norm": 0.7050511936168654,
      "learning_rate": 2.2878608218798057e-06,
      "loss": 0.209,
      "step": 8667
    },
    {
      "epoch": 5.740397350993377,
      "grad_norm": 0.7172396657055471,
      "learning_rate": 2.2762620771150653e-06,
      "loss": 0.2246,
      "step": 8668
    },
    {
      "epoch": 5.741059602649006,
      "grad_norm": 0.6966412225135395,
      "learning_rate": 2.2646925833440355e-06,
      "loss": 0.2236,
      "step": 8669
    },
    {
      "epoch": 5.741721854304636,
      "grad_norm": 0.6390564833159935,
      "learning_rate": 2.2531523428576005e-06,
      "loss": 0.1895,
      "step": 8670
    },
    {
      "epoch": 5.742384105960265,
      "grad_norm": 0.7010901656587819,
      "learning_rate": 2.241641357940849e-06,
      "loss": 0.2266,
      "step": 8671
    },
    {
      "epoch": 5.743046357615894,
      "grad_norm": 0.8219829680540488,
      "learning_rate": 2.230159630873124e-06,
      "loss": 0.2559,
      "step": 8672
    },
    {
      "epoch": 5.743708609271523,
      "grad_norm": 0.6978902351292718,
      "learning_rate": 2.218707163927924e-06,
      "loss": 0.2168,
      "step": 8673
    },
    {
      "epoch": 5.744370860927153,
      "grad_norm": 0.7383504338197789,
      "learning_rate": 2.207283959372935e-06,
      "loss": 0.2188,
      "step": 8674
    },
    {
      "epoch": 5.745033112582782,
      "grad_norm": 0.7485799926914288,
      "learning_rate": 2.195890019470131e-06,
      "loss": 0.2354,
      "step": 8675
    },
    {
      "epoch": 5.7456953642384105,
      "grad_norm": 0.7455558497780159,
      "learning_rate": 2.184525346475624e-06,
      "loss": 0.2207,
      "step": 8676
    },
    {
      "epoch": 5.7463576158940395,
      "grad_norm": 0.6702953788231936,
      "learning_rate": 2.1731899426397305e-06,
      "loss": 0.2139,
      "step": 8677
    },
    {
      "epoch": 5.747019867549669,
      "grad_norm": 0.7851674112158219,
      "learning_rate": 2.1618838102070046e-06,
      "loss": 0.2637,
      "step": 8678
    },
    {
      "epoch": 5.747682119205298,
      "grad_norm": 0.720888201542392,
      "learning_rate": 2.150606951416223e-06,
      "loss": 0.2402,
      "step": 8679
    },
    {
      "epoch": 5.748344370860927,
      "grad_norm": 0.7567741128796722,
      "learning_rate": 2.139359368500282e-06,
      "loss": 0.2617,
      "step": 8680
    },
    {
      "epoch": 5.749006622516556,
      "grad_norm": 0.7512194420382396,
      "learning_rate": 2.1281410636863504e-06,
      "loss": 0.2363,
      "step": 8681
    },
    {
      "epoch": 5.749668874172185,
      "grad_norm": 0.7641420717826817,
      "learning_rate": 2.116952039195785e-06,
      "loss": 0.249,
      "step": 8682
    },
    {
      "epoch": 5.750331125827815,
      "grad_norm": 0.747889420272799,
      "learning_rate": 2.1057922972441464e-06,
      "loss": 0.2354,
      "step": 8683
    },
    {
      "epoch": 5.750993377483444,
      "grad_norm": 0.726693111050286,
      "learning_rate": 2.094661840041184e-06,
      "loss": 0.2168,
      "step": 8684
    },
    {
      "epoch": 5.751655629139073,
      "grad_norm": 0.7429999503227016,
      "learning_rate": 2.083560669790868e-06,
      "loss": 0.2363,
      "step": 8685
    },
    {
      "epoch": 5.752317880794702,
      "grad_norm": 0.6694714308508266,
      "learning_rate": 2.0724887886913234e-06,
      "loss": 0.1963,
      "step": 8686
    },
    {
      "epoch": 5.752980132450331,
      "grad_norm": 0.7474054099951204,
      "learning_rate": 2.0614461989349295e-06,
      "loss": 0.2637,
      "step": 8687
    },
    {
      "epoch": 5.7536423841059605,
      "grad_norm": 0.7489807726623684,
      "learning_rate": 2.050432902708271e-06,
      "loss": 0.2695,
      "step": 8688
    },
    {
      "epoch": 5.7543046357615895,
      "grad_norm": 0.8461572490782385,
      "learning_rate": 2.0394489021920703e-06,
      "loss": 0.2852,
      "step": 8689
    },
    {
      "epoch": 5.754966887417218,
      "grad_norm": 0.8823206448204663,
      "learning_rate": 2.0284941995613037e-06,
      "loss": 0.3242,
      "step": 8690
    },
    {
      "epoch": 5.755629139072847,
      "grad_norm": 0.6187271290370921,
      "learning_rate": 2.0175687969851195e-06,
      "loss": 0.1895,
      "step": 8691
    },
    {
      "epoch": 5.756291390728477,
      "grad_norm": 0.663407771281543,
      "learning_rate": 2.0066726966268876e-06,
      "loss": 0.2109,
      "step": 8692
    },
    {
      "epoch": 5.756953642384106,
      "grad_norm": 0.7294767472237663,
      "learning_rate": 1.995805900644132e-06,
      "loss": 0.2402,
      "step": 8693
    },
    {
      "epoch": 5.757615894039735,
      "grad_norm": 0.8270008901078862,
      "learning_rate": 1.9849684111886477e-06,
      "loss": 0.2852,
      "step": 8694
    },
    {
      "epoch": 5.758278145695364,
      "grad_norm": 0.7418373565958783,
      "learning_rate": 1.9741602304063353e-06,
      "loss": 0.2559,
      "step": 8695
    },
    {
      "epoch": 5.758940397350994,
      "grad_norm": 0.6910156237449533,
      "learning_rate": 1.9633813604373492e-06,
      "loss": 0.2217,
      "step": 8696
    },
    {
      "epoch": 5.759602649006623,
      "grad_norm": 0.6094836185422668,
      "learning_rate": 1.952631803416066e-06,
      "loss": 0.1875,
      "step": 8697
    },
    {
      "epoch": 5.760264900662252,
      "grad_norm": 0.714375829297667,
      "learning_rate": 1.941911561470966e-06,
      "loss": 0.2197,
      "step": 8698
    },
    {
      "epoch": 5.760927152317881,
      "grad_norm": 0.7281800091808914,
      "learning_rate": 1.9312206367248172e-06,
      "loss": 0.2041,
      "step": 8699
    },
    {
      "epoch": 5.76158940397351,
      "grad_norm": 0.7824293676245234,
      "learning_rate": 1.920559031294544e-06,
      "loss": 0.2373,
      "step": 8700
    },
    {
      "epoch": 5.762251655629139,
      "grad_norm": 0.724803558744917,
      "learning_rate": 1.909926747291257e-06,
      "loss": 0.2363,
      "step": 8701
    },
    {
      "epoch": 5.762913907284768,
      "grad_norm": 0.7660724613158997,
      "learning_rate": 1.899323786820256e-06,
      "loss": 0.2012,
      "step": 8702
    },
    {
      "epoch": 5.763576158940397,
      "grad_norm": 0.6899600180029024,
      "learning_rate": 1.8887501519810778e-06,
      "loss": 0.1846,
      "step": 8703
    },
    {
      "epoch": 5.764238410596026,
      "grad_norm": 0.7528873231734479,
      "learning_rate": 1.8782058448673976e-06,
      "loss": 0.2539,
      "step": 8704
    },
    {
      "epoch": 5.764900662251655,
      "grad_norm": 0.7978965454652358,
      "learning_rate": 1.8676908675671287e-06,
      "loss": 0.2471,
      "step": 8705
    },
    {
      "epoch": 5.765562913907285,
      "grad_norm": 0.7742595522469865,
      "learning_rate": 1.8572052221623558e-06,
      "loss": 0.2422,
      "step": 8706
    },
    {
      "epoch": 5.766225165562914,
      "grad_norm": 0.7254702090895969,
      "learning_rate": 1.8467489107293509e-06,
      "loss": 0.209,
      "step": 8707
    },
    {
      "epoch": 5.766887417218543,
      "grad_norm": 0.6898917284902956,
      "learning_rate": 1.8363219353385584e-06,
      "loss": 0.2188,
      "step": 8708
    },
    {
      "epoch": 5.767549668874172,
      "grad_norm": 0.7212818688251796,
      "learning_rate": 1.825924298054693e-06,
      "loss": 0.2109,
      "step": 8709
    },
    {
      "epoch": 5.768211920529802,
      "grad_norm": 0.6858390986814731,
      "learning_rate": 1.8155560009365744e-06,
      "loss": 0.1914,
      "step": 8710
    },
    {
      "epoch": 5.768874172185431,
      "grad_norm": 0.841095462692886,
      "learning_rate": 1.8052170460372272e-06,
      "loss": 0.291,
      "step": 8711
    },
    {
      "epoch": 5.76953642384106,
      "grad_norm": 0.7233751242706888,
      "learning_rate": 1.794907435403914e-06,
      "loss": 0.2305,
      "step": 8712
    },
    {
      "epoch": 5.7701986754966885,
      "grad_norm": 0.6990412551066661,
      "learning_rate": 1.7846271710780514e-06,
      "loss": 0.1953,
      "step": 8713
    },
    {
      "epoch": 5.770860927152318,
      "grad_norm": 0.7677927325779764,
      "learning_rate": 1.774376255095228e-06,
      "loss": 0.2402,
      "step": 8714
    },
    {
      "epoch": 5.771523178807947,
      "grad_norm": 0.7767490531768444,
      "learning_rate": 1.7641546894852699e-06,
      "loss": 0.2617,
      "step": 8715
    },
    {
      "epoch": 5.772185430463576,
      "grad_norm": 0.6391866385137092,
      "learning_rate": 1.7539624762721581e-06,
      "loss": 0.2031,
