diff --git "a/checkpoint-1527/trainer_state.json" "b/checkpoint-1527/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1527/trainer_state.json"
@@ -0,0 +1,10742 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9996726677577742,
+  "eval_steps": 382,
+  "global_step": 1527,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.687138080596924,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 3.5097,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 3.6327099800109863,
+      "eval_runtime": 39.1673,
+      "eval_samples_per_second": 32.859,
+      "eval_steps_per_second": 8.221,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.729796886444092,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 3.6634,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.679180145263672,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.5559,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.81653356552124,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.1536,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.388213634490967,
+      "learning_rate": 0.00015,
+      "loss": 2.3092,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6662285327911377,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.2283,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9162248373031616,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.6207,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3946017026901245,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.2942,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3801995813846588,
+      "learning_rate": 0.00027,
+      "loss": 0.1143,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.2290647178888321,
+      "learning_rate": 0.0003,
+      "loss": 0.1152,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.2698324918746948,
+      "learning_rate": 0.00029999996457265966,
+      "loss": 0.0984,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.15049245953559875,
+      "learning_rate": 0.00029999985829065547,
+      "loss": 0.0925,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7001833319664001,
+      "learning_rate": 0.0002999996811540376,
+      "loss": 0.1215,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.22832374274730682,
+      "learning_rate": 0.00029999943316288974,
+      "loss": 0.0997,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.1290595531463623,
+      "learning_rate": 0.00029999911431732894,
+      "loss": 0.0973,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3555549383163452,
+      "learning_rate": 0.00029999872461750597,
+      "loss": 0.1108,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.04830395057797432,
+      "learning_rate": 0.0002999982640636048,
+      "loss": 0.0994,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.2727436125278473,
+      "learning_rate": 0.00029999773265584304,
+      "loss": 0.1144,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.03478335589170456,
+      "learning_rate": 0.0002999971303944716,
+      "loss": 0.0945,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.133951798081398,
+      "learning_rate": 0.00029999645727977505,
+      "loss": 0.0928,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.20885471999645233,
+      "learning_rate": 0.0002999957133120714,
+      "loss": 0.1056,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.030896561220288277,
+      "learning_rate": 0.00029999489849171195,
+      "loss": 0.0985,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.0476481132209301,
+      "learning_rate": 0.0002999940128190817,
+      "loss": 0.0993,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2006714642047882,
+      "learning_rate": 0.00029999305629459895,
+      "loss": 0.0971,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.150727316737175,
+      "learning_rate": 0.0002999920289187155,
+      "loss": 0.1016,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.03271281719207764,
+      "learning_rate": 0.0002999909306919168,
+      "loss": 0.1002,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.08288753032684326,
+      "learning_rate": 0.0002999897616147214,
+      "loss": 0.1009,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2443581521511078,
+      "learning_rate": 0.0002999885216876816,
+      "loss": 0.1036,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.16865722835063934,
+      "learning_rate": 0.00029998721091138323,
+      "loss": 0.0965,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.19362947344779968,
+      "learning_rate": 0.0002999858292864453,
+      "loss": 0.0952,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.039490532130002975,
+      "learning_rate": 0.0002999843768135205,
+      "loss": 0.0967,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.15848855674266815,
+      "learning_rate": 0.0002999828534932949,
+      "loss": 0.093,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2813495695590973,
+      "learning_rate": 0.0002999812593264881,
+      "loss": 0.1052,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.03380066901445389,
+      "learning_rate": 0.00029997959431385314,
+      "loss": 0.0974,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.050066880881786346,
+      "learning_rate": 0.0002999778584561764,
+      "loss": 0.0972,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2120673805475235,
+      "learning_rate": 0.00029997605175427803,
+      "loss": 0.0965,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.11290993541479111,
+      "learning_rate": 0.0002999741742090113,
+      "loss": 0.099,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.2454652190208435,
+      "learning_rate": 0.00029997222582126313,
+      "loss": 0.0898,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.10817914456129074,
+      "learning_rate": 0.0002999702065919539,
+      "loss": 0.0887,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3510904014110565,
+      "learning_rate": 0.00029996811652203737,
+      "loss": 0.1107,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3444919288158417,
+      "learning_rate": 0.0002999659556125009,
+      "loss": 0.1113,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.21621473133563995,
+      "learning_rate": 0.0002999637238643651,
+      "loss": 0.0991,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.0429786741733551,
+      "learning_rate": 0.00029996142127868426,
+      "loss": 0.0976,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.04371911287307739,
+      "learning_rate": 0.000299959047856546,
+      "loss": 0.0969,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.17956386506557465,
+      "learning_rate": 0.00029995660359907154,
+      "loss": 0.1027,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.05985981971025467,
+      "learning_rate": 0.0002999540885074153,
+      "loss": 0.0911,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.057165782898664474,
+      "learning_rate": 0.00029995150258276546,
+      "loss": 0.0944,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.06133668124675751,
+      "learning_rate": 0.00029994884582634345,
+      "loss": 0.0936,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.13429470360279083,
+      "learning_rate": 0.0002999461182394042,
+      "loss": 0.0932,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.08454808592796326,
+      "learning_rate": 0.00029994331982323625,
+      "loss": 0.0849,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.152529776096344,
+      "learning_rate": 0.0002999404505791613,
+      "loss": 0.0742,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9239559173583984,
+      "learning_rate": 0.0002999375105085348,
+      "loss": 0.1266,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.15827079117298126,
+      "learning_rate": 0.0002999344996127455,
+      "loss": 0.0685,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.11372745782136917,
+      "learning_rate": 0.0002999314178932156,
+      "loss": 0.0853,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.11947692930698395,
+      "learning_rate": 0.00029992826535140093,
+      "loss": 0.0871,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.09731484949588776,
+      "learning_rate": 0.00029992504198879047,
+      "loss": 0.0799,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40479809045791626,
+      "learning_rate": 0.0002999217478069069,
+      "loss": 0.1119,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.10651114583015442,
+      "learning_rate": 0.00029991838280730635,
+      "loss": 0.0741,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.13227766752243042,
+      "learning_rate": 0.0002999149469915782,
+      "loss": 0.067,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2328774333000183,
+      "learning_rate": 0.0002999114403613454,
+      "loss": 0.0872,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.15303733944892883,
+      "learning_rate": 0.0002999078629182645,
+      "loss": 0.077,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3285676836967468,
+      "learning_rate": 0.0002999042146640252,
+      "loss": 0.087,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.1548561453819275,
+      "learning_rate": 0.00029990049560035093,
+      "loss": 0.0521,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.1792415827512741,
+      "learning_rate": 0.0002998967057289983,
+      "loss": 0.0591,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.29741746187210083,
+      "learning_rate": 0.0002998928450517577,
+      "loss": 0.0955,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2590031325817108,
+      "learning_rate": 0.0002998889135704527,
+      "loss": 0.0443,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.2152624875307083,
+      "learning_rate": 0.0002998849112869403,
+      "loss": 0.0656,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.1976858377456665,
+      "learning_rate": 0.0002998808382031111,
+      "loss": 0.0256,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4642391502857208,
+      "learning_rate": 0.00029987669432088917,
+      "loss": 0.074,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43813541531562805,
+      "learning_rate": 0.0002998724796422318,
+      "loss": 0.0344,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8069552183151245,
+      "learning_rate": 0.0002998681941691299,
+      "loss": 0.1559,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3986961841583252,
+      "learning_rate": 0.00029986383790360776,
+      "loss": 0.0504,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.19154639542102814,
+      "learning_rate": 0.00029985941084772317,
+      "loss": 0.0638,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2110302895307541,
+      "learning_rate": 0.0002998549130035673,
+      "loss": 0.071,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17988017201423645,
+      "learning_rate": 0.00029985034437326477,
+      "loss": 0.0798,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.15195637941360474,
+      "learning_rate": 0.0002998457049589736,
+      "loss": 0.0575,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2465752810239792,
+      "learning_rate": 0.0002998409947628854,
+      "loss": 0.0669,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.10329095274209976,
+      "learning_rate": 0.0002998362137872249,
+      "loss": 0.0483,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.21354705095291138,
+      "learning_rate": 0.00029983136203425064,
+      "loss": 0.0522,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1916392743587494,
+      "learning_rate": 0.00029982643950625436,
+      "loss": 0.0797,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.12721975147724152,
+      "learning_rate": 0.0002998214462055613,
+      "loss": 0.0368,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.29551053047180176,
+      "learning_rate": 0.0002998163821345301,
+      "loss": 0.094,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3058943748474121,
+      "learning_rate": 0.00029981124729555283,
+      "loss": 0.0358,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7026583552360535,
+      "learning_rate": 0.00029980604169105497,
+      "loss": 0.1386,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37371405959129333,
+      "learning_rate": 0.00029980076532349557,
+      "loss": 0.0748,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.28942474722862244,
+      "learning_rate": 0.00029979541819536695,
+      "loss": 0.1037,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1699017435312271,
+      "learning_rate": 0.0002997900003091949,
+      "loss": 0.0631,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1061767190694809,
+      "learning_rate": 0.0002997845116675386,
+      "loss": 0.0557,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.13656219840049744,
+      "learning_rate": 0.0002997789522729908,
+      "loss": 0.0637,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.09874790161848068,
+      "learning_rate": 0.00029977332212817746,
+      "loss": 0.0495,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.24591276049613953,
+      "learning_rate": 0.0002997676212357581,
+      "loss": 0.0559,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.14663195610046387,
+      "learning_rate": 0.0002997618495984256,
+      "loss": 0.0804,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.08905334770679474,
+      "learning_rate": 0.0002997560072189062,
+      "loss": 0.0498,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.12921252846717834,
+      "learning_rate": 0.00029975009409995986,
+      "loss": 0.0365,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.08008511364459991,
+      "learning_rate": 0.0002997441102443795,
+      "loss": 0.03,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.2947149872779846,
+      "learning_rate": 0.0002997380556549918,
+      "loss": 0.0698,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3243441581726074,
+      "learning_rate": 0.0002997319303346567,
+      "loss": 0.0564,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.28058576583862305,
+      "learning_rate": 0.00029972573428626757,
+      "loss": 0.1262,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40957021713256836,
+      "learning_rate": 0.0002997194675127512,
+      "loss": 0.0471,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.12092690169811249,
+      "learning_rate": 0.00029971313001706787,
+      "loss": 0.0574,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.380398154258728,
