{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 30.0,
  "eval_steps": 500,
  "global_step": 9390,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 35.253448486328125,
      "eval_mse_0": 27.93633460998535,
      "eval_mse_1": 40.281654357910156,
      "eval_mse_2": 37.096221923828125,
      "eval_mse_3": 35.69955825805664,
      "eval_pcc_0": -0.12943421939487107,
      "eval_pcc_1": 0.11087851067313088,
      "eval_pcc_2": -0.01677494886701865,
      "eval_pcc_3": -0.06531007486316416,
      "eval_runtime": 125.3927,
      "eval_samples_per_second": 19.937,
      "eval_steps_per_second": 2.496,
      "step": 313
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.467518636847711e-06,
      "loss": 39.9026,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_loss": 24.858530044555664,
      "eval_mse_0": 16.21411895751953,
      "eval_mse_1": 29.686359405517578,
      "eval_mse_2": 26.701322555541992,
      "eval_mse_3": 26.832250595092773,
      "eval_pcc_0": -0.12953277782821082,
      "eval_pcc_1": 0.15263354753596575,
      "eval_pcc_2": -0.012667331337741387,
      "eval_pcc_3": -0.1740250623680245,
      "eval_runtime": 119.1615,
      "eval_samples_per_second": 20.98,
      "eval_steps_per_second": 2.627,
      "step": 626
    },
    {
      "epoch": 3.0,
      "eval_loss": 15.985442161560059,
      "eval_mse_0": 8.285962104797363,
      "eval_mse_1": 19.734556198120117,
      "eval_mse_2": 17.5140323638916,
      "eval_mse_3": 18.40717124938965,
      "eval_pcc_0": -0.046687682567983446,
      "eval_pcc_1": 0.2051045936246996,
      "eval_pcc_2": -0.026585957250347247,
      "eval_pcc_3": -0.17914015544544384,
      "eval_runtime": 120.0286,
      "eval_samples_per_second": 20.828,
      "eval_steps_per_second": 2.608,
      "step": 939
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.935037273695422e-06,
      "loss": 21.5881,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_loss": 9.054893493652344,
      "eval_mse_0": 4.047198295593262,
      "eval_mse_1": 11.194503784179688,
      "eval_mse_2": 9.912358283996582,
      "eval_mse_3": 11.065492630004883,
      "eval_pcc_0": 0.028015550873066734,
      "eval_pcc_1": 0.20084089515908607,
      "eval_pcc_2": -0.09479908506417078,
      "eval_pcc_3": -0.23376913241158687,
      "eval_runtime": 124.6145,
      "eval_samples_per_second": 20.062,
      "eval_steps_per_second": 2.512,
      "step": 1252
    },
    {
      "epoch": 4.79,
      "learning_rate": 8.402555910543132e-06,
      "loss": 9.6588,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "eval_loss": 4.534116268157959,
      "eval_mse_0": 2.5608792304992676,
      "eval_mse_1": 5.271782875061035,
      "eval_mse_2": 4.763275623321533,
      "eval_mse_3": 5.540524482727051,
      "eval_pcc_0": 0.18326387925089216,
      "eval_pcc_1": 0.29440170760540424,
      "eval_pcc_2": -0.023020681634067185,
      "eval_pcc_3": -0.18361875818361142,
      "eval_runtime": 123.2269,
      "eval_samples_per_second": 20.288,
      "eval_steps_per_second": 2.54,
      "step": 1565
    },
    {
      "epoch": 6.0,
      "eval_loss": 3.9766886234283447,
      "eval_mse_0": 3.775294065475464,
      "eval_mse_1": 4.050389289855957,
      "eval_mse_2": 3.85265851020813,
      "eval_mse_3": 4.228421688079834,
      "eval_pcc_0": 0.10950709526046151,
      "eval_pcc_1": 0.17878205237621464,
      "eval_pcc_2": 0.15411864066632502,
      "eval_pcc_3": 0.11296732255791778,
      "eval_runtime": 123.3863,
      "eval_samples_per_second": 20.262,
      "eval_steps_per_second": 2.537,
      "step": 1878
    },
    {
      "epoch": 6.39,
      "learning_rate": 7.870074547390843e-06,