      "step": 8716
    },
    {
      "epoch": 5.772847682119205,
      "grad_norm": 0.8168142306369967,
      "learning_rate": 1.7437996174740453e-06,
      "loss": 0.2656,
      "step": 8717
    },
    {
      "epoch": 5.773509933774834,
      "grad_norm": 0.7582981367273539,
      "learning_rate": 1.7336661151033215e-06,
      "loss": 0.2734,
      "step": 8718
    },
    {
      "epoch": 5.774172185430464,
      "grad_norm": 0.7932610039544626,
      "learning_rate": 1.723561971166515e-06,
      "loss": 0.2295,
      "step": 8719
    },
    {
      "epoch": 5.774834437086093,
      "grad_norm": 0.670023723496463,
      "learning_rate": 1.7134871876643586e-06,
      "loss": 0.2158,
      "step": 8720
    },
    {
      "epoch": 5.775496688741722,
      "grad_norm": 0.7403726199307219,
      "learning_rate": 1.7034417665918232e-06,
      "loss": 0.249,
      "step": 8721
    },
    {
      "epoch": 5.776158940397351,
      "grad_norm": 0.6895892629441779,
      "learning_rate": 1.6934257099379678e-06,
      "loss": 0.2188,
      "step": 8722
    },
    {
      "epoch": 5.77682119205298,
      "grad_norm": 0.8286805103921656,
      "learning_rate": 1.683439019686089e-06,
      "loss": 0.2715,
      "step": 8723
    },
    {
      "epoch": 5.77748344370861,
      "grad_norm": 0.8142738717599949,
      "learning_rate": 1.673481697813689e-06,
      "loss": 0.2891,
      "step": 8724
    },
    {
      "epoch": 5.7781456953642385,
      "grad_norm": 0.8240483476053179,
      "learning_rate": 1.663553746292423e-06,
      "loss": 0.2598,
      "step": 8725
    },
    {
      "epoch": 5.778807947019867,
      "grad_norm": 0.8894195355813784,
      "learning_rate": 1.653655167088136e-06,
      "loss": 0.3359,
      "step": 8726
    },
    {
      "epoch": 5.779470198675496,
      "grad_norm": 0.6747133327483652,
      "learning_rate": 1.6437859621608595e-06,
      "loss": 0.1924,
      "step": 8727
    },
    {
      "epoch": 5.780132450331126,
      "grad_norm": 0.6608283043488363,
      "learning_rate": 1.633946133464814e-06,
      "loss": 0.208,
      "step": 8728
    },
    {
      "epoch": 5.780794701986755,
      "grad_norm": 0.6906683344541861,
      "learning_rate": 1.624135682948391e-06,
      "loss": 0.2266,
      "step": 8729
    },
    {
      "epoch": 5.781456953642384,
      "grad_norm": 0.8170174013216691,
      "learning_rate": 1.6143546125541695e-06,
      "loss": 0.252,
      "step": 8730
    },
    {
      "epoch": 5.782119205298013,
      "grad_norm": 0.6776945246319762,
      "learning_rate": 1.604602924218934e-06,
      "loss": 0.1982,
      "step": 8731
    },
    {
      "epoch": 5.782781456953643,
      "grad_norm": 0.8505380538283959,
      "learning_rate": 1.5948806198736063e-06,
      "loss": 0.2617,
      "step": 8732
    },
    {
      "epoch": 5.783443708609272,
      "grad_norm": 0.6556298803771343,
      "learning_rate": 1.58518770144333e-06,
      "loss": 0.1963,
      "step": 8733
    },
    {
      "epoch": 5.784105960264901,
      "grad_norm": 0.7842342421880394,
      "learning_rate": 1.5755241708474198e-06,
      "loss": 0.25,
      "step": 8734
    },
    {
      "epoch": 5.78476821192053,
      "grad_norm": 0.6754407746985512,
      "learning_rate": 1.5658900299993448e-06,
      "loss": 0.21,
      "step": 8735
    },
    {
      "epoch": 5.785430463576159,
      "grad_norm": 0.6841954183643724,
      "learning_rate": 1.5562852808067794e-06,
      "loss": 0.2061,
      "step": 8736
    },
    {
      "epoch": 5.7860927152317885,
      "grad_norm": 0.7519840086843343,
      "learning_rate": 1.5467099251715852e-06,
      "loss": 0.2441,
      "step": 8737
    },
    {
      "epoch": 5.786754966887417,
      "grad_norm": 0.7071820034837071,
      "learning_rate": 1.537163964989796e-06,
      "loss": 0.2178,
      "step": 8738
    },
    {
      "epoch": 5.787417218543046,
      "grad_norm": 0.7670814629446293,
      "learning_rate": 1.5276474021515994e-06,
      "loss": 0.2383,
      "step": 8739
    },
    {
      "epoch": 5.788079470198675,
      "grad_norm": 0.7492701692483842,
      "learning_rate": 1.518160238541405e-06,
      "loss": 0.2334,
      "step": 8740
    },
    {
      "epoch": 5.788741721854304,
      "grad_norm": 0.7412888329440318,
      "learning_rate": 1.5087024760377598e-06,
      "loss": 0.2559,
      "step": 8741
    },
    {
      "epoch": 5.789403973509934,
      "grad_norm": 0.8636727376525105,
      "learning_rate": 1.4992741165134493e-06,
      "loss": 0.2656,
      "step": 8742
    },
    {
      "epoch": 5.790066225165563,
      "grad_norm": 0.8070364214414796,
      "learning_rate": 1.4898751618353466e-06,
      "loss": 0.2617,
      "step": 8743
    },
    {
      "epoch": 5.790728476821192,
      "grad_norm": 0.7898507532078801,
      "learning_rate": 1.4805056138645799e-06,
      "loss": 0.2637,
      "step": 8744
    },
    {
      "epoch": 5.791390728476821,
      "grad_norm": 0.6598634522465825,
      "learning_rate": 1.471165474456415e-06,
      "loss": 0.209,
      "step": 8745
    },
    {
      "epoch": 5.79205298013245,
      "grad_norm": 0.9350772134505817,
      "learning_rate": 1.4618547454603224e-06,
      "loss": 0.3281,
      "step": 8746
    },
    {
      "epoch": 5.79271523178808,
      "grad_norm": 0.8733134161967404,
      "learning_rate": 1.4525734287199276e-06,
      "loss": 0.3145,
      "step": 8747
    },
    {
      "epoch": 5.793377483443709,
      "grad_norm": 0.7272498549436502,
      "learning_rate": 1.443321526073027e-06,
      "loss": 0.207,
      "step": 8748
    },
    {
      "epoch": 5.794039735099338,
      "grad_norm": 0.7884254176266394,
      "learning_rate": 1.4340990393516216e-06,
      "loss": 0.2715,
      "step": 8749
    },
    {
      "epoch": 5.7947019867549665,
      "grad_norm": 0.7788952943897341,
      "learning_rate": 1.424905970381851e-06,
      "loss": 0.2412,
      "step": 8750
    },
    {
      "epoch": 5.795364238410596,
      "grad_norm": 0.7886484334670395,
      "learning_rate": 1.415742320984059e-06,
      "loss": 0.2598,
      "step": 8751
    },
    {
      "epoch": 5.796026490066225,
      "grad_norm": 0.910694243625502,
      "learning_rate": 1.4066080929727275e-06,
      "loss": 0.3398,
      "step": 8752
    },
    {
      "epoch": 5.796688741721854,
      "grad_norm": 0.6926864395615128,
      "learning_rate": 1.397503288156593e-06,
      "loss": 0.2285,
      "step": 8753
    },
    {
      "epoch": 5.797350993377483,
      "grad_norm": 0.5956281691630017,
      "learning_rate": 1.388427908338463e-06,
      "loss": 0.1602,
      "step": 8754
    },
    {
      "epoch": 5.798013245033113,
      "grad_norm": 0.7584226254609122,
      "learning_rate": 1.3793819553154007e-06,
      "loss": 0.2197,
      "step": 8755
    },
    {
      "epoch": 5.798675496688742,
      "grad_norm": 0.6907027126969195,
      "learning_rate": 1.3703654308785729e-06,
      "loss": 0.2031,
      "step": 8756
    },
    {
      "epoch": 5.799337748344371,
      "grad_norm": 0.7272030927217936,
      "learning_rate": 1.361378336813368e-06,
      "loss": 0.2451,
      "step": 8757
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.8679548329548463,
      "learning_rate": 1.3524206748993626e-06,
      "loss": 0.3047,
      "step": 8758
    },
    {
      "epoch": 5.800662251655629,
      "grad_norm": 0.8335746842737348,
      "learning_rate": 1.3434924469102382e-06,
      "loss": 0.2656,
      "step": 8759
    },
    {
      "epoch": 5.801324503311259,
      "grad_norm": 0.9611762232387562,
      "learning_rate": 1.3345936546138968e-06,
      "loss": 0.332,
      "step": 8760
    },
    {
      "epoch": 5.8019867549668875,
      "grad_norm": 0.7425841831413459,
      "learning_rate": 1.3257242997724128e-06,
      "loss": 0.2217,
      "step": 8761
    },
    {
      "epoch": 5.8026490066225165,
      "grad_norm": 0.716892351546579,
      "learning_rate": 1.3168843841420141e-06,
      "loss": 0.2246,
      "step": 8762
    },
    {
      "epoch": 5.803311258278145,
      "grad_norm": 0.8112737208608828,
      "learning_rate": 1.3080739094730841e-06,
      "loss": 0.2578,
      "step": 8763
    },
    {
      "epoch": 5.803973509933774,
      "grad_norm": 0.7355726184556421,
      "learning_rate": 1.299292877510244e-06,
      "loss": 0.2539,
      "step": 8764
    },
    {
      "epoch": 5.804635761589404,
      "grad_norm": 0.8615908112274114,
      "learning_rate": 1.2905412899921864e-06,
      "loss": 0.2891,
      "step": 8765
    },
    {
      "epoch": 5.805298013245033,
      "grad_norm": 0.7252477252003586,