+      "learning_rate": 0.0002997067218022111,
+      "loss": 0.1148,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13584262132644653,
+      "learning_rate": 0.0002997002428712079,
+      "loss": 0.0299,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13165581226348877,
+      "learning_rate": 0.00029969369322711874,
+      "loss": 0.0602,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1055503860116005,
+      "learning_rate": 0.00029968707287303744,
+      "loss": 0.0404,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.09600503742694855,
+      "learning_rate": 0.00029968038181209114,
+      "loss": 0.0497,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.05941639468073845,
+      "learning_rate": 0.0002996736200474406,
+      "loss": 0.0456,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1557297259569168,
+      "learning_rate": 0.0002996667875822797,
+      "loss": 0.077,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.14879021048545837,
+      "learning_rate": 0.00029965988441983595,
+      "loss": 0.0554,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13067294657230377,
+      "learning_rate": 0.00029965291056337006,
+      "loss": 0.0357,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.15178795158863068,
+      "learning_rate": 0.00029964586601617633,
+      "loss": 0.0433,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1176379844546318,
+      "learning_rate": 0.0002996387507815823,
+      "loss": 0.0432,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.048378992825746536,
+      "learning_rate": 0.000299631564862949,
+      "loss": 0.0338,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.09883740544319153,
+      "learning_rate": 0.0002996243082636708,
+      "loss": 0.0475,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.16062304377555847,
+      "learning_rate": 0.0002996169809871754,
+      "loss": 0.0595,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.06556422263383865,
+      "learning_rate": 0.00029960958303692397,
+      "loss": 0.0326,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7436458468437195,
+      "learning_rate": 0.000299602114416411,
+      "loss": 0.0512,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.12153153866529465,
+      "learning_rate": 0.00029959457512916454,
+      "loss": 0.0448,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.21684418618679047,
+      "learning_rate": 0.0002995869651787458,
+      "loss": 0.0754,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.13978178799152374,
+      "learning_rate": 0.0002995792845687494,
+      "loss": 0.03,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.08695519715547562,
+      "learning_rate": 0.0002995715333028034,
+      "loss": 0.0156,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2607383131980896,
+      "learning_rate": 0.0002995637113845693,
+      "loss": 0.0933,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.08398541808128357,
+      "learning_rate": 0.0002995558188177418,
+      "loss": 0.0368,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.14658145606517792,
+      "learning_rate": 0.0002995478556060492,
+      "loss": 0.0593,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.09054147452116013,
+      "learning_rate": 0.00029953982175325293,
+      "loss": 0.042,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.17315314710140228,
+      "learning_rate": 0.0002995317172631479,
+      "loss": 0.0754,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.20856395363807678,
+      "learning_rate": 0.0002995235421395624,
+      "loss": 0.0537,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.17539943754673004,
+      "learning_rate": 0.0002995152963863581,
+      "loss": 0.045,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1361098289489746,
+      "learning_rate": 0.00029950698000743,
+      "loss": 0.0622,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.05299444869160652,
+      "learning_rate": 0.00029949859300670644,
+      "loss": 0.0548,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.19711115956306458,
+      "learning_rate": 0.0002994901353881491,
+      "loss": 0.0721,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1288406252861023,
+      "learning_rate": 0.0002994816071557532,
+      "loss": 0.0408,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.08221332728862762,
+      "learning_rate": 0.000299473008313547,
+      "loss": 0.0526,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1506081223487854,
+      "learning_rate": 0.00029946433886559237,
+      "loss": 0.0542,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.293639600276947,
+      "learning_rate": 0.00029945559881598444,
+      "loss": 0.0769,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.06451396644115448,
+      "learning_rate": 0.0002994467881688517,
+      "loss": 0.0417,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2765437662601471,
+      "learning_rate": 0.00029943790692835604,
+      "loss": 0.0617,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.12035606801509857,
+      "learning_rate": 0.00029942895509869254,
+      "loss": 0.0429,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.09559385478496552,
+      "learning_rate": 0.0002994199326840898,
+      "loss": 0.044,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.13433387875556946,
+      "learning_rate": 0.00029941083968880965,
+      "loss": 0.036,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1325090080499649,
+      "learning_rate": 0.0002994016761171474,
+      "loss": 0.0762,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.19197365641593933,
+      "learning_rate": 0.00029939244197343143,
+      "loss": 0.0587,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.09238675236701965,
+      "learning_rate": 0.00029938313726202376,
+      "loss": 0.0262,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2584728002548218,
+      "learning_rate": 0.0002993737619873195,
+      "loss": 0.0382,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.30280745029449463,
+      "learning_rate": 0.00029936431615374727,
+      "loss": 0.0448,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41464564204216003,
+      "learning_rate": 0.00029935479976576896,
+      "loss": 0.0676,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4580010175704956,
+      "learning_rate": 0.00029934521282787974,
+      "loss": 0.1366,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1701657474040985,
+      "learning_rate": 0.0002993355553446081,
+      "loss": 0.0844,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.10784261673688889,
+      "learning_rate": 0.000299325827320516,
+      "loss": 0.0211,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.08266110718250275,
+      "learning_rate": 0.0002993160287601984,
+      "loss": 0.0181,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.20068615674972534,
+      "learning_rate": 0.00029930615966828407,
+      "loss": 0.0582,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.14237689971923828,
+      "learning_rate": 0.0002992962200494347,
+      "loss": 0.0549,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.09671233594417572,
+      "learning_rate": 0.0002992862099083453,
+      "loss": 0.0368,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.11356969177722931,
+      "learning_rate": 0.00029927612924974455,
+      "loss": 0.0851,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.17435969412326813,
+      "learning_rate": 0.00029926597807839394,
+      "loss": 0.0869,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.09785137325525284,
+      "learning_rate": 0.00029925575639908866,
+      "loss": 0.0463,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.143271341919899,
+      "learning_rate": 0.0002992454642166571,
+      "loss": 0.0532,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1381101906299591,
+      "learning_rate": 0.0002992351015359608,
+      "loss": 0.0512,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.0688018947839737,
+      "learning_rate": 0.0002992246683618948,
+      "loss": 0.0188,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.18138591945171356,
+      "learning_rate": 0.0002992141646993874,
+      "loss": 0.0737,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.0729256346821785,
+      "learning_rate": 0.0002992035905534001,
+      "loss": 0.0194,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.15414761006832123,
+      "learning_rate": 0.0002991929459289277,
+      "loss": 0.0412,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2506199777126312,
+      "learning_rate": 0.00029918223083099846,
+      "loss": 0.0789,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.16611520946025848,
+      "learning_rate": 0.00029917144526467375,
+      "loss": 0.046,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1828208565711975,
+      "learning_rate": 0.00029916058923504826,
+      "loss": 0.0324,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.08737993985414505,
+      "learning_rate": 0.00029914966274725006,
+      "loss": 0.0177,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.20271027088165283,
+      "learning_rate": 0.00029913866580644037,
+      "loss": 0.0455,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.04210209473967552,
+      "learning_rate": 0.00029912759841781383,
+      "loss": 0.0063,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.09085400402545929,
+      "learning_rate": 0.00029911646058659825,
+      "loss": 0.0174,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.18242572247982025,
+      "learning_rate": 0.00029910525231805466,
+      "loss": 0.053,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2796941101551056,
+      "learning_rate": 0.0002990939736174776,
+      "loss": 0.0348,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.18838226795196533,
+      "learning_rate": 0.00029908262449019463,
+      "loss": 0.0583,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.03574841469526291,
+      "learning_rate": 0.00029907120494156674,
+      "loss": 0.0058,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.18582922220230103,
+      "learning_rate": 0.00029905971497698805,
+      "loss": 0.0571,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.12871672213077545,
+      "learning_rate": 0.00029904815460188604,
+      "loss": 0.0618,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.0590621717274189,
+      "learning_rate": 0.00029903652382172143,
+      "loss": 0.0107,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.07922167330980301,
+      "learning_rate": 0.00029902482264198817,
+      "loss": 0.035,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3096056878566742,
+      "learning_rate": 0.0002990130510682135,
+      "loss": 0.0782,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1896304190158844,
+      "learning_rate": 0.00029900120910595783,
+      "loss": 0.036,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.11776513606309891,
+      "learning_rate": 0.000298989296760815,
+      "loss": 0.0521,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.11616750061511993,
+      "learning_rate": 0.00029897731403841194,
+      "loss": 0.0275,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.20179390907287598,
+      "learning_rate": 0.0002989652609444088,
+      "loss": 0.0514,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.14983738958835602,
+      "learning_rate": 0.00029895313748449907,
+      "loss": 0.077,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12123002856969833,
+      "learning_rate": 0.0002989409436644095,
+      "loss": 0.0485,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.314486026763916,
+      "learning_rate": 0.0002989286794898999,
+      "loss": 0.0931,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.132719025015831,
+      "learning_rate": 0.0002989163449667636,
+      "loss": 0.047,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.07938767969608307,
+      "learning_rate": 0.00029890394010082677,
+      "loss": 0.0364,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.08216488361358643,
+      "learning_rate": 0.00029889146489794926,
+      "loss": 0.0299,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.19339217245578766,
+      "learning_rate": 0.00029887891936402375,
+      "loss": 0.0408,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.30395349860191345,
+      "learning_rate": 0.0002988663035049763,
+      "loss": 0.0865,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.21264804899692535,
+      "learning_rate": 0.0002988536173267663,
+      "loss": 0.0584,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1590937227010727,
+      "learning_rate": 0.0002988408608353862,
+      "loss": 0.0442,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13069725036621094,
+      "learning_rate": 0.00029882803403686177,
+      "loss": 0.0416,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1968701034784317,
+      "learning_rate": 0.0002988151369372518,
+      "loss": 0.0586,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1478463113307953,
+      "learning_rate": 0.00029880216954264856,
+      "loss": 0.0595,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.06919383257627487,
+      "learning_rate": 0.0002987891318591773,
+      "loss": 0.0239,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.11905679851770401,
+      "learning_rate": 0.0002987760238929966,
+      "loss": 0.0345,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14240068197250366,
+      "learning_rate": 0.00029876284565029816,
+      "loss": 0.0467,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.16097158193588257,
+      "learning_rate": 0.000298749597137307,
+      "loss": 0.0554,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15597470104694366,
+      "learning_rate": 0.0002987362783602812,
+      "loss": 0.054,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.10321896523237228,
+      "learning_rate": 0.000298722889325512,
+      "loss": 0.0432,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.128427192568779,
+      "learning_rate": 0.000298709430039324,
+      "loss": 0.0315,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.11706223338842392,
+      "learning_rate": 0.00029869590050807487,
+      "loss": 0.0359,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15359801054000854,
+      "learning_rate": 0.0002986823007381555,
+      "loss": 0.034,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.10363847017288208,