      "loss": 3.4997,
      "step": 2000
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.9719064235687256,
      "eval_mse_0": 2.0252020359039307,
      "eval_mse_1": 1.9209357500076294,
      "eval_mse_2": 1.8631107807159424,
      "eval_mse_3": 2.078376054763794,
      "eval_pcc_0": 0.4697406407291476,
      "eval_pcc_1": 0.4714825247758767,
      "eval_pcc_2": 0.46683720988296845,
      "eval_pcc_3": 0.47567376150906243,
      "eval_runtime": 123.8915,
      "eval_samples_per_second": 20.179,
      "eval_steps_per_second": 2.526,
      "step": 2191
    },
    {
      "epoch": 7.99,
      "learning_rate": 7.337593184238552e-06,
      "loss": 1.5435,
      "step": 2500
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.7928707599639893,
      "eval_mse_0": 1.9913127422332764,
      "eval_mse_1": 1.6557341814041138,
      "eval_mse_2": 1.6641719341278076,
      "eval_mse_3": 1.8602606058120728,
      "eval_pcc_0": 0.49067170864681886,
      "eval_pcc_1": 0.48233546636694913,
      "eval_pcc_2": 0.47482439405251997,
      "eval_pcc_3": 0.48934525150382063,
      "eval_runtime": 123.1617,
      "eval_samples_per_second": 20.299,
      "eval_steps_per_second": 2.541,
      "step": 2504
    },
    {
      "epoch": 9.0,
      "eval_loss": 1.7764745950698853,
      "eval_mse_0": 1.91068434715271,
      "eval_mse_1": 1.6419528722763062,
      "eval_mse_2": 1.6560568809509277,
      "eval_mse_3": 1.8972089290618896,
      "eval_pcc_0": 0.4966193508589803,
      "eval_pcc_1": 0.4926786631014638,
      "eval_pcc_2": 0.4863374398117205,
      "eval_pcc_3": 0.49518858002390853,
      "eval_runtime": 118.3567,
      "eval_samples_per_second": 21.123,
      "eval_steps_per_second": 2.645,
      "step": 2817
    },
    {
      "epoch": 9.58,
      "learning_rate": 6.805111821086262e-06,
      "loss": 1.1265,
      "step": 3000
    },
    {
      "epoch": 10.0,
      "eval_loss": 1.501739263534546,
      "eval_mse_0": 1.6362265348434448,
      "eval_mse_1": 1.361143708229065,
      "eval_mse_2": 1.3952668905258179,
      "eval_mse_3": 1.614320158958435,
      "eval_pcc_0": 0.5806262984037475,
      "eval_pcc_1": 0.5939864858919732,
      "eval_pcc_2": 0.5797502168628982,
      "eval_pcc_3": 0.5833515543731059,
      "eval_runtime": 117.2862,
      "eval_samples_per_second": 21.315,
      "eval_steps_per_second": 2.669,
      "step": 3130
    },
    {
      "epoch": 11.0,
      "eval_loss": 1.3959943056106567,
      "eval_mse_0": 1.5413020849227905,
      "eval_mse_1": 1.2384960651397705,
      "eval_mse_2": 1.28584885597229,
      "eval_mse_3": 1.5183273553848267,
      "eval_pcc_0": 0.6110165037907326,
      "eval_pcc_1": 0.6366207620213354,
      "eval_pcc_2": 0.6233864115342613,
      "eval_pcc_3": 0.6147175651996913,
      "eval_runtime": 122.9223,
      "eval_samples_per_second": 20.338,
      "eval_steps_per_second": 2.546,
      "step": 3443
    },
    {
      "epoch": 11.18,
      "learning_rate": 6.272630457933972e-06,
      "loss": 0.9846,
      "step": 3500
    },
    {
      "epoch": 12.0,
      "eval_loss": 1.2620220184326172,
      "eval_mse_0": 1.4125484228134155,
      "eval_mse_1": 1.0894999504089355,
      "eval_mse_2": 1.1511507034301758,
      "eval_mse_3": 1.3948848247528076,
      "eval_pcc_0": 0.6386613163271186,
      "eval_pcc_1": 0.6823474627630972,
      "eval_pcc_2": 0.6630484829779204,
      "eval_pcc_3": 0.6485774820429591,
      "eval_runtime": 123.5626,
      "eval_samples_per_second": 20.233,
      "eval_steps_per_second": 2.533,
      "step": 3756
    },
    {
      "epoch": 12.78,
      "learning_rate": 5.7401490947816826e-06,
      "loss": 0.922,
      "step": 4000