      "learning_rate": 1.2818191486518748e-06,
      "loss": 0.2168,
      "step": 8766
    },
    {
      "epoch": 5.805960264900662,
      "grad_norm": 0.7061209259066067,
      "learning_rate": 1.273126455216361e-06,
      "loss": 0.2246,
      "step": 8767
    },
    {
      "epoch": 5.806622516556291,
      "grad_norm": 0.8034232511276235,
      "learning_rate": 1.2644632114069186e-06,
      "loss": 0.2539,
      "step": 8768
    },
    {
      "epoch": 5.807284768211921,
      "grad_norm": 0.7840057654053251,
      "learning_rate": 1.2558294189389251e-06,
      "loss": 0.2324,
      "step": 8769
    },
    {
      "epoch": 5.80794701986755,
      "grad_norm": 0.7159949058668483,
      "learning_rate": 1.2472250795220295e-06,
      "loss": 0.2168,
      "step": 8770
    },
    {
      "epoch": 5.808609271523179,
      "grad_norm": 0.6974423523440565,
      "learning_rate": 1.238650194859969e-06,
      "loss": 0.2139,
      "step": 8771
    },
    {
      "epoch": 5.809271523178808,
      "grad_norm": 0.7255283196960042,
      "learning_rate": 1.2301047666506359e-06,
      "loss": 0.2266,
      "step": 8772
    },
    {
      "epoch": 5.8099337748344375,
      "grad_norm": 0.6284756009842309,
      "learning_rate": 1.2215887965861592e-06,
      "loss": 0.207,
      "step": 8773
    },
    {
      "epoch": 5.8105960264900665,
      "grad_norm": 0.7950900473568635,
      "learning_rate": 1.2131022863528073e-06,
      "loss": 0.2168,
      "step": 8774
    },
    {
      "epoch": 5.811258278145695,
      "grad_norm": 0.7916523460458754,
      "learning_rate": 1.204645237630969e-06,
      "loss": 0.2793,
      "step": 8775
    },
    {
      "epoch": 5.811920529801324,
      "grad_norm": 0.7729154266479662,
      "learning_rate": 1.1962176520952715e-06,
      "loss": 0.2559,
      "step": 8776
    },
    {
      "epoch": 5.812582781456953,
      "grad_norm": 0.7693537816635067,
      "learning_rate": 1.1878195314144468e-06,
      "loss": 0.2227,
      "step": 8777
    },
    {
      "epoch": 5.813245033112583,
      "grad_norm": 0.7075189187198534,
      "learning_rate": 1.1794508772514478e-06,
      "loss": 0.2393,
      "step": 8778
    },
    {
      "epoch": 5.813907284768212,
      "grad_norm": 0.7660800578841785,
      "learning_rate": 1.1711116912633322e-06,
      "loss": 0.2266,
      "step": 8779
    },
    {
      "epoch": 5.814569536423841,
      "grad_norm": 0.7162288573551453,
      "learning_rate": 1.1628019751013794e-06,
      "loss": 0.2285,
      "step": 8780
    },
    {
      "epoch": 5.81523178807947,
      "grad_norm": 0.7226595902499757,
      "learning_rate": 1.1545217304109899e-06,
      "loss": 0.2139,
      "step": 8781
    },
    {
      "epoch": 5.815894039735099,
      "grad_norm": 0.7520538963404279,
      "learning_rate": 1.146270958831752e-06,
      "loss": 0.2676,
      "step": 8782
    },
    {
      "epoch": 5.816556291390729,
      "grad_norm": 0.7222313559097812,
      "learning_rate": 1.1380496619974255e-06,
      "loss": 0.2256,
      "step": 8783
    },
    {
      "epoch": 5.817218543046358,
      "grad_norm": 0.6732014415654022,
      "learning_rate": 1.1298578415359083e-06,
      "loss": 0.1943,
      "step": 8784
    },
    {
      "epoch": 5.817880794701987,
      "grad_norm": 0.6377840055003942,
      "learning_rate": 1.121695499069286e-06,
      "loss": 0.1826,
      "step": 8785
    },
    {
      "epoch": 5.8185430463576155,
      "grad_norm": 0.7124829007955021,
      "learning_rate": 1.1135626362137829e-06,
      "loss": 0.2246,
      "step": 8786
    },
    {
      "epoch": 5.819205298013245,
      "grad_norm": 0.7000372832124223,
      "learning_rate": 1.1054592545797935e-06,
      "loss": 0.2227,
      "step": 8787
    },
    {
      "epoch": 5.819867549668874,
      "grad_norm": 0.8226500063042513,
      "learning_rate": 1.0973853557719015e-06,
      "loss": 0.2734,
      "step": 8788
    },
    {
      "epoch": 5.820529801324503,
      "grad_norm": 0.7252901621277582,
      "learning_rate": 1.0893409413888276e-06,
      "loss": 0.2041,
      "step": 8789
    },
    {
      "epoch": 5.821192052980132,
      "grad_norm": 0.8632500813753755,
      "learning_rate": 1.081326013023448e-06,
      "loss": 0.3242,
      "step": 8790
    },
    {
      "epoch": 5.821854304635762,
      "grad_norm": 0.6814361391193241,
      "learning_rate": 1.0733405722628097e-06,
      "loss": 0.2119,
      "step": 8791
    },
    {
      "epoch": 5.822516556291391,
      "grad_norm": 0.7222564501665105,
      "learning_rate": 1.0653846206881479e-06,
      "loss": 0.2295,
      "step": 8792
    },
    {
      "epoch": 5.82317880794702,
      "grad_norm": 0.6882162258878533,
      "learning_rate": 1.0574581598747855e-06,
      "loss": 0.2236,
      "step": 8793
    },
    {
      "epoch": 5.823841059602649,
      "grad_norm": 0.7172868666355022,
      "learning_rate": 1.049561191392284e-06,
      "loss": 0.2236,
      "step": 8794
    },
    {
      "epoch": 5.824503311258278,
      "grad_norm": 1.0630013153167195,
      "learning_rate": 1.0416937168043427e-06,
      "loss": 0.3848,
      "step": 8795
    },
    {
      "epoch": 5.825165562913908,
      "grad_norm": 1.008966116816096,
      "learning_rate": 1.0338557376687817e-06,
      "loss": 0.3457,
      "step": 8796
    },
    {
      "epoch": 5.825827814569537,
      "grad_norm": 0.7668784775182778,
      "learning_rate": 1.0260472555376264e-06,
      "loss": 0.249,
      "step": 8797
    },
    {
      "epoch": 5.8264900662251655,
      "grad_norm": 0.7244742605881664,
      "learning_rate": 1.0182682719570401e-06,
      "loss": 0.2432,
      "step": 8798
    },
    {
      "epoch": 5.8271523178807945,
      "grad_norm": 0.7196204347604843,
      "learning_rate": 1.010518788467357e-06,
      "loss": 0.2148,
      "step": 8799
    },
    {
      "epoch": 5.827814569536423,
      "grad_norm": 0.7634842518411153,
      "learning_rate": 1.0027988066030835e-06,
      "loss": 0.2148,
      "step": 8800
    },
    {
      "epoch": 5.828476821192053,
      "grad_norm": 0.7880795574879496,
      "learning_rate": 9.951083278928295e-07,
      "loss": 0.2314,
      "step": 8801
    },
    {
      "epoch": 5.829139072847682,
      "grad_norm": 0.769842612920095,
      "learning_rate": 9.874473538594108e-07,
      "loss": 0.2373,
      "step": 8802
    },
    {
      "epoch": 5.829801324503311,
      "grad_norm": 0.7312587351830787,
      "learning_rate": 9.798158860197803e-07,
      "loss": 0.2139,
      "step": 8803
    },
    {
      "epoch": 5.83046357615894,
      "grad_norm": 0.7369781543893006,
      "learning_rate": 9.722139258850791e-07,
      "loss": 0.2236,
      "step": 8804
    },
    {
      "epoch": 5.83112582781457,
      "grad_norm": 0.9024122470073395,
      "learning_rate": 9.646414749605702e-07,
      "loss": 0.3145,
      "step": 8805
    },
    {
      "epoch": 5.831788079470199,
      "grad_norm": 0.6839335656006554,
      "learning_rate": 9.570985347456705e-07,
      "loss": 0.1914,
      "step": 8806
    },
    {
      "epoch": 5.832450331125828,
      "grad_norm": 0.776911919304328,
      "learning_rate": 9.495851067339854e-07,
      "loss": 0.2559,
      "step": 8807
    },
    {
      "epoch": 5.833112582781457,
      "grad_norm": 0.6811499303306172,
      "learning_rate": 9.421011924132748e-07,
      "loss": 0.2129,
      "step": 8808
    },
    {
      "epoch": 5.8337748344370866,
      "grad_norm": 0.755387615201497,
      "learning_rate": 9.346467932654034e-07,
      "loss": 0.248,
      "step": 8809
    },
    {
      "epoch": 5.8344370860927155,
      "grad_norm": 0.6900663772662814,
      "learning_rate": 9.272219107664403e-07,
      "loss": 0.2109,
      "step": 8810
    },
    {
      "epoch": 5.835099337748344,
      "grad_norm": 0.8268035056234583,
      "learning_rate": 9.198265463866095e-07,
      "loss": 0.3301,
      "step": 8811
    },
    {
      "epoch": 5.835761589403973,
      "grad_norm": 0.6679224855409073,
      "learning_rate": 9.124607015902729e-07,
      "loss": 0.1973,
      "step": 8812
    },
    {
      "epoch": 5.836423841059602,
      "grad_norm": 0.6858376461150592,
      "learning_rate": 9.051243778359641e-07,
      "loss": 0.208,
      "step": 8813
    },
    {
      "epoch": 5.837086092715232,
      "grad_norm": 0.8138605455107571,
      "learning_rate": 8.978175765763207e-07,
      "loss": 0.3008,
      "step": 8814
    },
    {
      "epoch": 5.837748344370861,
      "grad_norm": 0.6213164073960943,
      "learning_rate": 8.905402992582023e-07,
      "loss": 0.1768,
      "step": 8815
    },
    {