+      "learning_rate": 0.0002986686307359899,
+      "loss": 0.0261,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12338493019342422,
+      "learning_rate": 0.0002986548905080353,
+      "loss": 0.0287,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.16201013326644897,
+      "learning_rate": 0.00029864108006078205,
+      "loss": 0.0173,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.04950540140271187,
+      "learning_rate": 0.00029862719940075387,
+      "loss": 0.0098,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.20930823683738708,
+      "learning_rate": 0.0002986132485345073,
+      "loss": 0.0652,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12760238349437714,
+      "learning_rate": 0.0002985992274686324,
+      "loss": 0.0342,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2107914686203003,
+      "learning_rate": 0.00029858513620975216,
+      "loss": 0.015,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.21169154345989227,
+      "learning_rate": 0.0002985709747645227,
+      "loss": 0.072,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.18555670976638794,
+      "learning_rate": 0.00029855674313963355,
+      "loss": 0.0359,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1801125705242157,
+      "learning_rate": 0.00029854244134180707,
+      "loss": 0.038,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.10735122859477997,
+      "learning_rate": 0.000298528069377799,
+      "loss": 0.037,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.20155467092990875,
+      "learning_rate": 0.0002985136272543982,
+      "loss": 0.0505,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1130833774805069,
+      "learning_rate": 0.0002984991149784265,
+      "loss": 0.0202,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1932414174079895,
+      "learning_rate": 0.00029848453255673906,
+      "loss": 0.0803,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.18907181918621063,
+      "learning_rate": 0.0002984698799962241,
+      "loss": 0.0562,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.11439274251461029,
+      "learning_rate": 0.0002984551573038029,
+      "loss": 0.0474,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.19350704550743103,
+      "learning_rate": 0.00029844036448643,
+      "loss": 0.0335,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.19873294234275818,
+      "learning_rate": 0.000298425501551093,
+      "loss": 0.0616,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2024085968732834,
+      "learning_rate": 0.00029841056850481265,
+      "loss": 0.0567,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.09004423022270203,
+      "learning_rate": 0.0002983955653546427,
+      "loss": 0.0291,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.19469811022281647,
+      "learning_rate": 0.00029838049210767015,
+      "loss": 0.0487,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2525189518928528,
+      "learning_rate": 0.00029836534877101514,
+      "loss": 0.0629,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12139023840427399,
+      "learning_rate": 0.0002983501353518307,
+      "loss": 0.0457,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.06411401927471161,
+      "learning_rate": 0.00029833485185730326,
+      "loss": 0.0186,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.024475542828440666,
+      "learning_rate": 0.00029831949829465214,
+      "loss": 0.004,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.15951114892959595,
+      "learning_rate": 0.0002983040746711298,
+      "loss": 0.0297,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.03694155812263489,
+      "learning_rate": 0.0002982885809940218,
+      "loss": 0.0073,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13100893795490265,
+      "learning_rate": 0.0002982730172706468,
+      "loss": 0.0272,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.08929093927145004,
+      "learning_rate": 0.00029825738350835665,
+      "loss": 0.0146,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1474764049053192,
+      "learning_rate": 0.0002982416797145361,
+      "loss": 0.0422,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13874994218349457,
+      "learning_rate": 0.00029822590589660306,
+      "loss": 0.0353,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.048271678388118744,
+      "learning_rate": 0.00029821006206200856,
+      "loss": 0.0072,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.29017898440361023,
+      "learning_rate": 0.0002981941482182366,
+      "loss": 0.0607,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3267674446105957,
+      "learning_rate": 0.0002981781643728044,
+      "loss": 0.101,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.17602747678756714,
+      "learning_rate": 0.00029816211053326216,
+      "loss": 0.0236,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.08361077308654785,
+      "learning_rate": 0.00029814598670719304,
+      "loss": 0.0277,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.08593238145112991,
+      "learning_rate": 0.00029812979290221346,
+      "loss": 0.0291,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.08858275413513184,
+      "learning_rate": 0.00029811352912597277,
+      "loss": 0.0329,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.08017202466726303,
+      "learning_rate": 0.0002980971953861534,
+      "loss": 0.0287,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.06615002453327179,
+      "learning_rate": 0.0002980807916904709,
+      "loss": 0.0269,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12813499569892883,
+      "learning_rate": 0.00029806431804667364,
+      "loss": 0.0321,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05528206750750542,
+      "learning_rate": 0.0002980477744625433,
+      "loss": 0.0089,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.10161186009645462,
+      "learning_rate": 0.00029803116094589445,
+      "loss": 0.0294,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.09885023534297943,
+      "learning_rate": 0.00029801447750457476,
+      "loss": 0.0232,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.20870375633239746,
+      "learning_rate": 0.00029799772414646484,
+      "loss": 0.0478,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2730790674686432,
+      "learning_rate": 0.00029798090087947843,
+      "loss": 0.042,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.20371069014072418,
+      "learning_rate": 0.0002979640077115622,
+      "loss": 0.0634,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14660406112670898,
+      "learning_rate": 0.0002979470446506959,
+      "loss": 0.0201,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.19971100986003876,
+      "learning_rate": 0.0002979300117048923,
+      "loss": 0.0431,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.14965400099754333,
+      "learning_rate": 0.0002979129088821971,
+      "loss": 0.041,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2110958695411682,
+      "learning_rate": 0.0002978957361906892,
+      "loss": 0.028,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13050246238708496,
+      "learning_rate": 0.0002978784936384802,
+      "loss": 0.0258,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.0885690301656723,
+      "learning_rate": 0.000297861181233715,
+      "loss": 0.0337,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.26541608572006226,
+      "learning_rate": 0.0002978437989845713,
+      "loss": 0.1142,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.14441104233264923,
+      "learning_rate": 0.0002978263468992599,
+      "loss": 0.0368,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.11450188606977463,
+      "learning_rate": 0.0002978088249860245,
+      "loss": 0.0243,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3472074568271637,
+      "learning_rate": 0.00029779123325314184,
+      "loss": 0.0786,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.07867071032524109,
+      "learning_rate": 0.0002977735717089217,
+      "loss": 0.0356,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1661967933177948,
+      "learning_rate": 0.0002977558403617067,
+      "loss": 0.047,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.17638400197029114,
+      "learning_rate": 0.00029773803921987244,
+      "loss": 0.0527,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.05885611101984978,
+      "learning_rate": 0.0002977201682918277,
+      "loss": 0.0156,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.07076411694288254,
+      "learning_rate": 0.00029770222758601395,
+      "loss": 0.0418,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.06245988979935646,
+      "learning_rate": 0.0002976842171109058,
+      "loss": 0.0199,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.08311894536018372,
+      "learning_rate": 0.0002976661368750107,
+      "loss": 0.028,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11093831807374954,
+      "learning_rate": 0.0002976479868868692,
+      "loss": 0.0298,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.17683441936969757,
+      "learning_rate": 0.00029762976715505464,
+      "loss": 0.0539,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13351142406463623,
+      "learning_rate": 0.00029761147768817345,
+      "loss": 0.0593,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.07717160880565643,
+      "learning_rate": 0.0002975931184948648,
+      "loss": 0.0227,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11211559176445007,
+      "learning_rate": 0.0002975746895838011,
+      "loss": 0.0385,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.09209641814231873,
+      "learning_rate": 0.00029755619096368734,
+      "loss": 0.0086,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.0850004106760025,
+      "learning_rate": 0.0002975376226432617,
+      "loss": 0.0343,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.17711663246154785,
+      "learning_rate": 0.0002975189846312952,
+      "loss": 0.0665,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13066548109054565,
+      "learning_rate": 0.0002975002769365918,
+      "loss": 0.0551,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.07509409636259079,
+      "learning_rate": 0.00029748149956798826,
+      "loss": 0.0087,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3725223243236542,
+      "learning_rate": 0.0002974626525343544,
+      "loss": 0.026,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20973052084445953,
+      "learning_rate": 0.0002974437358445929,
+      "loss": 0.015,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.25902581214904785,
+      "learning_rate": 0.0002974247495076393,
+      "loss": 0.0617,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.22490067780017853,
+      "learning_rate": 0.000297405693532462,
+      "loss": 0.0456,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2885708510875702,
+      "learning_rate": 0.0002973865679280626,
+      "loss": 0.1066,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2658590078353882,
+      "learning_rate": 0.00029736737270347517,
+      "loss": 0.0931,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11531944572925568,
+      "learning_rate": 0.00029734810786776687,
+      "loss": 0.0238,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.0557803250849247,
+      "learning_rate": 0.00029732877343003776,
+      "loss": 0.0257,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.10880523920059204,
+      "learning_rate": 0.00029730936939942077,
+      "loss": 0.0387,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.09500639885663986,
+      "learning_rate": 0.0002972898957850816,
+      "loss": 0.0308,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11504241824150085,
+      "learning_rate": 0.0002972703525962189,
+      "loss": 0.0292,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.10513140261173248,
+      "learning_rate": 0.0002972507398420643,
+      "loss": 0.0245,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.20218555629253387,
+      "learning_rate": 0.000297231057531882,
+      "loss": 0.0394,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.053536418825387955,
+      "learning_rate": 0.00029721130567496936,
+      "loss": 0.0129,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.15879443287849426,
+      "learning_rate": 0.0002971914842806564,
+      "loss": 0.054,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11933678388595581,
+      "learning_rate": 0.00029717159335830606,
+      "loss": 0.0206,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.14436180889606476,
+      "learning_rate": 0.0002971516329173141,
+      "loss": 0.024,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.01978749968111515,
+      "learning_rate": 0.0002971316029671091,
+      "loss": 0.0047,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1731237769126892,
+      "learning_rate": 0.00029711150351715253,
+      "loss": 0.0605,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.059307076036930084,
+      "learning_rate": 0.00029709133457693867,
+      "loss": 0.0308,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3645476996898651,
+      "learning_rate": 0.00029707109615599456,
+      "loss": 0.0566,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.10670791566371918,
+      "learning_rate": 0.0002970507882638801,
+      "loss": 0.0234,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.10919758677482605,
+      "learning_rate": 0.0002970304109101881,
+      "loss": 0.0157,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.08173630386590958,
+      "learning_rate": 0.00029700996410454407,
+      "loss": 0.0371,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.13943839073181152,
+      "learning_rate": 0.00029698944785660635,
+      "loss": 0.0781,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.342821329832077,
+      "learning_rate": 0.00029696886217606605,
+      "loss": 0.0476,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.048615969717502594,
+      "learning_rate": 0.0002969482070726472,
+      "loss": 0.0083,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1213599145412445,
+      "learning_rate": 0.0002969274825561064,
+      "loss": 0.0258,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1914874166250229,
+      "learning_rate": 0.0002969066886362333,
+      "loss": 0.034,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.14067624509334564,