    },
    {
      "epoch": 13.0,
      "eval_loss": 1.2597845792770386,
      "eval_mse_0": 1.4059348106384277,
      "eval_mse_1": 1.0759116411209106,
      "eval_mse_2": 1.1540552377700806,
      "eval_mse_3": 1.403237223625183,
      "eval_pcc_0": 0.6445167475467969,
      "eval_pcc_1": 0.6877805024993177,
      "eval_pcc_2": 0.6675128211257984,
      "eval_pcc_3": 0.6496152231146246,
      "eval_runtime": 124.9289,
      "eval_samples_per_second": 20.011,
      "eval_steps_per_second": 2.505,
      "step": 4069
    },
    {
      "epoch": 14.0,
      "eval_loss": 1.286251187324524,
      "eval_mse_0": 1.466301441192627,
      "eval_mse_1": 1.1047468185424805,
      "eval_mse_2": 1.1404074430465698,
      "eval_mse_3": 1.4335484504699707,
      "eval_pcc_0": 0.6362665135071693,
      "eval_pcc_1": 0.6924939693575668,
      "eval_pcc_2": 0.6764887019377555,
      "eval_pcc_3": 0.6482415782472952,
      "eval_runtime": 120.8305,
      "eval_samples_per_second": 20.69,
      "eval_steps_per_second": 2.59,
      "step": 4382
    },
    {
      "epoch": 14.38,
      "learning_rate": 5.207667731629393e-06,
      "loss": 0.8126,
      "step": 4500
    },
    {
      "epoch": 15.0,
      "eval_loss": 1.1709247827529907,
      "eval_mse_0": 1.3320404291152954,
      "eval_mse_1": 0.9946478605270386,
      "eval_mse_2": 1.050872564315796,
      "eval_mse_3": 1.3061367273330688,
      "eval_pcc_0": 0.6633085453601341,
      "eval_pcc_1": 0.7144007675516612,
      "eval_pcc_2": 0.699356776928208,
      "eval_pcc_3": 0.6753245217035396,
      "eval_runtime": 121.0479,
      "eval_samples_per_second": 20.653,
      "eval_steps_per_second": 2.586,
      "step": 4695
    },
    {
      "epoch": 15.97,
      "learning_rate": 4.675186368477103e-06,
      "loss": 0.7783,
      "step": 5000
    },
    {
      "epoch": 16.0,
      "eval_loss": 1.2495259046554565,
      "eval_mse_0": 1.4424058198928833,
      "eval_mse_1": 1.0512551069259644,
      "eval_mse_2": 1.1075762510299683,
      "eval_mse_3": 1.3968638181686401,
      "eval_pcc_0": 0.6352550449549746,
      "eval_pcc_1": 0.6967280653723789,
      "eval_pcc_2": 0.6792741476546669,
      "eval_pcc_3": 0.64945214920132,
      "eval_runtime": 121.0064,
      "eval_samples_per_second": 20.66,
      "eval_steps_per_second": 2.587,
      "step": 5008
    },
    {
      "epoch": 17.0,
      "eval_loss": 1.4459205865859985,
      "eval_mse_0": 1.6440175771713257,
      "eval_mse_1": 1.267660140991211,
      "eval_mse_2": 1.279537558555603,
      "eval_mse_3": 1.592469334602356,
      "eval_pcc_0": 0.6292826434289844,
      "eval_pcc_1": 0.6901424064792703,
      "eval_pcc_2": 0.6738298495150237,
      "eval_pcc_3": 0.6447228899623312,
      "eval_runtime": 114.6109,
      "eval_samples_per_second": 21.813,
      "eval_steps_per_second": 2.731,
      "step": 5321
    },
    {
      "epoch": 17.57,
      "learning_rate": 4.142705005324814e-06,
      "loss": 0.7406,
      "step": 5500
    },
    {
      "epoch": 18.0,
      "eval_loss": 1.2125182151794434,
      "eval_mse_0": 1.3871902227401733,
      "eval_mse_1": 1.0241502523422241,
      "eval_mse_2": 1.084183692932129,
      "eval_mse_3": 1.3545498847961426,
      "eval_pcc_0": 0.648592932654662,
      "eval_pcc_1": 0.7049778702998659,
      "eval_pcc_2": 0.6896955826045925,
      "eval_pcc_3": 0.6621401050383381,
      "eval_runtime": 119.8271,
      "eval_samples_per_second": 20.863,
      "eval_steps_per_second": 2.612,
      "step": 5634
    },
    {
      "epoch": 19.0,
      "eval_loss": 1.2266896963119507,
      "eval_mse_0": 1.3873472213745117,
      "eval_mse_1": 1.039366364479065,
      "eval_mse_2": 1.1103386878967285,