      "epoch": 5.83841059602649,
      "grad_norm": 0.8104656426558374,
      "learning_rate": 8.83292547322606e-07,
      "loss": 0.2676,
      "step": 8816
    },
    {
      "epoch": 5.839072847682119,
      "grad_norm": 0.7674065282150813,
      "learning_rate": 8.76074322204634e-07,
      "loss": 0.2471,
      "step": 8817
    },
    {
      "epoch": 5.839735099337748,
      "grad_norm": 0.7722854098254966,
      "learning_rate": 8.688856253336096e-07,
      "loss": 0.2422,
      "step": 8818
    },
    {
      "epoch": 5.840397350993378,
      "grad_norm": 0.6787048824580929,
      "learning_rate": 8.617264581329441e-07,
      "loss": 0.1865,
      "step": 8819
    },
    {
      "epoch": 5.841059602649007,
      "grad_norm": 0.6700804943048952,
      "learning_rate": 8.54596822020287e-07,
      "loss": 0.1904,
      "step": 8820
    },
    {
      "epoch": 5.841721854304636,
      "grad_norm": 0.7815751969666703,
      "learning_rate": 8.474967184073255e-07,
      "loss": 0.2559,
      "step": 8821
    },
    {
      "epoch": 5.842384105960265,
      "grad_norm": 0.6604765104902409,
      "learning_rate": 8.404261487000019e-07,
      "loss": 0.21,
      "step": 8822
    },
    {
      "epoch": 5.843046357615894,
      "grad_norm": 0.7198683729863349,
      "learning_rate": 8.333851142983627e-07,
      "loss": 0.2207,
      "step": 8823
    },
    {
      "epoch": 5.843708609271523,
      "grad_norm": 0.6732531887742457,
      "learning_rate": 8.263736165966095e-07,
      "loss": 0.1904,
      "step": 8824
    },
    {
      "epoch": 5.844370860927152,
      "grad_norm": 0.7149857968683632,
      "learning_rate": 8.193916569830983e-07,
      "loss": 0.2168,
      "step": 8825
    },
    {
      "epoch": 5.845033112582781,
      "grad_norm": 0.825933570723644,
      "learning_rate": 8.124392368403399e-07,
      "loss": 0.2471,
      "step": 8826
    },
    {
      "epoch": 5.845695364238411,
      "grad_norm": 0.683456241245201,
      "learning_rate": 8.05516357544983e-07,
      "loss": 0.1973,
      "step": 8827
    },
    {
      "epoch": 5.84635761589404,
      "grad_norm": 0.7540353810219281,
      "learning_rate": 7.986230204678645e-07,
      "loss": 0.2617,
      "step": 8828
    },
    {
      "epoch": 5.847019867549669,
      "grad_norm": 0.6227638896625928,
      "learning_rate": 7.917592269739093e-07,
      "loss": 0.2021,
      "step": 8829
    },
    {
      "epoch": 5.847682119205298,
      "grad_norm": 0.7833528210932177,
      "learning_rate": 7.849249784222466e-07,
      "loss": 0.2578,
      "step": 8830
    },
    {
      "epoch": 5.848344370860927,
      "grad_norm": 0.7793830097070877,
      "learning_rate": 7.781202761661442e-07,
      "loss": 0.2324,
      "step": 8831
    },
    {
      "epoch": 5.849006622516557,
      "grad_norm": 0.6483548444734437,
      "learning_rate": 7.713451215529909e-07,
      "loss": 0.1924,
      "step": 8832
    },
    {
      "epoch": 5.849668874172186,
      "grad_norm": 0.7505857592029226,
      "learning_rate": 7.645995159243467e-07,
      "loss": 0.2598,
      "step": 8833
    },
    {
      "epoch": 5.850331125827815,
      "grad_norm": 0.8172479951880373,
      "learning_rate": 7.578834606159434e-07,
      "loss": 0.2441,
      "step": 8834
    },
    {
      "epoch": 5.8509933774834435,
      "grad_norm": 0.7942814547343147,
      "learning_rate": 7.511969569576171e-07,
      "loss": 0.2246,
      "step": 8835
    },
    {
      "epoch": 5.8516556291390724,
      "grad_norm": 0.8057364412369703,
      "learning_rate": 7.445400062733586e-07,
      "loss": 0.2676,
      "step": 8836
    },
    {
      "epoch": 5.852317880794702,
      "grad_norm": 0.817122951823774,
      "learning_rate": 7.379126098813637e-07,
      "loss": 0.2734,
      "step": 8837
    },
    {
      "epoch": 5.852980132450331,
      "grad_norm": 0.7220753659998398,
      "learning_rate": 7.313147690938826e-07,
      "loss": 0.209,
      "step": 8838
    },
    {
      "epoch": 5.85364238410596,
      "grad_norm": 0.8462978987368671,
      "learning_rate": 7.247464852174034e-07,
      "loss": 0.2852,
      "step": 8839
    },
    {
      "epoch": 5.854304635761589,
      "grad_norm": 0.7226459728010415,
      "learning_rate": 7.18207759552486e-07,
      "loss": 0.2354,
      "step": 8840
    },
    {
      "epoch": 5.854966887417218,
      "grad_norm": 0.785150262551872,
      "learning_rate": 7.116985933939112e-07,
      "loss": 0.2168,
      "step": 8841
    },
    {
      "epoch": 5.855629139072848,
      "grad_norm": 0.7103113242134287,
      "learning_rate": 7.052189880305314e-07,
      "loss": 0.2275,
      "step": 8842
    },
    {
      "epoch": 5.856291390728477,
      "grad_norm": 0.7323601748136837,
      "learning_rate": 6.987689447454203e-07,
      "loss": 0.2236,
      "step": 8843
    },
    {
      "epoch": 5.856953642384106,
      "grad_norm": 0.7462629165651844,
      "learning_rate": 6.92348464815723e-07,
      "loss": 0.2324,
      "step": 8844
    },
    {
      "epoch": 5.857615894039736,
      "grad_norm": 0.760776808726537,
      "learning_rate": 6.859575495128055e-07,
      "loss": 0.2236,
      "step": 8845
    },
    {
      "epoch": 5.8582781456953645,
      "grad_norm": 0.6777784020406845,
      "learning_rate": 6.79596200102106e-07,
      "loss": 0.2256,
      "step": 8846
    },
    {
      "epoch": 5.8589403973509935,
      "grad_norm": 0.8607366020066697,
      "learning_rate": 6.732644178432667e-07,
      "loss": 0.3125,
      "step": 8847
    },
    {
      "epoch": 5.859602649006622,
      "grad_norm": 0.936704976906442,
      "learning_rate": 6.669622039900513e-07,
      "loss": 0.3418,
      "step": 8848
    },
    {
      "epoch": 5.860264900662251,
      "grad_norm": 0.8323622978783115,
      "learning_rate": 6.606895597903617e-07,
      "loss": 0.2637,
      "step": 8849
    },
    {
      "epoch": 5.860927152317881,
      "grad_norm": 0.7143824425782659,
      "learning_rate": 6.544464864862708e-07,
      "loss": 0.2383,
      "step": 8850
    },
    {
      "epoch": 5.86158940397351,
      "grad_norm": 0.7392291070381241,
      "learning_rate": 6.482329853139401e-07,
      "loss": 0.25,
      "step": 8851
    },
    {
      "epoch": 5.862251655629139,
      "grad_norm": 0.7314031024492902,
      "learning_rate": 6.420490575037684e-07,
      "loss": 0.2598,
      "step": 8852
    },
    {
      "epoch": 5.862913907284768,
      "grad_norm": 0.6806140369204269,
      "learning_rate": 6.358947042802098e-07,
      "loss": 0.21,
      "step": 8853
    },
    {
      "epoch": 5.863576158940397,
      "grad_norm": 0.7971253164028728,
      "learning_rate": 6.297699268618894e-07,
      "loss": 0.2354,
      "step": 8854
    },
    {
      "epoch": 5.864238410596027,
      "grad_norm": 0.8815921132321878,
      "learning_rate": 6.236747264616038e-07,
      "loss": 0.2988,
      "step": 8855
    },
    {
      "epoch": 5.864900662251656,
      "grad_norm": 0.8093700739453823,
      "learning_rate": 6.176091042862708e-07,
      "loss": 0.2734,
      "step": 8856
    },
    {
      "epoch": 5.865562913907285,
      "grad_norm": 0.7230415535205806,
      "learning_rate": 6.115730615369463e-07,
      "loss": 0.2354,
      "step": 8857
    },
    {
      "epoch": 5.866225165562914,
      "grad_norm": 0.7250978244691269,
      "learning_rate": 6.05566599408841e-07,
      "loss": 0.2275,
      "step": 8858
    },
    {
      "epoch": 5.866887417218543,
      "grad_norm": 0.7538741973102229,
      "learning_rate": 5.995897190913035e-07,
      "loss": 0.2197,
      "step": 8859
    },
    {
      "epoch": 5.867549668874172,
      "grad_norm": 0.9003633042542556,
      "learning_rate": 5.936424217678038e-07,
      "loss": 0.3105,
      "step": 8860
    },
    {
      "epoch": 5.868211920529801,
      "grad_norm": 0.6372154700034829,
      "learning_rate": 5.877247086159998e-07,
      "loss": 0.1826,
      "step": 8861
    },
    {
      "epoch": 5.86887417218543,
      "grad_norm": 0.8012555531283756,
      "learning_rate": 5.818365808076708e-07,
      "loss": 0.2656,
      "step": 8862
    },
    {
      "epoch": 5.869536423841059,
      "grad_norm": 0.6777449116273798,
      "learning_rate": 5.759780395087176e-07,
      "loss": 0.2041,
      "step": 8863
    },
    {
      "epoch": 5.870198675496689,
      "grad_norm": 0.6604036290285409,
      "learning_rate": 5.701490858791958e-07,
      "loss": 0.2041,
      "step": 8864
    },
    {
      "epoch": 5.870860927152318,
      "grad_norm": 0.8786069880195906,
      "learning_rate": 5.64349721073315e-07,
      "loss": 0.3086,