+      "learning_rate": 0.0002968858253228502,
+      "loss": 0.0395,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.08359983563423157,
+      "learning_rate": 0.00029686489262581217,
+      "loss": 0.0315,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11551601439714432,
+      "learning_rate": 0.000296843890555007,
+      "loss": 0.058,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12968787550926208,
+      "learning_rate": 0.00029682281912035545,
+      "loss": 0.0347,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.10182147473096848,
+      "learning_rate": 0.0002968016783318109,
+      "loss": 0.0165,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.06534916907548904,
+      "learning_rate": 0.00029678046819935934,
+      "loss": 0.0218,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12587250769138336,
+      "learning_rate": 0.0002967591887330199,
+      "loss": 0.0498,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.06701786816120148,
+      "learning_rate": 0.0002967378399428441,
+      "loss": 0.0484,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.10836692154407501,
+      "learning_rate": 0.00029671642183891643,
+      "loss": 0.0412,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.061415113508701324,
+      "learning_rate": 0.00029669493443135403,
+      "loss": 0.0172,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.20760087668895721,
+      "learning_rate": 0.0002966733777303068,
+      "loss": 0.0494,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.11503862589597702,
+      "learning_rate": 0.00029665175174595736,
+      "loss": 0.0385,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.07366505265235901,
+      "learning_rate": 0.000296630056488521,
+      "loss": 0.0403,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.036951594054698944,
+      "learning_rate": 0.00029660829196824577,
+      "loss": 0.0092,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.08457314223051071,
+      "learning_rate": 0.0002965864581954126,
+      "loss": 0.0445,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24513787031173706,
+      "learning_rate": 0.0002965645551803349,
+      "loss": 0.0716,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.08235831558704376,
+      "learning_rate": 0.00029654258293335887,
+      "loss": 0.029,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.08004003018140793,
+      "learning_rate": 0.00029652054146486344,
+      "loss": 0.0365,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.14928393065929413,
+      "learning_rate": 0.0002964984307852602,
+      "loss": 0.039,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1802273988723755,
+      "learning_rate": 0.00029647625090499345,
+      "loss": 0.0324,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.18169750273227692,
+      "learning_rate": 0.00029645400183454026,
+      "loss": 0.0427,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.13121691346168518,
+      "learning_rate": 0.0002964316835844102,
+      "loss": 0.0274,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.27358877658843994,
+      "learning_rate": 0.0002964092961651456,
+      "loss": 0.0537,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.16992299258708954,
+      "learning_rate": 0.0002963868395873216,
+      "loss": 0.0797,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2110740691423416,
+      "learning_rate": 0.0002963643138615458,
+      "loss": 0.0835,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.17114487290382385,
+      "learning_rate": 0.0002963417189984586,
+      "loss": 0.0619,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.09492560476064682,
+      "learning_rate": 0.000296319055008733,
+      "loss": 0.0212,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.19000209867954254,
+      "learning_rate": 0.0002962963219030746,
+      "loss": 0.0802,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11632812023162842,
+      "learning_rate": 0.0002962735196922219,
+      "loss": 0.0426,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.15153561532497406,
+      "learning_rate": 0.0002962506483869456,
+      "loss": 0.07,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.0691797137260437,
+      "learning_rate": 0.00029622770799804944,
+      "loss": 0.0246,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.0731196403503418,
+      "learning_rate": 0.0002962046985363697,
+      "loss": 0.0413,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1449161171913147,
+      "learning_rate": 0.00029618162001277513,
+      "loss": 0.023,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.13844870030879974,
+      "learning_rate": 0.0002961584724381672,
+      "loss": 0.055,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.08192728459835052,
+      "learning_rate": 0.00029613525582348007,
+      "loss": 0.0274,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.030294157564640045,
+      "learning_rate": 0.0002961119701796804,
+      "loss": 0.0332,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12008962035179138,
+      "learning_rate": 0.0002960886155177675,
+      "loss": 0.0293,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.22829335927963257,
+      "learning_rate": 0.0002960651918487734,
+      "loss": 0.049,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.09662315249443054,
+      "learning_rate": 0.00029604169918376246,
+      "loss": 0.019,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.056000061333179474,
+      "learning_rate": 0.0002960181375338318,
+      "loss": 0.0077,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.04742419347167015,
+      "learning_rate": 0.00029599450691011116,
+      "loss": 0.0216,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.17151907086372375,
+      "learning_rate": 0.0002959708073237628,
+      "loss": 0.0364,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3108668923377991,
+      "learning_rate": 0.00029594703878598155,
+      "loss": 0.0288,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05538111925125122,
+      "learning_rate": 0.00029592320130799487,
+      "loss": 0.0048,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2907853126525879,
+      "learning_rate": 0.00029589929490106263,
+      "loss": 0.0443,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.19189013540744781,
+      "learning_rate": 0.0002958753195764775,
+      "loss": 0.0688,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3744778037071228,
+      "learning_rate": 0.00029585127534556446,
+      "loss": 0.0726,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.02139083668589592,
+      "learning_rate": 0.00029582716221968124,
+      "loss": 0.003,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3209889531135559,
+      "learning_rate": 0.00029580298021021796,
+      "loss": 0.068,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.13530127704143524,
+      "learning_rate": 0.0002957787293285974,
+      "loss": 0.0229,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.04955355450510979,
+      "learning_rate": 0.00029575440958627485,
+      "loss": 0.007,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05992133542895317,
+      "learning_rate": 0.0002957300209947379,
+      "loss": 0.014,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.08975626528263092,
+      "learning_rate": 0.0002957055635655071,
+      "loss": 0.0419,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3397723436355591,
+      "learning_rate": 0.00029568103731013513,
+      "loss": 0.093,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.05291612446308136,
+      "learning_rate": 0.00029565644224020733,
+      "loss": 0.0137,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.16154609620571136,
+      "learning_rate": 0.0002956317783673416,
+      "loss": 0.0414,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12861596047878265,
+      "learning_rate": 0.0002956070457031882,
+      "loss": 0.0372,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.09462448954582214,
+      "learning_rate": 0.00029558224425943003,
+      "loss": 0.0292,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14290063083171844,
+      "learning_rate": 0.00029555737404778233,
+      "loss": 0.0572,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11055822670459747,
+      "learning_rate": 0.00029553243507999307,
+      "loss": 0.0372,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.10231087356805801,
+      "learning_rate": 0.00029550742736784237,
+      "loss": 0.0368,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.09969429671764374,
+      "learning_rate": 0.00029548235092314304,
+      "loss": 0.0416,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1207612007856369,
+      "learning_rate": 0.00029545720575774033,
+      "loss": 0.0307,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11535090953111649,
+      "learning_rate": 0.0002954319918835119,
+      "loss": 0.0296,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1460224986076355,
+      "learning_rate": 0.00029540670931236786,
+      "loss": 0.0587,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.10432720184326172,
+      "learning_rate": 0.0002953813580562509,
+      "loss": 0.0397,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.2140846997499466,
+      "learning_rate": 0.0002953559381271359,
+      "loss": 0.0538,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12050808221101761,
+      "learning_rate": 0.00029533044953703044,
+      "loss": 0.0439,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.07928888499736786,
+      "learning_rate": 0.0002953048922979744,
+      "loss": 0.0163,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.08733994513750076,
+      "learning_rate": 0.0002952792664220402,
+      "loss": 0.0219,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.18080447614192963,
+      "learning_rate": 0.0002952535719213325,
+      "loss": 0.0469,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.08348793536424637,
+      "learning_rate": 0.0002952278088079884,
+      "loss": 0.035,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1347195953130722,
+      "learning_rate": 0.00029520197709417763,
+      "loss": 0.029,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11075679957866669,
+      "learning_rate": 0.0002951760767921021,
+      "loss": 0.0257,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13172994554042816,
+      "learning_rate": 0.0002951501079139962,
+      "loss": 0.0302,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.114262655377388,
+      "learning_rate": 0.0002951240704721267,
+      "loss": 0.0492,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.034534960985183716,
+      "eval_runtime": 39.6959,
+      "eval_samples_per_second": 32.421,
+      "eval_steps_per_second": 8.112,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.08364730328321457,
+      "learning_rate": 0.0002950979644787928,
+      "loss": 0.0185,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.16603770852088928,
+      "learning_rate": 0.000295071789946326,
+      "loss": 0.0443,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1269228458404541,
+      "learning_rate": 0.00029504554688709027,
+      "loss": 0.0217,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.15612861514091492,
+      "learning_rate": 0.0002950192353134819,
+      "loss": 0.0377,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.056646961718797684,
+      "learning_rate": 0.00029499285523792946,
+      "loss": 0.0133,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23394975066184998,
+      "learning_rate": 0.000294966406672894,
+      "loss": 0.0767,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.21382953226566315,
+      "learning_rate": 0.00029493988963086895,
+      "loss": 0.0729,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.27641353011131287,
+      "learning_rate": 0.00029491330412438,
+      "loss": 0.1022,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.0760459303855896,
+      "learning_rate": 0.0002948866501659852,
+      "loss": 0.0269,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5418729186058044,
+      "learning_rate": 0.0002948599277682748,
+      "loss": 0.1523,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13234178721904755,
+      "learning_rate": 0.00029483313694387165,
+      "loss": 0.0292,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.07174021750688553,
+      "learning_rate": 0.00029480627770543086,
+      "loss": 0.0395,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.09958759695291519,
+      "learning_rate": 0.00029477935006563957,
+      "loss": 0.0559,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.07592346519231796,
+      "learning_rate": 0.00029475235403721763,
+      "loss": 0.0488,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.10129998624324799,
+      "learning_rate": 0.00029472528963291685,
+      "loss": 0.0287,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.08051212131977081,
+      "learning_rate": 0.00029469815686552163,
+      "loss": 0.0386,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.0695783942937851,
+      "learning_rate": 0.0002946709557478485,
+      "loss": 0.0201,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1511554718017578,
+      "learning_rate": 0.00029464368629274624,
+      "loss": 0.0464,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.075484499335289,
+      "learning_rate": 0.00029461634851309597,
+      "loss": 0.031,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.08108027279376984,
+      "learning_rate": 0.00029458894242181114,
+      "loss": 0.0271,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.07254958897829056,
+      "learning_rate": 0.00029456146803183745,
+      "loss": 0.0187,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.215089812874794,
+      "learning_rate": 0.00029453392535615274,
+      "loss": 0.0463,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.034637995064258575,
+      "learning_rate": 0.0002945063144077672,
+      "loss": 0.0084,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12073606252670288,
+      "learning_rate": 0.00029447863519972337,
+      "loss": 0.0401,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13762198388576508,
+      "learning_rate": 0.00029445088774509583,
+      "loss": 0.0244,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2537041902542114,
+      "learning_rate": 0.00029442307205699154,