      "eval_mse_3": 1.369706630706787,
      "eval_pcc_0": 0.6531011068847803,
      "eval_pcc_1": 0.7109674417540542,
      "eval_pcc_2": 0.6965897285366162,
      "eval_pcc_3": 0.6668592560212646,
      "eval_runtime": 120.6159,
      "eval_samples_per_second": 20.727,
      "eval_steps_per_second": 2.595,
      "step": 5947
    },
    {
      "epoch": 19.17,
      "learning_rate": 3.6102236421725247e-06,
      "loss": 0.727,
      "step": 6000
    },
    {
      "epoch": 20.0,
      "eval_loss": 1.2268439531326294,
      "eval_mse_0": 1.396562933921814,
      "eval_mse_1": 1.057541847229004,
      "eval_mse_2": 1.0942394733428955,
      "eval_mse_3": 1.359034538269043,
      "eval_pcc_0": 0.6507776199758031,
      "eval_pcc_1": 0.7003761762305559,
      "eval_pcc_2": 0.6872739839173044,
      "eval_pcc_3": 0.6644769465648772,
      "eval_runtime": 119.4331,
      "eval_samples_per_second": 20.932,
      "eval_steps_per_second": 2.621,
      "step": 6260
    },
    {
      "epoch": 20.77,
      "learning_rate": 3.0777422790202343e-06,
      "loss": 0.6663,
      "step": 6500
    },
    {
      "epoch": 21.0,
      "eval_loss": 1.2125563621520996,
      "eval_mse_0": 1.4073569774627686,
      "eval_mse_1": 1.0193030834197998,
      "eval_mse_2": 1.0586122274398804,
      "eval_mse_3": 1.3649543523788452,
      "eval_pcc_0": 0.6471936491552426,
      "eval_pcc_1": 0.7111482932311268,
      "eval_pcc_2": 0.6984204198361936,
      "eval_pcc_3": 0.6627486264786455,
      "eval_runtime": 120.2217,
      "eval_samples_per_second": 20.795,
      "eval_steps_per_second": 2.604,
      "step": 6573
    },
    {
      "epoch": 22.0,
      "eval_loss": 1.2184520959854126,
      "eval_mse_0": 1.4079796075820923,
      "eval_mse_1": 1.0253150463104248,
      "eval_mse_2": 1.0776245594024658,
      "eval_mse_3": 1.3628888130187988,
      "eval_pcc_0": 0.6521598121302783,
      "eval_pcc_1": 0.7078635737733601,
      "eval_pcc_2": 0.6953593800918589,
      "eval_pcc_3": 0.6670740488150717,
      "eval_runtime": 121.07,
      "eval_samples_per_second": 20.649,
      "eval_steps_per_second": 2.585,
      "step": 6886
    },
    {
      "epoch": 22.36,
      "learning_rate": 2.5452609158679447e-06,
      "loss": 0.679,
      "step": 7000
    },
    {
      "epoch": 23.0,
      "eval_loss": 1.210086703300476,
      "eval_mse_0": 1.3777058124542236,
      "eval_mse_1": 1.025907278060913,
      "eval_mse_2": 1.0858049392700195,
      "eval_mse_3": 1.3509302139282227,
      "eval_pcc_0": 0.6573554933056146,
      "eval_pcc_1": 0.708297350110508,
      "eval_pcc_2": 0.6949233524759934,
      "eval_pcc_3": 0.6696772655430471,
      "eval_runtime": 121.1393,
      "eval_samples_per_second": 20.637,
      "eval_steps_per_second": 2.584,
      "step": 7199
    },
    {
      "epoch": 23.96,
      "learning_rate": 2.012779552715655e-06,
      "loss": 0.6528,
      "step": 7500
    },
    {
      "epoch": 24.0,
      "eval_loss": 1.2004283666610718,
      "eval_mse_0": 1.3663476705551147,
      "eval_mse_1": 1.0228403806686401,
      "eval_mse_2": 1.0746558904647827,
      "eval_mse_3": 1.3378705978393555,
      "eval_pcc_0": 0.6582689400749444,
      "eval_pcc_1": 0.7062594658983674,
      "eval_pcc_2": 0.693820643427176,
      "eval_pcc_3": 0.6701055929214624,
      "eval_runtime": 114.7453,
      "eval_samples_per_second": 21.787,
      "eval_steps_per_second": 2.728,
      "step": 7512
    },
    {
      "epoch": 25.0,
      "eval_loss": 1.2322232723236084,
      "eval_mse_0": 1.4174244403839111,
      "eval_mse_1": 1.048292636871338,
      "eval_mse_2": 1.0903046131134033,
      "eval_mse_3": 1.3728687763214111,