      "step": 8865
    },
    {
      "epoch": 5.871523178807947,
      "grad_norm": 0.8724540945043789,
      "learning_rate": 5.585799462394236e-07,
      "loss": 0.2852,
      "step": 8866
    },
    {
      "epoch": 5.872185430463576,
      "grad_norm": 0.700663342969884,
      "learning_rate": 5.528397625199743e-07,
      "loss": 0.2236,
      "step": 8867
    },
    {
      "epoch": 5.872847682119206,
      "grad_norm": 0.754637812746562,
      "learning_rate": 5.471291710516246e-07,
      "loss": 0.2393,
      "step": 8868
    },
    {
      "epoch": 5.873509933774835,
      "grad_norm": 0.7473045072514733,
      "learning_rate": 5.414481729651199e-07,
      "loss": 0.2432,
      "step": 8869
    },
    {
      "epoch": 5.874172185430464,
      "grad_norm": 0.7307426481796577,
      "learning_rate": 5.357967693853438e-07,
      "loss": 0.2441,
      "step": 8870
    },
    {
      "epoch": 5.8748344370860925,
      "grad_norm": 0.7437882297845524,
      "learning_rate": 5.301749614313844e-07,
      "loss": 0.2461,
      "step": 8871
    },
    {
      "epoch": 5.8754966887417215,
      "grad_norm": 0.8303225964519886,
      "learning_rate": 5.245827502163847e-07,
      "loss": 0.2695,
      "step": 8872
    },
    {
      "epoch": 5.876158940397351,
      "grad_norm": 0.7629807179782687,
      "learning_rate": 5.190201368476754e-07,
      "loss": 0.2314,
      "step": 8873
    },
    {
      "epoch": 5.87682119205298,
      "grad_norm": 0.8345768921431717,
      "learning_rate": 5.134871224267256e-07,
      "loss": 0.2539,
      "step": 8874
    },
    {
      "epoch": 5.877483443708609,
      "grad_norm": 0.7180517480575306,
      "learning_rate": 5.079837080491422e-07,
      "loss": 0.2363,
      "step": 8875
    },
    {
      "epoch": 5.878145695364238,
      "grad_norm": 0.7623139790550106,
      "learning_rate": 5.025098948046369e-07,
      "loss": 0.2197,
      "step": 8876
    },
    {
      "epoch": 5.878807947019867,
      "grad_norm": 0.71260140977013,
      "learning_rate": 4.970656837771258e-07,
      "loss": 0.252,
      "step": 8877
    },
    {
      "epoch": 5.879470198675497,
      "grad_norm": 0.7070401338359867,
      "learning_rate": 4.916510760445969e-07,
      "loss": 0.2256,
      "step": 8878
    },
    {
      "epoch": 5.880132450331126,
      "grad_norm": 0.6636400467097158,
      "learning_rate": 4.862660726792089e-07,
      "loss": 0.1797,
      "step": 8879
    },
    {
      "epoch": 5.880794701986755,
      "grad_norm": 0.6409366375442862,
      "learning_rate": 4.809106747472591e-07,
      "loss": 0.1865,
      "step": 8880
    },
    {
      "epoch": 5.881456953642384,
      "grad_norm": 0.7512881296308841,
      "learning_rate": 4.7558488330916576e-07,
      "loss": 0.2344,
      "step": 8881
    },
    {
      "epoch": 5.882119205298014,
      "grad_norm": 0.6202887503837247,
      "learning_rate": 4.7028869941951875e-07,
      "loss": 0.1846,
      "step": 8882
    },
    {
      "epoch": 5.8827814569536425,
      "grad_norm": 0.9623708625567875,
      "learning_rate": 4.650221241270125e-07,
      "loss": 0.2793,
      "step": 8883
    },
    {
      "epoch": 5.8834437086092715,
      "grad_norm": 0.7449521470704898,
      "learning_rate": 4.5978515847449603e-07,
      "loss": 0.2412,
      "step": 8884
    },
    {
      "epoch": 5.8841059602649,
      "grad_norm": 0.8056556332346643,
      "learning_rate": 4.545778034989234e-07,
      "loss": 0.2656,
      "step": 8885
    },
    {
      "epoch": 5.88476821192053,
      "grad_norm": 0.8237688634946688,
      "learning_rate": 4.494000602314363e-07,
      "loss": 0.2676,
      "step": 8886
    },
    {
      "epoch": 5.885430463576159,
      "grad_norm": 0.7460964113136672,
      "learning_rate": 4.442519296972813e-07,
      "loss": 0.2344,
      "step": 8887
    },
    {
      "epoch": 5.886092715231788,
      "grad_norm": 0.7121795779345035,
      "learning_rate": 4.39133412915843e-07,
      "loss": 0.1982,
      "step": 8888
    },
    {
      "epoch": 5.886754966887417,
      "grad_norm": 0.8654211500698257,
      "learning_rate": 4.340445109006441e-07,
      "loss": 0.2715,
      "step": 8889
    },
    {
      "epoch": 5.887417218543046,
      "grad_norm": 0.6716284569433371,
      "learning_rate": 4.28985224659345e-07,
      "loss": 0.2227,
      "step": 8890
    },
    {
      "epoch": 5.888079470198676,
      "grad_norm": 0.759695587640271,
      "learning_rate": 4.2395555519376124e-07,
      "loss": 0.2617,
      "step": 8891
    },
    {
      "epoch": 5.888741721854305,
      "grad_norm": 0.8834560624636014,
      "learning_rate": 4.18955503499796e-07,
      "loss": 0.2734,
      "step": 8892
    },
    {
      "epoch": 5.889403973509934,
      "grad_norm": 0.6672665112428086,
      "learning_rate": 4.139850705675407e-07,
      "loss": 0.2061,
      "step": 8893
    },
    {
      "epoch": 5.890066225165563,
      "grad_norm": 0.8136199319104247,
      "learning_rate": 4.090442573811914e-07,
      "loss": 0.25,
      "step": 8894
    },
    {
      "epoch": 5.890728476821192,
      "grad_norm": 0.775986462916124,
      "learning_rate": 4.0413306491908216e-07,
      "loss": 0.2754,
      "step": 8895
    },
    {
      "epoch": 5.891390728476821,
      "grad_norm": 0.7325669466224816,
      "learning_rate": 3.992514941536851e-07,
      "loss": 0.2314,
      "step": 8896
    },
    {
      "epoch": 5.89205298013245,
      "grad_norm": 0.739937539137077,
      "learning_rate": 3.943995460516103e-07,
      "loss": 0.252,
      "step": 8897
    },
    {
      "epoch": 5.892715231788079,
      "grad_norm": 0.7104803824206763,
      "learning_rate": 3.895772215735893e-07,
      "loss": 0.2061,
      "step": 8898
    },
    {
      "epoch": 5.893377483443708,
      "grad_norm": 0.6503873762066041,
      "learning_rate": 3.847845216745249e-07,
      "loss": 0.1953,
      "step": 8899
    },
    {
      "epoch": 5.894039735099338,
      "grad_norm": 0.8189973613440557,
      "learning_rate": 3.8002144730339137e-07,
      "loss": 0.2988,
      "step": 8900
    },
    {
      "epoch": 5.894701986754967,
      "grad_norm": 0.6713833839486647,
      "learning_rate": 3.7528799940335086e-07,
      "loss": 0.2227,
      "step": 8901
    },
    {
      "epoch": 5.895364238410596,
      "grad_norm": 0.8079638182630176,
      "learning_rate": 3.7058417891167034e-07,
      "loss": 0.2754,
      "step": 8902
    },
    {
      "epoch": 5.896026490066225,
      "grad_norm": 0.7691070948660405,
      "learning_rate": 3.659099867597881e-07,
      "loss": 0.2471,
      "step": 8903
    },
    {
      "epoch": 5.896688741721855,
      "grad_norm": 0.7692274268185151,
      "learning_rate": 3.6126542387321375e-07,
      "loss": 0.2363,
      "step": 8904
    },
    {
      "epoch": 5.897350993377484,
      "grad_norm": 0.830780542318256,
      "learning_rate": 3.56650491171645e-07,
      "loss": 0.2559,
      "step": 8905
    },
    {
      "epoch": 5.898013245033113,
      "grad_norm": 0.6730085175320445,
      "learning_rate": 3.5206518956886756e-07,
      "loss": 0.2109,
      "step": 8906
    },
    {
      "epoch": 5.898675496688742,
      "grad_norm": 0.6704224132716022,
      "learning_rate": 3.4750951997285505e-07,
      "loss": 0.2061,
      "step": 8907
    },
    {
      "epoch": 5.8993377483443705,
      "grad_norm": 0.7898065347201618,
      "learning_rate": 3.4298348328566927e-07,
      "loss": 0.2314,
      "step": 8908
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.7944145808036633,
      "learning_rate": 3.3848708040351004e-07,
      "loss": 0.2715,
      "step": 8909
    },
    {
      "epoch": 5.900662251655629,
      "grad_norm": 0.7903490770275601,
      "learning_rate": 3.340203122167151e-07,
      "loss": 0.2451,
      "step": 8910
    },
    {
      "epoch": 5.901324503311258,
      "grad_norm": 0.8199336167792031,
      "learning_rate": 3.2958317960977696e-07,
      "loss": 0.2314,
      "step": 8911
    },
    {
      "epoch": 5.901986754966887,
      "grad_norm": 0.7106047653223646,
      "learning_rate": 3.251756834612762e-07,
      "loss": 0.2139,
      "step": 8912
    },
    {
      "epoch": 5.902649006622516,
      "grad_norm": 0.7526616204225441,
      "learning_rate": 3.2079782464396464e-07,
      "loss": 0.25,
      "step": 8913
    },
    {
      "epoch": 5.903311258278146,
      "grad_norm": 0.7944706242158172,
      "learning_rate": 3.164496040246822e-07,
      "loss": 0.25,
      "step": 8914
    },
    {
      "epoch": 5.903973509933775,
      "grad_norm": 0.6792056366835123,