+      "loss": 0.0574,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1401953399181366,
+      "learning_rate": 0.00029439518814854956,
+      "loss": 0.0202,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13872119784355164,
+      "learning_rate": 0.0002943672360329413,
+      "loss": 0.0373,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3436320126056671,
+      "learning_rate": 0.00029433921572337044,
+      "loss": 0.0944,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.20004349946975708,
+      "learning_rate": 0.00029431112723307266,
+      "loss": 0.0625,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.10176026076078415,
+      "learning_rate": 0.00029428297057531607,
+      "loss": 0.023,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.08603208512067795,
+      "learning_rate": 0.0002942547457634008,
+      "loss": 0.0141,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.03601311519742012,
+      "learning_rate": 0.0002942264528106592,
+      "loss": 0.0071,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1434870958328247,
+      "learning_rate": 0.000294198091730456,
+      "loss": 0.0362,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1505521684885025,
+      "learning_rate": 0.0002941696625361879,
+      "loss": 0.0211,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.14390698075294495,
+      "learning_rate": 0.0002941411652412838,
+      "loss": 0.054,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.21683859825134277,
+      "learning_rate": 0.00029411259985920486,
+      "loss": 0.0482,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12036791443824768,
+      "learning_rate": 0.0002940839664034444,
+      "loss": 0.0444,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.09479566663503647,
+      "learning_rate": 0.00029405526488752775,
+      "loss": 0.035,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.14229558408260345,
+      "learning_rate": 0.0002940264953250125,
+      "loss": 0.0573,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.22773970663547516,
+      "learning_rate": 0.00029399765772948844,
+      "loss": 0.061,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.11387961357831955,
+      "learning_rate": 0.0002939687521145774,
+      "loss": 0.057,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1798745095729828,
+      "learning_rate": 0.00029393977849393333,
+      "loss": 0.0392,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.07203508168458939,
+      "learning_rate": 0.0002939107368812424,
+      "loss": 0.0152,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.04569177329540253,
+      "learning_rate": 0.0002938816272902228,
+      "loss": 0.0113,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.0927419438958168,
+      "learning_rate": 0.0002938524497346249,
+      "loss": 0.0246,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.16807597875595093,
+      "learning_rate": 0.0002938232042282311,
+      "loss": 0.0364,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12006795406341553,
+      "learning_rate": 0.00029379389078485596,
+      "loss": 0.0118,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.0377679318189621,
+      "learning_rate": 0.0002937645094183461,
+      "loss": 0.0063,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.27051666378974915,
+      "learning_rate": 0.00029373506014258025,
+      "loss": 0.0682,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.228448748588562,
+      "learning_rate": 0.0002937055429714692,
+      "loss": 0.0733,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.18427824974060059,
+      "learning_rate": 0.00029367595791895577,
+      "loss": 0.0338,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.25813257694244385,
+      "learning_rate": 0.00029364630499901503,
+      "loss": 0.0323,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.17406705021858215,
+      "learning_rate": 0.0002936165842256538,
+      "loss": 0.0398,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5199068188667297,
+      "learning_rate": 0.0002935867956129112,
+      "loss": 0.0486,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3251938223838806,
+      "learning_rate": 0.0002935569391748583,
+      "loss": 0.049,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.057003892958164215,
+      "learning_rate": 0.00029352701492559827,
+      "loss": 0.0114,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.15188859403133392,
+      "learning_rate": 0.00029349702287926623,
+      "loss": 0.0323,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.17942048609256744,
+      "learning_rate": 0.0002934669630500293,
+      "loss": 0.0437,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.06396406143903732,
+      "learning_rate": 0.0002934368354520867,
+      "loss": 0.0097,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1496248096227646,
+      "learning_rate": 0.00029340664009966974,
+      "loss": 0.0316,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.0654374286532402,
+      "learning_rate": 0.00029337637700704156,
+      "loss": 0.0083,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.04386695846915245,
+      "learning_rate": 0.0002933460461884973,
+      "loss": 0.0094,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.14928901195526123,
+      "learning_rate": 0.0002933156476583643,
+      "loss": 0.0484,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12666364014148712,
+      "learning_rate": 0.0002932851814310017,
+      "loss": 0.0148,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.023791933432221413,
+      "learning_rate": 0.0002932546475208006,
+      "loss": 0.003,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.022256718948483467,
+      "learning_rate": 0.0002932240459421842,
+      "loss": 0.0044,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12194914370775223,
+      "learning_rate": 0.0002931933767096076,
+      "loss": 0.009,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.29687178134918213,
+      "learning_rate": 0.0002931626398375578,
+      "loss": 0.0691,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24758018553256989,
+      "learning_rate": 0.00029313183534055386,
+      "loss": 0.0589,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.10298270732164383,
+      "learning_rate": 0.0002931009632331468,
+      "loss": 0.0187,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1447860449552536,
+      "learning_rate": 0.00029307002352991937,
+      "loss": 0.0297,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2590334117412567,
+      "learning_rate": 0.00029303901624548644,
+      "loss": 0.0892,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.07339983433485031,
+      "learning_rate": 0.00029300794139449477,
+      "loss": 0.0249,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.16213186085224152,
+      "learning_rate": 0.000292976798991623,
+      "loss": 0.0493,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.03418932110071182,
+      "learning_rate": 0.0002929455890515818,
+      "loss": 0.0066,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.18771564960479736,
+      "learning_rate": 0.0002929143115891134,
+      "loss": 0.03,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13976161181926727,
+      "learning_rate": 0.00029288296661899243,
+      "loss": 0.0451,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.07075387239456177,
+      "learning_rate": 0.00029285155415602495,
+      "loss": 0.0201,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1304980367422104,
+      "learning_rate": 0.0002928200742150492,
+      "loss": 0.0286,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.06026493385434151,
+      "learning_rate": 0.00029278852681093514,
+      "loss": 0.0159,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.08018484711647034,
+      "learning_rate": 0.0002927569119585847,
+      "loss": 0.0333,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.21171532571315765,
+      "learning_rate": 0.0002927252296729315,
+      "loss": 0.034,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14055241644382477,
+      "learning_rate": 0.0002926934799689413,
+      "loss": 0.0504,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.17434647679328918,
+      "learning_rate": 0.0002926616628616113,
+      "loss": 0.0519,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12710362672805786,
+      "learning_rate": 0.00029262977836597105,
+      "loss": 0.0154,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.16046389937400818,
+      "learning_rate": 0.0002925978264970814,
+      "loss": 0.0398,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.23207533359527588,
+      "learning_rate": 0.00029256580727003543,
+      "loss": 0.0562,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.29609429836273193,
+      "learning_rate": 0.0002925337206999579,
+      "loss": 0.137,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.15176476538181305,
+      "learning_rate": 0.00029250156680200526,
+      "loss": 0.025,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.14394959807395935,
+      "learning_rate": 0.00029246934559136597,
+      "loss": 0.0519,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.08391053229570389,
+      "learning_rate": 0.00029243705708326015,
+      "loss": 0.0184,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.09384860098361969,
+      "learning_rate": 0.00029240470129293975,
+      "loss": 0.0229,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12083159387111664,
+      "learning_rate": 0.00029237227823568845,
+      "loss": 0.0219,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.19567762315273285,
+      "learning_rate": 0.0002923397879268218,
+      "loss": 0.0728,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.07342015206813812,
+      "learning_rate": 0.0002923072303816871,
+      "loss": 0.0412,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.06717100739479065,
+      "learning_rate": 0.00029227460561566333,
+      "loss": 0.0309,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.09244221448898315,
+      "learning_rate": 0.0002922419136441613,
+      "loss": 0.0508,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.052494604140520096,
+      "learning_rate": 0.0002922091544826235,
+      "loss": 0.0319,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.14286155998706818,
+      "learning_rate": 0.00029217632814652417,
+      "loss": 0.0654,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.06442811340093613,
+      "learning_rate": 0.00029214343465136945,
+      "loss": 0.0132,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.05420248210430145,
+      "learning_rate": 0.0002921104740126969,
+      "loss": 0.0115,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.04951406642794609,
+      "learning_rate": 0.0002920774462460761,
+      "loss": 0.0086,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.08321358263492584,
+      "learning_rate": 0.00029204435136710803,
+      "loss": 0.0445,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.11665898561477661,
+      "learning_rate": 0.0002920111893914257,
+      "loss": 0.0262,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1829105019569397,
+      "learning_rate": 0.00029197796033469356,
+      "loss": 0.0308,
+      "step": 488
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20940159261226654,
+      "learning_rate": 0.00029194466421260786,
+      "loss": 0.0299,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20697347819805145,
+      "learning_rate": 0.0002919113010408965,
+      "loss": 0.0405,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.051994968205690384,
+      "learning_rate": 0.000291877870835319,
+      "loss": 0.01,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1463523805141449,
+      "learning_rate": 0.00029184437361166676,
+      "loss": 0.0555,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.09110219031572342,
+      "learning_rate": 0.00029181080938576255,
+      "loss": 0.0371,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.04076121374964714,
+      "learning_rate": 0.00029177717817346097,
+      "loss": 0.0065,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.11555450409650803,
+      "learning_rate": 0.0002917434799906482,
+      "loss": 0.0115,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.15579824149608612,
+      "learning_rate": 0.0002917097148532421,
+      "loss": 0.0332,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41938668489456177,
+      "learning_rate": 0.000291675882777192,
+      "loss": 0.0678,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.16764874756336212,
+      "learning_rate": 0.0002916419837784791,
+      "loss": 0.0683,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1291145384311676,
+      "learning_rate": 0.00029160801787311613,
+      "loss": 0.0376,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.06120933219790459,
+      "learning_rate": 0.0002915739850771472,
+      "loss": 0.0307,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.09218423068523407,
+      "learning_rate": 0.0002915398854066483,
+      "loss": 0.0545,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12664952874183655,
+      "learning_rate": 0.00029150571887772694,
+      "loss": 0.0274,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.0705379918217659,
+      "learning_rate": 0.0002914714855065221,
+      "loss": 0.0198,
+      "step": 503
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.03559693694114685,
+      "learning_rate": 0.00029143718530920447,
+      "loss": 0.0114,
+      "step": 504
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.051283448934555054,
+      "learning_rate": 0.0002914028183019762,
+      "loss": 0.0327,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12527117133140564,
+      "learning_rate": 0.0002913683845010711,
+      "loss": 0.0316,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.0627032071352005,
+      "learning_rate": 0.0002913338839227544,
+      "loss": 0.0185,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.07235468178987503,
+      "learning_rate": 0.000291299316583323,
+      "loss": 0.0605,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.07697612792253494,
+      "learning_rate": 0.0002912646824991053,
+      "loss": 0.031,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.08240342885255814,
+      "learning_rate": 0.0002912299816864612,