      "eval_pcc_0": 0.6477830243165981,
      "eval_pcc_1": 0.7000484427666047,
      "eval_pcc_2": 0.6879032701994684,
      "eval_pcc_3": 0.6619093166774149,
      "eval_runtime": 116.7561,
      "eval_samples_per_second": 21.412,
      "eval_steps_per_second": 2.681,
      "step": 7825
    },
    {
      "epoch": 25.56,
      "learning_rate": 1.4802981895633654e-06,
      "loss": 0.6595,
      "step": 8000
    },
    {
      "epoch": 26.0,
      "eval_loss": 1.2048563957214355,
      "eval_mse_0": 1.379071831703186,
      "eval_mse_1": 1.0233663320541382,
      "eval_mse_2": 1.0703182220458984,
      "eval_mse_3": 1.346666932106018,
      "eval_pcc_0": 0.6500023241358558,
      "eval_pcc_1": 0.7045484661520249,
      "eval_pcc_2": 0.6920802171890341,
      "eval_pcc_3": 0.6637560007475978,
      "eval_runtime": 121.8052,
      "eval_samples_per_second": 20.525,
      "eval_steps_per_second": 2.57,
      "step": 8138
    },
    {
      "epoch": 27.0,
      "eval_loss": 1.228061556816101,
      "eval_mse_0": 1.4074989557266235,
      "eval_mse_1": 1.0446228981018066,
      "eval_mse_2": 1.09303617477417,
      "eval_mse_3": 1.3670885562896729,
      "eval_pcc_0": 0.6502650614159742,
      "eval_pcc_1": 0.7012770306789152,
      "eval_pcc_2": 0.6889389448570339,
      "eval_pcc_3": 0.6642077424768584,
      "eval_runtime": 120.9832,
      "eval_samples_per_second": 20.664,
      "eval_steps_per_second": 2.587,
      "step": 8451
    },
    {
      "epoch": 27.16,
      "learning_rate": 9.478168264110757e-07,
      "loss": 0.6276,
      "step": 8500
    },
    {
      "epoch": 28.0,
      "eval_loss": 1.2095798254013062,
      "eval_mse_0": 1.3814703226089478,
      "eval_mse_1": 1.0255109071731567,
      "eval_mse_2": 1.0790736675262451,
      "eval_mse_3": 1.352264404296875,
      "eval_pcc_0": 0.6484841338961417,
      "eval_pcc_1": 0.7040187033038017,
      "eval_pcc_2": 0.6907324570253328,
      "eval_pcc_3": 0.6621290104932265,
      "eval_runtime": 121.0164,
      "eval_samples_per_second": 20.658,
      "eval_steps_per_second": 2.586,
      "step": 8764
    },
    {
      "epoch": 28.75,
      "learning_rate": 4.1533546325878595e-07,
      "loss": 0.6087,
      "step": 9000
    },
    {
      "epoch": 29.0,
      "eval_loss": 1.209524393081665,
      "eval_mse_0": 1.386349081993103,
      "eval_mse_1": 1.02353835105896,
      "eval_mse_2": 1.0750586986541748,
      "eval_mse_3": 1.353151798248291,
      "eval_pcc_0": 0.6495366534228407,
      "eval_pcc_1": 0.704988942154168,
      "eval_pcc_2": 0.6921612298465728,
      "eval_pcc_3": 0.663405210333282,
      "eval_runtime": 120.3378,
      "eval_samples_per_second": 20.775,
      "eval_steps_per_second": 2.601,
      "step": 9077
    },
    {
      "epoch": 30.0,
      "eval_loss": 1.207377552986145,
      "eval_mse_0": 1.383621096611023,
      "eval_mse_1": 1.0227596759796143,
      "eval_mse_2": 1.0722577571868896,
      "eval_mse_3": 1.3508676290512085,
      "eval_pcc_0": 0.6492893725954768,
      "eval_pcc_1": 0.7047307122301355,
      "eval_pcc_2": 0.6919477211029743,
      "eval_pcc_3": 0.6631503671275302,
      "eval_runtime": 120.172,
      "eval_samples_per_second": 20.804,
      "eval_steps_per_second": 2.605,
      "step": 9390
    },
    {
      "epoch": 30.0,
      "step": 9390,
      "total_flos": 0.0,
      "train_loss": 4.614258646127134,
      "train_runtime": 10767.5202,
      "train_samples_per_second": 6.965,
      "train_steps_per_second": 0.872
    }
  ],
  "logging_steps": 500,
  "max_steps": 9390,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 30,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}