      "learning_rate": 3.121310224644569e-07,
      "loss": 0.207,
      "step": 8915
    },
    {
      "epoch": 5.904635761589404,
      "grad_norm": 0.7420015346272435,
      "learning_rate": 3.0784208081840457e-07,
      "loss": 0.2314,
      "step": 8916
    },
    {
      "epoch": 5.905298013245033,
      "grad_norm": 0.711953305718888,
      "learning_rate": 3.035827799357793e-07,
      "loss": 0.2217,
      "step": 8917
    },
    {
      "epoch": 5.905960264900663,
      "grad_norm": 0.7561913498532168,
      "learning_rate": 2.9935312065998973e-07,
      "loss": 0.2412,
      "step": 8918
    },
    {
      "epoch": 5.906622516556292,
      "grad_norm": 0.7552253886960993,
      "learning_rate": 2.951531038285326e-07,
      "loss": 0.252,
      "step": 8919
    },
    {
      "epoch": 5.9072847682119205,
      "grad_norm": 0.7655349594693397,
      "learning_rate": 2.9098273027307604e-07,
      "loss": 0.2676,
      "step": 8920
    },
    {
      "epoch": 5.907947019867549,
      "grad_norm": 0.6124692977778528,
      "learning_rate": 2.868420008193928e-07,
      "loss": 0.2051,
      "step": 8921
    },
    {
      "epoch": 5.908609271523179,
      "grad_norm": 0.7442683914941131,
      "learning_rate": 2.8273091628739365e-07,
      "loss": 0.2266,
      "step": 8922
    },
    {
      "epoch": 5.909271523178808,
      "grad_norm": 0.7271860891384281,
      "learning_rate": 2.786494774911274e-07,
      "loss": 0.2236,
      "step": 8923
    },
    {
      "epoch": 5.909933774834437,
      "grad_norm": 0.7444541760562383,
      "learning_rate": 2.7459768523878103e-07,
      "loss": 0.2383,
      "step": 8924
    },
    {
      "epoch": 5.910596026490066,
      "grad_norm": 0.6595465670496837,
      "learning_rate": 2.7057554033261263e-07,
      "loss": 0.207,
      "step": 8925
    },
    {
      "epoch": 5.911258278145695,
      "grad_norm": 0.7503823419992977,
      "learning_rate": 2.6658304356910187e-07,
      "loss": 0.252,
      "step": 8926
    },
    {
      "epoch": 5.911920529801325,
      "grad_norm": 0.6541646241929427,
      "learning_rate": 2.626201957387664e-07,
      "loss": 0.2061,
      "step": 8927
    },
    {
      "epoch": 5.912582781456954,
      "grad_norm": 0.7714786689901325,
      "learning_rate": 2.586869976263284e-07,
      "loss": 0.2676,
      "step": 8928
    },
    {
      "epoch": 5.913245033112583,
      "grad_norm": 0.7076523115736347,
      "learning_rate": 2.547834500105983e-07,
      "loss": 0.1885,
      "step": 8929
    },
    {
      "epoch": 5.913907284768212,
      "grad_norm": 0.7177180222041608,
      "learning_rate": 2.509095536645078e-07,
      "loss": 0.2012,
      "step": 8930
    },
    {
      "epoch": 5.914569536423841,
      "grad_norm": 0.6845776312434835,
      "learning_rate": 2.4706530935514337e-07,
      "loss": 0.2148,
      "step": 8931
    },
    {
      "epoch": 5.9152317880794705,
      "grad_norm": 0.8254567230512837,
      "learning_rate": 2.4325071784371265e-07,
      "loss": 0.2715,
      "step": 8932
    },
    {
      "epoch": 5.915894039735099,
      "grad_norm": 0.8788908583005908,
      "learning_rate": 2.3946577988554484e-07,
      "loss": 0.332,
      "step": 8933
    },
    {
      "epoch": 5.916556291390728,
      "grad_norm": 0.6701938657144477,
      "learning_rate": 2.3571049623009042e-07,
      "loss": 0.1973,
      "step": 8934
    },
    {
      "epoch": 5.917218543046357,
      "grad_norm": 0.7826204246774577,
      "learning_rate": 2.3198486762095458e-07,
      "loss": 0.2539,
      "step": 8935
    },
    {
      "epoch": 5.917880794701987,
      "grad_norm": 0.7442361396024337,
      "learning_rate": 2.2828889479586387e-07,
      "loss": 0.2246,
      "step": 8936
    },
    {
      "epoch": 5.918543046357616,
      "grad_norm": 0.7795026848096929,
      "learning_rate": 2.2462257848663289e-07,
      "loss": 0.2354,
      "step": 8937
    },
    {
      "epoch": 5.919205298013245,
      "grad_norm": 0.7936331109401208,
      "learning_rate": 2.2098591941926425e-07,
      "loss": 0.2637,
      "step": 8938
    },
    {
      "epoch": 5.919867549668874,
      "grad_norm": 0.8852704433846098,
      "learning_rate": 2.1737891831383192e-07,
      "loss": 0.3066,
      "step": 8939
    },
    {
      "epoch": 5.920529801324504,
      "grad_norm": 0.7566920775364205,
      "learning_rate": 2.138015758845646e-07,
      "loss": 0.2334,
      "step": 8940
    },
    {
      "epoch": 5.921192052980133,
      "grad_norm": 0.6295109339026174,
      "learning_rate": 2.102538928398456e-07,
      "loss": 0.1885,
      "step": 8941
    },
    {
      "epoch": 5.921854304635762,
      "grad_norm": 0.6435344531999437,
      "learning_rate": 2.0673586988212976e-07,
      "loss": 0.1865,
      "step": 8942
    },
    {
      "epoch": 5.922516556291391,
      "grad_norm": 0.6851958843778964,
      "learning_rate": 2.032475077080431e-07,
      "loss": 0.2285,
      "step": 8943
    },
    {
      "epoch": 5.92317880794702,
      "grad_norm": 0.7597811284828032,
      "learning_rate": 1.9978880700831647e-07,
      "loss": 0.25,
      "step": 8944
    },
    {
      "epoch": 5.923841059602649,
      "grad_norm": 0.6853322428700527,
      "learning_rate": 1.96359768467802e-07,
      "loss": 0.2295,
      "step": 8945
    },
    {
      "epoch": 5.924503311258278,
      "grad_norm": 0.7301361681712475,
      "learning_rate": 1.9296039276550658e-07,
      "loss": 0.2441,
      "step": 8946
    },
    {
      "epoch": 5.925165562913907,
      "grad_norm": 0.8623177801798151,
      "learning_rate": 1.8959068057452508e-07,
      "loss": 0.291,
      "step": 8947
    },
    {
      "epoch": 5.925827814569536,
      "grad_norm": 0.7786931342149757,
      "learning_rate": 1.8625063256210714e-07,
      "loss": 0.2383,
      "step": 8948
    },
    {
      "epoch": 5.926490066225165,
      "grad_norm": 0.7546414152781539,
      "learning_rate": 1.829402493896237e-07,
      "loss": 0.2441,
      "step": 8949
    },
    {
      "epoch": 5.927152317880795,
      "grad_norm": 0.7945899176684996,
      "learning_rate": 1.7965953171256708e-07,
      "loss": 0.2832,
      "step": 8950
    },
    {
      "epoch": 5.927814569536424,
      "grad_norm": 0.7405557193678926,
      "learning_rate": 1.7640848018056765e-07,
      "loss": 0.2275,
      "step": 8951
    },
    {
      "epoch": 5.928476821192053,
      "grad_norm": 0.8294018284339708,
      "learning_rate": 1.7318709543734376e-07,
      "loss": 0.3008,
      "step": 8952
    },
    {
      "epoch": 5.929139072847682,
      "grad_norm": 0.8166003948271976,
      "learning_rate": 1.6999537812080188e-07,
      "loss": 0.2578,
      "step": 8953
    },
    {
      "epoch": 5.929801324503311,
      "grad_norm": 0.7577792984753371,
      "learning_rate": 1.6683332886291979e-07,
      "loss": 0.248,
      "step": 8954
    },
    {
      "epoch": 5.930463576158941,
      "grad_norm": 0.721794264715447,
      "learning_rate": 1.6370094828979664e-07,
      "loss": 0.2334,
      "step": 8955
    },
    {
      "epoch": 5.9311258278145695,
      "grad_norm": 0.6716264902714889,
      "learning_rate": 1.605982370217196e-07,
      "loss": 0.2295,
      "step": 8956
    },
    {
      "epoch": 5.9317880794701985,
      "grad_norm": 0.7211981896507037,
      "learning_rate": 1.5752519567304721e-07,
      "loss": 0.2051,
      "step": 8957
    },
    {
      "epoch": 5.932450331125828,
      "grad_norm": 0.7167560264605896,
      "learning_rate": 1.5448182485225945e-07,
      "loss": 0.2285,
      "step": 8958
    },
    {
      "epoch": 5.933112582781457,
      "grad_norm": 0.8019413025919309,
      "learning_rate": 1.5146812516200757e-07,
      "loss": 0.2393,
      "step": 8959
    },
    {
      "epoch": 5.933774834437086,
      "grad_norm": 0.7087708373301929,
      "learning_rate": 1.484840971990142e-07,
      "loss": 0.2295,
      "step": 8960
    },
    {
      "epoch": 5.934437086092715,
      "grad_norm": 0.644850188859022,
      "learning_rate": 1.4552974155417342e-07,
      "loss": 0.1963,
      "step": 8961
    },
    {
      "epoch": 5.935099337748344,
      "grad_norm": 0.8498228833874456,
      "learning_rate": 1.426050588124672e-07,
      "loss": 0.2832,
      "step": 8962
    },
    {
      "epoch": 5.935761589403974,
      "grad_norm": 0.8501592837948648,
      "learning_rate": 1.3971004955301568e-07,
      "loss": 0.2988,
      "step": 8963
    },
    {
      "epoch": 5.936423841059603,
      "grad_norm": 0.7226528478314316,
      "learning_rate": 1.368447143490603e-07,
      "loss": 0.2451,
      "step": 8964
    },
    {