+      "loss": 0.0211,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.07725581526756287,
+      "learning_rate": 0.0002911952141617821,
+      "loss": 0.0311,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.14777988195419312,
+      "learning_rate": 0.000291160379941491,
+      "loss": 0.038,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.11423151195049286,
+      "learning_rate": 0.0002911254790420423,
+      "loss": 0.0594,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.07308260351419449,
+      "learning_rate": 0.000291090511479922,
+      "loss": 0.0416,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.11171098798513412,
+      "learning_rate": 0.00029105547727164747,
+      "loss": 0.0509,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.29647496342658997,
+      "learning_rate": 0.00029102037643376764,
+      "loss": 0.0421,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.08812320232391357,
+      "learning_rate": 0.00029098520898286303,
+      "loss": 0.0559,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13493718206882477,
+      "learning_rate": 0.00029094997493554525,
+      "loss": 0.0257,
+      "step": 518
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1292780339717865,
+      "learning_rate": 0.0002909146743084579,
+      "loss": 0.0699,
+      "step": 519
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.03736162185668945,
+      "learning_rate": 0.0002908793071182755,
+      "loss": 0.0113,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.20628990232944489,
+      "learning_rate": 0.00029084387338170435,
+      "loss": 0.1039,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13702163100242615,
+      "learning_rate": 0.0002908083731154821,
+      "loss": 0.0715,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.10376426577568054,
+      "learning_rate": 0.0002907728063363779,
+      "loss": 0.0566,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.03796597197651863,
+      "learning_rate": 0.00029073717306119206,
+      "loss": 0.0131,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12588168680667877,
+      "learning_rate": 0.0002907014733067566,
+      "loss": 0.0754,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.18614119291305542,
+      "learning_rate": 0.00029066570708993474,
+      "loss": 0.0839,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.08624828606843948,
+      "learning_rate": 0.0002906298744276212,
+      "loss": 0.0519,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.09907104074954987,
+      "learning_rate": 0.00029059397533674216,
+      "loss": 0.0554,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.05135316029191017,
+      "learning_rate": 0.00029055800983425494,
+      "loss": 0.0374,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.10954371839761734,
+      "learning_rate": 0.00029052197793714844,
+      "loss": 0.03,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13733310997486115,
+      "learning_rate": 0.0002904858796624428,
+      "loss": 0.0345,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.09171781688928604,
+      "learning_rate": 0.00029044971502718966,
+      "loss": 0.0285,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.08643066138029099,
+      "learning_rate": 0.00029041348404847177,
+      "loss": 0.0225,
+      "step": 533
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3179713487625122,
+      "learning_rate": 0.00029037718674340343,
+      "loss": 0.1167,
+      "step": 534
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.09737833589315414,
+      "learning_rate": 0.0002903408231291303,
+      "loss": 0.047,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.15587852895259857,
+      "learning_rate": 0.00029030439322282904,
+      "loss": 0.0406,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.07560009509325027,
+      "learning_rate": 0.0002902678970417081,
+      "loss": 0.0387,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12732967734336853,
+      "learning_rate": 0.00029023133460300677,
+      "loss": 0.0434,
+      "step": 538
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.06021510064601898,
+      "learning_rate": 0.00029019470592399593,
+      "loss": 0.0149,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.09609080851078033,
+      "learning_rate": 0.0002901580110219777,
+      "loss": 0.0203,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1442640721797943,
+      "learning_rate": 0.0002901212499142854,
+      "loss": 0.0345,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.15236537158489227,
+      "learning_rate": 0.0002900844226182837,
+      "loss": 0.041,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.14138057827949524,
+      "learning_rate": 0.00029004752915136854,
+      "loss": 0.0413,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.16659876704216003,
+      "learning_rate": 0.000290010569530967,
+      "loss": 0.0202,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.16970619559288025,
+      "learning_rate": 0.0002899735437745376,
+      "loss": 0.0373,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.044596217572689056,
+      "learning_rate": 0.00028993645189956987,
+      "loss": 0.0202,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.07182051986455917,
+      "learning_rate": 0.00028989929392358484,
+      "loss": 0.0137,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2593410313129425,
+      "learning_rate": 0.0002898620698641345,
+      "loss": 0.0373,
+      "step": 548
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.17339394986629486,
+      "learning_rate": 0.0002898247797388023,
+      "loss": 0.0217,
+      "step": 549
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13247337937355042,
+      "learning_rate": 0.00028978742356520256,
+      "loss": 0.0621,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.04582560807466507,
+      "learning_rate": 0.00028975000136098123,
+      "loss": 0.0051,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.04409830644726753,
+      "learning_rate": 0.0002897125131438151,
+      "loss": 0.0042,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.11188169568777084,
+      "learning_rate": 0.0002896749589314123,
+      "loss": 0.0307,
+      "step": 553
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.10103113949298859,
+      "learning_rate": 0.00028963733874151225,
+      "loss": 0.0132,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13099652528762817,
+      "learning_rate": 0.0002895996525918852,
+      "loss": 0.0348,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.07826762646436691,
+      "learning_rate": 0.0002895619005003328,
+      "loss": 0.0232,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.053435299545526505,
+      "learning_rate": 0.00028952408248468785,
+      "loss": 0.0113,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.07408218830823898,
+      "learning_rate": 0.00028948619856281423,
+      "loss": 0.0099,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.08491642028093338,
+      "learning_rate": 0.00028944824875260693,
+      "loss": 0.0122,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.0294903963804245,
+      "learning_rate": 0.00028941023307199214,
+      "loss": 0.0044,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.16142538189888,
+      "learning_rate": 0.000289372151538927,
+      "loss": 0.0721,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.11368390917778015,
+      "learning_rate": 0.0002893340041714,
+      "loss": 0.0109,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1799473911523819,
+      "learning_rate": 0.0002892957909874306,
+      "loss": 0.0487,
+      "step": 563
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1448475420475006,
+      "learning_rate": 0.0002892575120050693,
+      "loss": 0.0601,
+      "step": 564
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.07079991698265076,
+      "learning_rate": 0.00028921916724239773,
+      "loss": 0.0089,
+      "step": 565
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13462460041046143,
+      "learning_rate": 0.0002891807567175287,
+      "loss": 0.0361,
+      "step": 566
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.08166678249835968,
+      "learning_rate": 0.00028914228044860584,
+      "loss": 0.0412,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.09470119327306747,
+      "learning_rate": 0.00028910373845380405,
+      "loss": 0.036,
+      "step": 568
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.0957297682762146,
+      "learning_rate": 0.00028906513075132917,
+      "loss": 0.0302,
+      "step": 569
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.17004123330116272,
+      "learning_rate": 0.00028902645735941814,
+      "loss": 0.0559,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.10910087823867798,
+      "learning_rate": 0.0002889877182963389,
+      "loss": 0.0765,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1027827113866806,
+      "learning_rate": 0.0002889489135803904,
+      "loss": 0.0261,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1182394027709961,
+      "learning_rate": 0.00028891004322990254,
+      "loss": 0.0413,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.08422794938087463,
+      "learning_rate": 0.00028887110726323644,
+      "loss": 0.048,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.10699556767940521,
+      "learning_rate": 0.00028883210569878397,
+      "loss": 0.0193,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06325127184391022,
+      "learning_rate": 0.00028879303855496805,
+      "loss": 0.0248,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.10081582516431808,
+      "learning_rate": 0.00028875390585024274,
+      "loss": 0.0211,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.062216054648160934,
+      "learning_rate": 0.00028871470760309285,
+      "loss": 0.0185,
+      "step": 578
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.086198590695858,
+      "learning_rate": 0.00028867544383203423,
+      "loss": 0.0544,
+      "step": 579
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.11464603990316391,
+      "learning_rate": 0.00028863611455561374,
+      "loss": 0.0482,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1089998185634613,
+      "learning_rate": 0.0002885967197924092,
+      "loss": 0.0496,
+      "step": 581
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1297656148672104,
+      "learning_rate": 0.00028855725956102913,
+      "loss": 0.0286,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12966851890087128,
+      "learning_rate": 0.0002885177338801133,
+      "loss": 0.0271,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1413564682006836,
+      "learning_rate": 0.00028847814276833215,
+      "loss": 0.0334,
+      "step": 584
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.08366623520851135,
+      "learning_rate": 0.0002884384862443871,
+      "loss": 0.0252,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.11143944412469864,
+      "learning_rate": 0.0002883987643270106,
+      "loss": 0.0347,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.018316002562642097,
+      "learning_rate": 0.0002883589770349658,
+      "loss": 0.0041,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.02275553159415722,
+      "learning_rate": 0.0002883191243870467,
+      "loss": 0.0049,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.14462235569953918,
+      "learning_rate": 0.0002882792064020785,
+      "loss": 0.0745,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.10231613367795944,
+      "learning_rate": 0.0002882392230989169,
+      "loss": 0.0211,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.013464580290019512,
+      "learning_rate": 0.00028819917449644865,
+      "loss": 0.0027,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1707848161458969,
+      "learning_rate": 0.0002881590606135912,
+      "loss": 0.0292,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.021210182458162308,
+      "learning_rate": 0.00028811888146929303,
+      "loss": 0.0034,
+      "step": 593
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.09697694331407547,
+      "learning_rate": 0.00028807863708253326,
+      "loss": 0.0134,
+      "step": 594
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.014497664757072926,
+      "learning_rate": 0.000288038327472322,
+      "loss": 0.0033,
+      "step": 595
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25384795665740967,
+      "learning_rate": 0.00028799795265770003,
+      "loss": 0.0258,
+      "step": 596
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.0065186647698283195,
+      "learning_rate": 0.00028795751265773894,
+      "loss": 0.0012,
+      "step": 597
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.03637157753109932,
+      "learning_rate": 0.00028791700749154124,
+      "loss": 0.004,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.039990831166505814,
+      "learning_rate": 0.00028787643717824007,
+      "loss": 0.0067,
+      "step": 599
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.18821458518505096,
+      "learning_rate": 0.0002878358017369994,
+      "loss": 0.0233,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12891234457492828,
+      "learning_rate": 0.00028779510118701404,
+      "loss": 0.0121,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1731066256761551,
+      "learning_rate": 0.0002877543355475094,
+      "loss": 0.0535,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5192031264305115,
+      "learning_rate": 0.0002877135048377418,
+      "loss": 0.1073,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13350637257099152,
+      "learning_rate": 0.0002876726090769982,
+      "loss": 0.0157,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12136203050613403,
+      "learning_rate": 0.0002876316482845963,
+      "loss": 0.0132,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5036077499389648,
+      "learning_rate": 0.0002875906224798844,
+      "loss": 0.1366,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.22896146774291992,
+      "learning_rate": 0.0002875495316822419,
+      "loss": 0.08,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.15327180922031403,
+      "learning_rate": 0.0002875083759110785,
+      "loss": 0.0322,
+      "step": 608
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.0520663745701313,
+      "learning_rate": 0.0002874671551858346,
+      "loss": 0.0202,
+      "step": 609
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.08731318265199661,
+      "learning_rate": 0.00028742586952598155,
+      "loss": 0.0414,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.11570514738559723,
+      "learning_rate": 0.0002873845189510213,
+      "loss": 0.0625,
+      "step": 611
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1604083925485611,
+      "learning_rate": 0.0002873431034804862,
+      "loss": 0.0644,
+      "step": 612
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06147552654147148,
+      "learning_rate": 0.0002873016231339396,
+      "loss": 0.0168,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12419867515563965,
+      "learning_rate": 0.00028726007793097527,
+      "loss": 0.0438,
+      "step": 614
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06133590638637543,
+      "learning_rate": 0.0002872184678912177,
+      "loss": 0.024,
+      "step": 615
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.10245617479085922,
+      "learning_rate": 0.00028717679303432207,
+      "loss": 0.0468,
+      "step": 616
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.11957762390375137,
+      "learning_rate": 0.000287135053379974,
+      "loss": 0.0442,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12896914780139923,
+      "learning_rate": 0.0002870932489478899,
+      "loss": 0.019,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1816866546869278,
+      "learning_rate": 0.0002870513797578167,
+      "loss": 0.0465,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3061673045158386,
+      "learning_rate": 0.00028700944582953184,
+      "loss": 0.0356,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12940478324890137,
+      "learning_rate": 0.0002869674471828435,
+      "loss": 0.0447,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2569711208343506,
+      "learning_rate": 0.0002869253838375903,
+      "loss": 0.0383,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.17063623666763306,
+      "learning_rate": 0.0002868832558136415,
+      "loss": 0.0394,
+      "step": 623
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.16775226593017578,
+      "learning_rate": 0.00028684106313089686,
+      "loss": 0.0314,
+      "step": 624
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12676480412483215,
+      "learning_rate": 0.00028679880580928676,
+      "loss": 0.0397,
+      "step": 625
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.19791187345981598,
+      "learning_rate": 0.0002867564838687721,
+      "loss": 0.0668,
+      "step": 626
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.18982940912246704,
+      "learning_rate": 0.0002867140973293441,
+      "loss": 0.0472,
+      "step": 627
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06308908015489578,
+      "learning_rate": 0.00028667164621102475,
+      "loss": 0.0166,
+      "step": 628
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.09570673853158951,
+      "learning_rate": 0.0002866291305338665,
+      "loss": 0.0156,
+      "step": 629
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12950573861598969,
+      "learning_rate": 0.00028658655031795215,
+      "loss": 0.0381,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.30905017256736755,
+      "learning_rate": 0.00028654390558339516,
+      "loss": 0.0386,
+      "step": 631
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2680380940437317,
+      "learning_rate": 0.0002865011963503394,
+      "loss": 0.0307,
+      "step": 632
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.15153923630714417,
+      "learning_rate": 0.00028645842263895916,
+      "loss": 0.0448,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06900045275688171,
+      "learning_rate": 0.0002864155844694592,
+      "loss": 0.0134,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39054739475250244,
+      "learning_rate": 0.00028637268186207474,
+      "loss": 0.0562,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06766320765018463,
+      "learning_rate": 0.0002863297148370716,
+      "loss": 0.0135,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12230436503887177,
+      "learning_rate": 0.0002862866834147457,
+      "loss": 0.0189,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.10021094232797623,
+      "learning_rate": 0.00028624358761542365,
+      "loss": 0.021,
+      "step": 638
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1645062267780304,
+      "learning_rate": 0.0002862004274594623,
+      "loss": 0.0284,
+      "step": 639
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3108697831630707,
+      "learning_rate": 0.00028615720296724906,
+      "loss": 0.0792,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12834666669368744,
+      "learning_rate": 0.0002861139141592017,
+      "loss": 0.0162,
+      "step": 641
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.11455690860748291,
+      "learning_rate": 0.00028607056105576806,
+      "loss": 0.0374,
+      "step": 642
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.14810198545455933,
+      "learning_rate": 0.0002860271436774269,
+      "loss": 0.0132,
+      "step": 643
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1764562875032425,
+      "learning_rate": 0.00028598366204468694,
+      "loss": 0.0641,
+      "step": 644
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.10819990932941437,
+      "learning_rate": 0.0002859401161780873,
+      "loss": 0.036,
+      "step": 645
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.10301560163497925,
+      "learning_rate": 0.00028589650609819764,
+      "loss": 0.0272,
+      "step": 646
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13949047029018402,
+      "learning_rate": 0.00028585283182561773,
+      "loss": 0.0396,
+      "step": 647
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.20076854526996613,
+      "learning_rate": 0.0002858090933809777,
+      "loss": 0.0304,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12382891029119492,
+      "learning_rate": 0.0002857652907849381,
+      "loss": 0.0317,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.03410351276397705,
+      "learning_rate": 0.0002857214240581897,
+      "loss": 0.0075,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.10016089677810669,
+      "learning_rate": 0.00028567749322145367,
+      "loss": 0.0179,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.24712401628494263,
+      "learning_rate": 0.00028563349829548125,
+      "loss": 0.0857,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.10354748368263245,
+      "learning_rate": 0.00028558943930105413,
+      "loss": 0.0276,
+      "step": 653
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13952110707759857,
+      "learning_rate": 0.00028554531625898434,
+      "loss": 0.0352,
+      "step": 654
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.25892096757888794,
+      "learning_rate": 0.0002855011291901138,
+      "loss": 0.0635,
+      "step": 655
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1324494630098343,
+      "learning_rate": 0.0002854568781153151,
+      "loss": 0.0404,
+      "step": 656
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2835068702697754,
+      "learning_rate": 0.0002854125630554908,
+      "loss": 0.0913,
+      "step": 657
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06329616159200668,
+      "learning_rate": 0.00028536818403157387,
+      "loss": 0.0146,
+      "step": 658
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.07758588343858719,
+      "learning_rate": 0.0002853237410645272,
+      "loss": 0.022,
+      "step": 659
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.0839746966958046,
+      "learning_rate": 0.00028527923417534425,
+      "loss": 0.0175,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06847698986530304,
+      "learning_rate": 0.0002852346633850484,
+      "loss": 0.0257,
+      "step": 661
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.05117741599678993,
+      "learning_rate": 0.0002851900287146933,
+      "loss": 0.0136,
+      "step": 662
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12874063849449158,
+      "learning_rate": 0.0002851453301853628,
+      "loss": 0.0525,
+      "step": 663
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1822301298379898,
+      "learning_rate": 0.000285100567818171,
+      "loss": 0.0889,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.11295532435178757,
+      "learning_rate": 0.0002850557416342619,
+      "loss": 0.0242,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13836829364299774,
+      "learning_rate": 0.0002850108516548099,
+      "loss": 0.0441,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.10722105205059052,
+      "learning_rate": 0.0002849658979010194,
+      "loss": 0.0401,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1335124373435974,
+      "learning_rate": 0.000284920880394125,
+      "loss": 0.0355,
+      "step": 668
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.0793779119849205,
+      "learning_rate": 0.00028487579915539136,
+      "loss": 0.0653,
+      "step": 669
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06469617038965225,
+      "learning_rate": 0.00028483065420611313,
+      "loss": 0.0212,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.11224417388439178,
+      "learning_rate": 0.0002847854455676154,
+      "loss": 0.0689,
+      "step": 671
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.08650530874729156,
+      "learning_rate": 0.00028474017326125296,
+      "loss": 0.0301,
+      "step": 672
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.0688636302947998,
+      "learning_rate": 0.0002846948373084109,
+      "loss": 0.0161,
+      "step": 673
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13195598125457764,
+      "learning_rate": 0.0002846494377305043,
+      "loss": 0.0529,
+      "step": 674
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1887226700782776,
+      "learning_rate": 0.0002846039745489783,
+      "loss": 0.0615,
+      "step": 675
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06736018508672714,
+      "learning_rate": 0.0002845584477853082,
+      "loss": 0.0246,
+      "step": 676
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1488025039434433,
+      "learning_rate": 0.0002845128574609992,
+      "loss": 0.0361,
+      "step": 677
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.09811149537563324,
+      "learning_rate": 0.0002844672035975864,
+      "loss": 0.0228,
+      "step": 678
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06320784986019135,
+      "learning_rate": 0.0002844214862166352,
+      "loss": 0.0182,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.0695585086941719,
+      "learning_rate": 0.00028437570533974084,
+      "loss": 0.0393,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.08886481821537018,
+      "learning_rate": 0.00028432986098852857,
+      "loss": 0.0293,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.09019115567207336,
+      "learning_rate": 0.0002842839531846537,
+      "loss": 0.0436,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1718403697013855,
+      "learning_rate": 0.0002842379819498013,
+      "loss": 0.0512,
+      "step": 683
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1692350208759308,
+      "learning_rate": 0.0002841919473056867,
+      "loss": 0.0637,
+      "step": 684
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.15840108692646027,
+      "learning_rate": 0.00028414584927405497,
+      "loss": 0.0224,
+      "step": 685
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12710994482040405,
+      "learning_rate": 0.0002840996878766812,
+      "loss": 0.042,
+      "step": 686
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.07157866656780243,
+      "learning_rate": 0.0002840534631353704,
+      "loss": 0.0279,
+      "step": 687
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.046079106628894806,
+      "learning_rate": 0.0002840071750719575,
+      "loss": 0.0093,
+      "step": 688
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.09579090774059296,
+      "learning_rate": 0.00028396082370830733,
+      "loss": 0.027,
+      "step": 689
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.201382577419281,
+      "learning_rate": 0.0002839144090663146,
+      "loss": 0.072,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1724497377872467,
+      "learning_rate": 0.000283867931167904,
+      "loss": 0.052,
+      "step": 691
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06867794692516327,
+      "learning_rate": 0.00028382139003503006,
+      "loss": 0.0168,
+      "step": 692
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06690815091133118,
+      "learning_rate": 0.00028377478568967704,
+      "loss": 0.0137,
+      "step": 693
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.17312408983707428,
+      "learning_rate": 0.0002837281181538593,
+      "loss": 0.0226,
+      "step": 694
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.08470922708511353,
+      "learning_rate": 0.0002836813874496208,
+      "loss": 0.0126,
+      "step": 695
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.0778060331940651,
+      "learning_rate": 0.00028363459359903565,
+      "loss": 0.0235,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.05282146856188774,
+      "learning_rate": 0.00028358773662420745,
+      "loss": 0.0086,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.09441710263490677,
+      "learning_rate": 0.00028354081654726984,
+      "loss": 0.049,
+      "step": 698
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.050999149680137634,
+      "learning_rate": 0.00028349383339038617,
+      "loss": 0.0086,
+      "step": 699
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.10146886110305786,
+      "learning_rate": 0.0002834467871757497,
+      "loss": 0.0246,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.03706188499927521,
+      "learning_rate": 0.0002833996779255833,
+      "loss": 0.0056,
+      "step": 701
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.10421967506408691,
+      "learning_rate": 0.0002833525056621397,
+      "loss": 0.0241,
+      "step": 702
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.18308381736278534,
+      "learning_rate": 0.00028330527040770146,
+      "loss": 0.042,
+      "step": 703
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2132684737443924,