      "epoch": 5.937086092715232,
      "grad_norm": 0.7848160160347092,
      "learning_rate": 1.3400905376798055e-07,
      "loss": 0.2754,
      "step": 8965
    },
    {
      "epoch": 5.937748344370861,
      "grad_norm": 0.6933778326261003,
      "learning_rate": 1.312030683712606e-07,
      "loss": 0.2061,
      "step": 8966
    },
    {
      "epoch": 5.93841059602649,
      "grad_norm": 0.7635571846794397,
      "learning_rate": 1.2842675871452267e-07,
      "loss": 0.2656,
      "step": 8967
    },
    {
      "epoch": 5.9390728476821195,
      "grad_norm": 0.6742102297033106,
      "learning_rate": 1.2568012534751038e-07,
      "loss": 0.21,
      "step": 8968
    },
    {
      "epoch": 5.9397350993377485,
      "grad_norm": 0.7167534303218163,
      "learning_rate": 1.229631688141053e-07,
      "loss": 0.2295,
      "step": 8969
    },
    {
      "epoch": 5.940397350993377,
      "grad_norm": 0.6898771344894374,
      "learning_rate": 1.2027588965226043e-07,
      "loss": 0.2266,
      "step": 8970
    },
    {
      "epoch": 5.941059602649006,
      "grad_norm": 0.8328464812824243,
      "learning_rate": 1.1761828839410014e-07,
      "loss": 0.252,
      "step": 8971
    },
    {
      "epoch": 5.941721854304635,
      "grad_norm": 0.8059568970953456,
      "learning_rate": 1.1499036556587015e-07,
      "loss": 0.2715,
      "step": 8972
    },
    {
      "epoch": 5.942384105960265,
      "grad_norm": 0.8818303296257425,
      "learning_rate": 1.1239212168792089e-07,
      "loss": 0.2695,
      "step": 8973
    },
    {
      "epoch": 5.943046357615894,
      "grad_norm": 0.724471518722728,
      "learning_rate": 1.0982355727474079e-07,
      "loss": 0.2314,
      "step": 8974
    },
    {
      "epoch": 5.943708609271523,
      "grad_norm": 0.7727995916615588,
      "learning_rate": 1.0728467283492304e-07,
      "loss": 0.2812,
      "step": 8975
    },
    {
      "epoch": 5.944370860927152,
      "grad_norm": 0.6947875783966496,
      "learning_rate": 1.0477546887121546e-07,
      "loss": 0.1982,
      "step": 8976
    },
    {
      "epoch": 5.945033112582782,
      "grad_norm": 0.8502345682682994,
      "learning_rate": 1.0229594588043732e-07,
      "loss": 0.3184,
      "step": 8977
    },
    {
      "epoch": 5.945695364238411,
      "grad_norm": 0.634425220976975,
      "learning_rate": 9.984610435359586e-08,
      "loss": 0.1934,
      "step": 8978
    },
    {
      "epoch": 5.94635761589404,
      "grad_norm": 1.151886437656613,
      "learning_rate": 9.742594477578636e-08,
      "loss": 0.4141,
      "step": 8979
    },
    {
      "epoch": 5.947019867549669,
      "grad_norm": 0.7204509985431728,
      "learning_rate": 9.503546762620884e-08,
      "loss": 0.2305,
      "step": 8980
    },
    {
      "epoch": 5.947682119205298,
      "grad_norm": 0.7320720243540163,
      "learning_rate": 9.267467337820134e-08,
      "loss": 0.2324,
      "step": 8981
    },
    {
      "epoch": 5.948344370860927,
      "grad_norm": 0.683142897070884,
      "learning_rate": 9.034356249925656e-08,
      "loss": 0.2002,
      "step": 8982
    },
    {
      "epoch": 5.949006622516556,
      "grad_norm": 0.7600911883561219,
      "learning_rate": 8.804213545093864e-08,
      "loss": 0.2246,
      "step": 8983
    },
    {
      "epoch": 5.949668874172185,
      "grad_norm": 0.7608084260086367,
      "learning_rate": 8.577039268894969e-08,
      "loss": 0.2422,
      "step": 8984
    },
    {
      "epoch": 5.950331125827814,
      "grad_norm": 0.8187098813903155,
      "learning_rate": 8.352833466314657e-08,
      "loss": 0.2637,
      "step": 8985
    },
    {
      "epoch": 5.950993377483444,
      "grad_norm": 0.7884743340572617,
      "learning_rate": 8.131596181747413e-08,
      "loss": 0.2598,
      "step": 8986
    },
    {
      "epoch": 5.951655629139073,
      "grad_norm": 0.7079191689803381,
      "learning_rate": 7.913327458999863e-08,
      "loss": 0.2305,
      "step": 8987
    },
    {
      "epoch": 5.952317880794702,
      "grad_norm": 0.7190076557478448,
      "learning_rate": 7.698027341292434e-08,
      "loss": 0.2373,
      "step": 8988
    },
    {
      "epoch": 5.952980132450331,
      "grad_norm": 0.661997952224156,
      "learning_rate": 7.485695871256026e-08,
      "loss": 0.2129,
      "step": 8989
    },
    {
      "epoch": 5.95364238410596,
      "grad_norm": 0.7830668521858762,
      "learning_rate": 7.276333090935338e-08,
      "loss": 0.249,
      "step": 8990
    },
    {
      "epoch": 5.95430463576159,
      "grad_norm": 0.8346824930967212,
      "learning_rate": 7.069939041787209e-08,
      "loss": 0.2334,
      "step": 8991
    },
    {
      "epoch": 5.954966887417219,
      "grad_norm": 0.7530102295444446,
      "learning_rate": 6.86651376467895e-08,
      "loss": 0.2217,
      "step": 8992
    },
    {
      "epoch": 5.9556291390728475,
      "grad_norm": 0.709389326019918,
      "learning_rate": 6.66605729989167e-08,
      "loss": 0.2275,
      "step": 8993
    },
    {
      "epoch": 5.9562913907284765,
      "grad_norm": 0.772154107130087,
      "learning_rate": 6.46856968711862e-08,
      "loss": 0.2266,
      "step": 8994
    },
    {
      "epoch": 5.956953642384106,
      "grad_norm": 0.7888967870102803,
      "learning_rate": 6.274050965463517e-08,
      "loss": 0.2773,
      "step": 8995
    },
    {
      "epoch": 5.957615894039735,
      "grad_norm": 0.6383178439909735,
      "learning_rate": 6.082501173443887e-08,
      "loss": 0.2012,
      "step": 8996
    },
    {
      "epoch": 5.958278145695364,
      "grad_norm": 0.8201207578033066,
      "learning_rate": 5.893920348987724e-08,
      "loss": 0.2578,
      "step": 8997
    },
    {
      "epoch": 5.958940397350993,
      "grad_norm": 0.6680759579425741,
      "learning_rate": 5.708308529438488e-08,
      "loss": 0.208,
      "step": 8998
    },
    {
      "epoch": 5.959602649006623,
      "grad_norm": 0.6923520621417799,
      "learning_rate": 5.525665751548447e-08,
      "loss": 0.2354,
      "step": 8999
    },
    {
      "epoch": 5.960264900662252,
      "grad_norm": 0.722502093629824,
      "learning_rate": 5.3459920514820065e-08,
      "loss": 0.2148,
      "step": 9000
    },
    {
      "epoch": 5.960927152317881,
      "grad_norm": 0.6002838986280469,
      "learning_rate": 5.169287464815708e-08,
      "loss": 0.1826,
      "step": 9001
    },
    {
      "epoch": 5.96158940397351,
      "grad_norm": 0.8085497432595289,
      "learning_rate": 4.995552026543226e-08,
      "loss": 0.2754,
      "step": 9002
    },
    {
      "epoch": 5.962251655629139,
      "grad_norm": 0.7009574085482572,
      "learning_rate": 4.8247857710620455e-08,
      "loss": 0.2148,
      "step": 9003
    },
    {
      "epoch": 5.9629139072847686,
      "grad_norm": 0.7387097878668679,
      "learning_rate": 4.656988732188449e-08,
      "loss": 0.2344,
      "step": 9004
    },
    {
      "epoch": 5.9635761589403975,
      "grad_norm": 0.6900575763483855,
      "learning_rate": 4.492160943145861e-08,
      "loss": 0.2051,
      "step": 9005
    },
    {
      "epoch": 5.964238410596026,
      "grad_norm": 0.7958175892699306,
      "learning_rate": 4.330302436573174e-08,
      "loss": 0.248,
      "step": 9006
    },
    {
      "epoch": 5.964900662251655,
      "grad_norm": 0.7228270617125553,
      "learning_rate": 4.171413244521416e-08,
      "loss": 0.2314,
      "step": 9007
    },
    {
      "epoch": 5.965562913907284,
      "grad_norm": 0.7947810738875899,
      "learning_rate": 4.01549339845042e-08,
      "loss": 0.2715,
      "step": 9008
    },
    {
      "epoch": 5.966225165562914,
      "grad_norm": 0.7546574346759855,
      "learning_rate": 3.86254292923549e-08,
      "loss": 0.2432,
      "step": 9009
    },
    {
      "epoch": 5.966887417218543,
      "grad_norm": 0.6110909130843617,
      "learning_rate": 3.7125618671624e-08,
      "loss": 0.1719,
      "step": 9010
    },
    {
      "epoch": 5.967549668874172,
      "grad_norm": 0.6475206357154525,
      "learning_rate": 3.5655502419290605e-08,
      "loss": 0.2109,
      "step": 9011
    },
    {
      "epoch": 5.968211920529801,
      "grad_norm": 0.7599164505975731,
      "learning_rate": 3.4215080826455186e-08,
      "loss": 0.2188,
      "step": 9012
    },
    {
      "epoch": 5.968874172185431,
      "grad_norm": 0.6196737687773664,
      "learning_rate": 3.280435417832294e-08,
      "loss": 0.1621,
      "step": 9013
    },
    {
      "epoch": 5.96953642384106,
      "grad_norm": 0.7950140765419488,
      "learning_rate": 3.142332275425374e-08,
      "loss": 0.2988,
      "step": 9014
    },
    {
      "epoch": 5.970198675496689,
      "grad_norm": 0.7900795565787098,
      "learning_rate": 3.007198682769552e-08,
      "loss": 0.252,
      "step": 9015
    },
    {
      "epoch": 5.970860927152318,
      "grad_norm": 0.6788574921060473,
      "learning_rate": 2.8750346666250868e-08,
      "loss": 0.2168,
      "step": 9016
    },
    {
      "epoch": 5.9715231788079475,
      "grad_norm": 0.8099272988068814,
      "learning_rate": 2.745840253157716e-08,
      "loss": 0.2754,
      "step": 9017
    },
    {
      "epoch": 5.972185430463576,
      "grad_norm": 0.7357799257356616,
      "learning_rate": 2.6196154679536395e-08,
      "loss": 0.2246,
      "step": 9018
    },
    {
      "epoch": 5.972847682119205,
      "grad_norm": 0.8013839698399275,
      "learning_rate": 2.4963603360045327e-08,
      "loss": 0.2656,
      "step": 9019
    },
    {
      "epoch": 5.973509933774834,
      "grad_norm": 0.6660548114662181,
      "learning_rate": 2.3760748817158726e-08,
      "loss": 0.2041,
      "step": 9020
    },
    {
      "epoch": 5.974172185430463,
      "grad_norm": 0.7349700601059688,
      "learning_rate": 2.2587591289086047e-08,
      "loss": 0.2637,
      "step": 9021
    },
    {
      "epoch": 5.974834437086093,
      "grad_norm": 0.667400940470393,
      "learning_rate": 2.1444131008091503e-08,
      "loss": 0.21,
      "step": 9022
    },
    {
      "epoch": 5.975496688741722,
      "grad_norm": 0.7559925456982503,
      "learning_rate": 2.0330368200610624e-08,
      "loss": 0.2324,
      "step": 9023
    },
    {
      "epoch": 5.976158940397351,
      "grad_norm": 0.7652894571373917,
      "learning_rate": 1.9246303087167016e-08,
      "loss": 0.2354,
      "step": 9024
    },
    {
      "epoch": 5.97682119205298,
      "grad_norm": 0.6945270360811201,
      "learning_rate": 1.8191935882438945e-08,
      "loss": 0.2217,
      "step": 9025
    },
    {
      "epoch": 5.977483443708609,
      "grad_norm": 0.6966980302305292,
      "learning_rate": 1.7167266795192757e-08,
      "loss": 0.2344,
      "step": 9026
    },
    {
      "epoch": 5.978145695364239,
      "grad_norm": 0.7081063474240175,
      "learning_rate": 1.6172296028332808e-08,
      "loss": 0.2295,
      "step": 9027
    },
    {
      "epoch": 5.978807947019868,
      "grad_norm": 0.8486262265249113,
      "learning_rate": 1.5207023778851522e-08,
      "loss": 0.2793,
      "step": 9028
    },
    {
      "epoch": 5.979470198675497,
      "grad_norm": 0.8455903720013623,
      "learning_rate": 1.4271450237895998e-08,
      "loss": 0.2852,
      "step": 9029
    },
    {
      "epoch": 5.9801324503311255,
      "grad_norm": 0.7789746189338085,
      "learning_rate": 1.3365575590734701e-08,
      "loss": 0.2656,
      "step": 9030
    },
    {
      "epoch": 5.980794701986755,
      "grad_norm": 0.6698492092125282,
      "learning_rate": 1.2489400016740814e-08,
      "loss": 0.1787,
      "step": 9031
    },
    {
      "epoch": 5.981456953642384,
      "grad_norm": 1.0380567101179006,
      "learning_rate": 1.1642923689375582e-08,
      "loss": 0.3359,
      "step": 9032
    },
    {
      "epoch": 5.982119205298013,
      "grad_norm": 0.6496764689730753,
      "learning_rate": 1.0826146776288237e-08,
      "loss": 0.2012,
      "step": 9033
    },
    {
      "epoch": 5.982781456953642,
      "grad_norm": 0.7098449100817591,
      "learning_rate": 1.0039069439182757e-08,
      "loss": 0.2383,
      "step": 9034
    },
    {
      "epoch": 5.983443708609272,
      "grad_norm": 0.8301135912109756,
      "learning_rate": 9.281691833917804e-09,
      "loss": 0.2637,
      "step": 9035
    },
    {
      "epoch": 5.984105960264901,
      "grad_norm": 0.8514348291367677,
      "learning_rate": 8.554014110473406e-09,
      "loss": 0.2617,
      "step": 9036
    },
    {
      "epoch": 5.98476821192053,
      "grad_norm": 0.7502464940641256,
      "learning_rate": 7.856036412917654e-09,
      "loss": 0.2236,
      "step": 9037
    },
    {
      "epoch": 5.985430463576159,
      "grad_norm": 0.8053979136224846,
      "learning_rate": 7.187758879489969e-09,
      "loss": 0.252,
      "step": 9038
    },
    {
      "epoch": 5.986092715231788,
      "grad_norm": 0.6914402351157083,
      "learning_rate": 6.5491816424845244e-09,
      "loss": 0.2217,
      "step": 9039
    },
    {
      "epoch": 5.986754966887418,
      "grad_norm": 0.7787907859863512,
      "learning_rate": 5.940304828366827e-09,
      "loss": 0.2695,
      "step": 9040
    },
    {
      "epoch": 5.9874172185430465,
      "grad_norm": 0.7673742816924171,
      "learning_rate": 5.361128557690442e-09,
      "loss": 0.2656,
      "step": 9041
    },
    {
      "epoch": 5.9880794701986755,
      "grad_norm": 0.7257990142145353,
      "learning_rate": 4.81165294514696e-09,
      "loss": 0.2461,
      "step": 9042
    },
    {
      "epoch": 5.988741721854304,
      "grad_norm": 0.6471936496950759,
      "learning_rate": 4.291878099516033e-09,
      "loss": 0.207,
      "step": 9043
    },
    {
      "epoch": 5.989403973509933,
      "grad_norm": 0.7031518017806361,
      "learning_rate": 3.8018041237652955e-09,
      "loss": 0.2246,
      "step": 9044
    },
    {
      "epoch": 5.990066225165563,
      "grad_norm": 0.7661618121761746,
      "learning_rate": 3.34143111488383e-09,
      "loss": 0.2461,
      "step": 9045
    },
    {
      "epoch": 5.990728476821192,
      "grad_norm": 0.7643603317054753,
      "learning_rate": 2.9107591640820106e-09,
      "loss": 0.2207,
      "step": 9046
    },
    {
      "epoch": 5.991390728476821,
      "grad_norm": 0.8316668984638713,
      "learning_rate": 2.5097883565916577e-09,
      "loss": 0.2871,
      "step": 9047
    },
    {
      "epoch": 5.99205298013245,
      "grad_norm": 0.6584138238965226,
      "learning_rate": 2.138518771832576e-09,
      "loss": 0.2139,
      "step": 9048
    },
    {
      "epoch": 5.99271523178808,
      "grad_norm": 0.9476265383849117,
      "learning_rate": 1.7969504833126314e-09,
      "loss": 0.3652,
      "step": 9049
    },
    {
      "epoch": 5.993377483443709,
      "grad_norm": 0.6243357553043218,
      "learning_rate": 1.4850835586610598e-09,
      "loss": 0.1904,
      "step": 9050
    },
    {
      "epoch": 5.994039735099338,
      "grad_norm": 0.7253004500732355,
      "learning_rate": 1.2029180596617727e-09,
      "loss": 0.2314,
      "step": 9051
    },
    {
      "epoch": 5.994701986754967,
      "grad_norm": 0.880250012640625,
      "learning_rate": 9.504540421534369e-10,
      "loss": 0.3047,
      "step": 9052
    },
    {
      "epoch": 5.9953642384105965,
      "grad_norm": 0.694671528865614,
      "learning_rate": 7.27691556129395e-10,
      "loss": 0.2344,
      "step": 9053
    },
    {
      "epoch": 5.9960264900662255,
      "grad_norm": 0.7138490846613068,
      "learning_rate": 5.346306457210125e-10,
      "loss": 0.207,
      "step": 9054
    },
    {
      "epoch": 5.996688741721854,
      "grad_norm": 0.7706952956004545,
      "learning_rate": 3.7127134914771706e-10,
      "loss": 0.2295,
      "step": 9055
    },
    {
      "epoch": 5.997350993377483,
      "grad_norm": 0.8234903775824498,
      "learning_rate": 2.376136987336519e-10,
      "loss": 0.2793,
      "step": 9056
    },
    {
      "epoch": 5.998013245033112,
      "grad_norm": 0.740101684295858,
      "learning_rate": 1.336577209742895e-10,
      "loss": 0.25,
      "step": 9057
    },
    {
      "epoch": 5.998675496688742,
      "grad_norm": 0.7083457684665877,
      "learning_rate": 5.940343645316481e-11,
      "loss": 0.2334,
      "step": 9058
    },
    {
      "epoch": 5.999337748344371,
      "grad_norm": 0.822543277060649,
      "learning_rate": 1.4850859841875063e-11,
      "loss": 0.2656,
      "step": 9059
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.9273724863004569,
      "learning_rate": 0.0,
      "loss": 0.3105,
      "step": 9060
    },
    {
      "epoch": 6.0,
      "eval_loss": 2.969489574432373,
      "eval_runtime": 34.0302,
      "eval_samples_per_second": 9.932,
      "eval_steps_per_second": 9.932,
      "step": 9060
    },
    {
      "epoch": 6.0,
      "step": 9060,
      "total_flos": 14820153753600.0,
      "train_loss": 1.7485878647557946,
      "train_runtime": 43424.1913,
      "train_samples_per_second": 0.209,
      "train_steps_per_second": 0.209
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 9060,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 14820153753600.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}