+      "learning_rate": 0.0002832579721845809,
+      "loss": 0.0226,
+      "step": 704
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5411369204521179,
+      "learning_rate": 0.00028321061101511984,
+      "loss": 0.0702,
+      "step": 705
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3440389335155487,
+      "learning_rate": 0.0002831631869216902,
+      "loss": 0.0225,
+      "step": 706
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.16572096943855286,
+      "learning_rate": 0.00028311569992669333,
+      "loss": 0.0352,
+      "step": 707
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.15913799405097961,
+      "learning_rate": 0.0002830681500525604,
+      "loss": 0.0266,
+      "step": 708
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1818440854549408,
+      "learning_rate": 0.0002830205373217524,
+      "loss": 0.0688,
+      "step": 709
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.17044726014137268,
+      "learning_rate": 0.0002829728617567598,
+      "loss": 0.0515,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15003225207328796,
+      "learning_rate": 0.0002829251233801028,
+      "loss": 0.0757,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.11599011719226837,
+      "learning_rate": 0.00028287732221433145,
+      "loss": 0.0402,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.11937666684389114,
+      "learning_rate": 0.0002828294582820252,
+      "loss": 0.0391,
+      "step": 713
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2134632170200348,
+      "learning_rate": 0.0002827815316057933,
+      "loss": 0.0748,
+      "step": 714
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.29407811164855957,
+      "learning_rate": 0.00028273354220827477,
+      "loss": 0.0679,
+      "step": 715
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.059820059686899185,
+      "learning_rate": 0.00028268549011213785,
+      "loss": 0.0372,
+      "step": 716
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06392081081867218,
+      "learning_rate": 0.0002826373753400808,
+      "loss": 0.0212,
+      "step": 717
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15369698405265808,
+      "learning_rate": 0.0002825891979148313,
+      "loss": 0.0258,
+      "step": 718
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.11246192455291748,
+      "learning_rate": 0.00028254095785914667,
+      "loss": 0.0357,
+      "step": 719
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.09747990220785141,
+      "learning_rate": 0.0002824926551958138,
+      "loss": 0.0393,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06651636213064194,
+      "learning_rate": 0.0002824442899476491,
+      "loss": 0.0222,
+      "step": 721
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2205415815114975,
+      "learning_rate": 0.00028239586213749866,
+      "loss": 0.0391,
+      "step": 722
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.15302903950214386,
+      "learning_rate": 0.000282347371788238,
+      "loss": 0.019,
+      "step": 723
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.0960957333445549,
+      "learning_rate": 0.00028229881892277237,
+      "loss": 0.0142,
+      "step": 724
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13723668456077576,
+      "learning_rate": 0.00028225020356403624,
+      "loss": 0.0595,
+      "step": 725
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.11208293586969376,
+      "learning_rate": 0.00028220152573499394,
+      "loss": 0.0283,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.11818848550319672,
+      "learning_rate": 0.000282152785458639,
+      "loss": 0.0337,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1239568442106247,
+      "learning_rate": 0.0002821039827579948,
+      "loss": 0.0406,
+      "step": 728
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2423425167798996,
+      "learning_rate": 0.0002820551176561138,
+      "loss": 0.0502,
+      "step": 729
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.11348683387041092,
+      "learning_rate": 0.0002820061901760783,
+      "loss": 0.0356,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1281062662601471,
+      "learning_rate": 0.00028195720034099976,
+      "loss": 0.0172,
+      "step": 731
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.017542295157909393,
+      "learning_rate": 0.0002819081481740193,
+      "loss": 0.0035,
+      "step": 732
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13793808221817017,
+      "learning_rate": 0.00028185903369830757,
+      "loss": 0.0142,
+      "step": 733
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.0807032361626625,
+      "learning_rate": 0.0002818098569370643,
+      "loss": 0.0093,
+      "step": 734
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13335250318050385,
+      "learning_rate": 0.0002817606179135189,
+      "loss": 0.0249,
+      "step": 735
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.18836469948291779,
+      "learning_rate": 0.0002817113166509302,
+      "loss": 0.0702,
+      "step": 736
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.08416979014873505,
+      "learning_rate": 0.0002816619531725863,
+      "loss": 0.0115,
+      "step": 737
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.054892800748348236,
+      "learning_rate": 0.00028161252750180486,
+      "loss": 0.0051,
+      "step": 738
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.11004303395748138,
+      "learning_rate": 0.0002815630396619327,
+      "loss": 0.025,
+      "step": 739
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.08595911413431168,
+      "learning_rate": 0.00028151348967634613,
+      "loss": 0.0247,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.19074852764606476,
+      "learning_rate": 0.0002814638775684509,
+      "loss": 0.0548,
+      "step": 741
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.15080730617046356,
+      "learning_rate": 0.0002814142033616819,
+      "loss": 0.0865,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1833682358264923,
+      "learning_rate": 0.00028136446707950353,
+      "loss": 0.0697,
+      "step": 743
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.049776408821344376,
+      "learning_rate": 0.00028131466874540943,
+      "loss": 0.0078,
+      "step": 744
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.28232109546661377,
+      "learning_rate": 0.00028126480838292254,
+      "loss": 0.0283,
+      "step": 745
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.11048179864883423,
+      "learning_rate": 0.0002812148860155952,
+      "loss": 0.0333,
+      "step": 746
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3188829720020294,
+      "learning_rate": 0.0002811649016670089,
+      "loss": 0.0524,
+      "step": 747
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.17602626979351044,
+      "learning_rate": 0.0002811148553607745,
+      "loss": 0.0559,
+      "step": 748
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1038702130317688,
+      "learning_rate": 0.0002810647471205321,
+      "loss": 0.0409,
+      "step": 749
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13291211426258087,
+      "learning_rate": 0.00028101457696995104,
+      "loss": 0.0343,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3924441933631897,
+      "learning_rate": 0.0002809643449327299,
+      "loss": 0.1051,
+      "step": 751
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4292432367801666,
+      "learning_rate": 0.0002809140510325966,
+      "loss": 0.0985,
+      "step": 752
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2376273274421692,
+      "learning_rate": 0.0002808636952933081,
+      "loss": 0.0372,
+      "step": 753
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.18019931018352509,
+      "learning_rate": 0.0002808132777386507,
+      "loss": 0.0449,
+      "step": 754
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.09754154831171036,
+      "learning_rate": 0.0002807627983924399,
+      "loss": 0.0399,
+      "step": 755
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.15885044634342194,
+      "learning_rate": 0.0002807122572785203,
+      "loss": 0.0231,
+      "step": 756
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.15031558275222778,
+      "learning_rate": 0.0002806616544207657,
+      "loss": 0.0562,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.03422224149107933,
+      "learning_rate": 0.00028061098984307923,
+      "loss": 0.0055,
+      "step": 758
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.11442912369966507,
+      "learning_rate": 0.0002805602635693929,
+      "loss": 0.0318,
+      "step": 759
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.10043658316135406,
+      "learning_rate": 0.0002805094756236681,
+      "loss": 0.0293,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06093163788318634,
+      "learning_rate": 0.00028045862602989516,
+      "loss": 0.0062,
+      "step": 761
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17897385358810425,
+      "learning_rate": 0.0002804077148120937,
+      "loss": 0.0387,
+      "step": 762
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14174053072929382,
+      "learning_rate": 0.0002803567419943124,
+      "loss": 0.0742,
+      "step": 763
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.12193652242422104,
+      "learning_rate": 0.0002803057076006289,
+      "loss": 0.0237,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.03274312615394592,
+      "eval_runtime": 39.9218,
+      "eval_samples_per_second": 32.238,
+      "eval_steps_per_second": 8.066,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.17127107083797455,
+      "learning_rate": 0.00028025461165515016,
+      "loss": 0.0269,
+      "step": 765
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.03232162818312645,
+      "learning_rate": 0.00028020345418201196,
+      "loss": 0.0053,
+      "step": 766
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.29065465927124023,
+      "learning_rate": 0.0002801522352053794,
+      "loss": 0.0621,
+      "step": 767
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14432761073112488,
+      "learning_rate": 0.00028010095474944647,
+      "loss": 0.0556,
+      "step": 768
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.11056532710790634,
+      "learning_rate": 0.00028004961283843624,
+      "loss": 0.0111,
+      "step": 769
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.18930545449256897,
+      "learning_rate": 0.0002799982094966007,
+      "loss": 0.0548,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14607198536396027,
+      "learning_rate": 0.00027994674474822115,
+      "loss": 0.0296,
+      "step": 771
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1470440924167633,
+      "learning_rate": 0.0002798952186176076,
+      "loss": 0.0366,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.0858917385339737,
+      "learning_rate": 0.0002798436311290992,
+      "loss": 0.0149,
+      "step": 773
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23497411608695984,
+      "learning_rate": 0.000279791982307064,
+      "loss": 0.031,
+      "step": 774
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.05533986538648605,
+      "learning_rate": 0.00027974027217589917,
+      "loss": 0.0149,
+      "step": 775
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18971019983291626,
+      "learning_rate": 0.00027968850076003066,
+      "loss": 0.0339,
+      "step": 776
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.18975158035755157,
+      "learning_rate": 0.00027963666808391343,
+      "loss": 0.0192,
+      "step": 777
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.04416336864233017,
+      "learning_rate": 0.0002795847741720315,
+      "loss": 0.0073,
+      "step": 778
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.20576409995555878,
+      "learning_rate": 0.00027953281904889764,
+      "loss": 0.0418,
+      "step": 779
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1331322193145752,
+      "learning_rate": 0.0002794808027390536,
+      "loss": 0.011,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3062935769557953,
+      "learning_rate": 0.0002794287252670701,
+      "loss": 0.07,
+      "step": 781
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1513393074274063,
+      "learning_rate": 0.0002793765866575466,
+      "loss": 0.0384,
+      "step": 782
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.08953273296356201,
+      "learning_rate": 0.0002793243869351116,
+      "loss": 0.0342,
+      "step": 783
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.16210728883743286,
+      "learning_rate": 0.00027927212612442243,
+      "loss": 0.0403,
+      "step": 784
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.10174558311700821,
+      "learning_rate": 0.0002792198042501652,
+      "loss": 0.0304,
+      "step": 785
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.15236282348632812,
+      "learning_rate": 0.0002791674213370549,
+      "loss": 0.0378,
+      "step": 786
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06242475286126137,
+      "learning_rate": 0.0002791149774098353,
+      "loss": 0.0092,
+      "step": 787
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.20970316231250763,
+      "learning_rate": 0.0002790624724932792,
+      "loss": 0.0479,
+      "step": 788
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.15189824998378754,
+      "learning_rate": 0.0002790099066121879,
+      "loss": 0.0118,
+      "step": 789
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.061395786702632904,
+      "learning_rate": 0.0002789572797913918,
+      "loss": 0.0151,
+      "step": 790
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5034061074256897,
+      "learning_rate": 0.00027890459205574987,
+      "loss": 0.0864,
+      "step": 791
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.07147826254367828,
+      "learning_rate": 0.0002788518434301499,
+      "loss": 0.0191,
+      "step": 792
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1329374462366104,
+      "learning_rate": 0.0002787990339395085,
+      "loss": 0.0331,
+      "step": 793
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.15126173198223114,
+      "learning_rate": 0.0002787461636087711,
+      "loss": 0.0143,
+      "step": 794
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.09816433489322662,
+      "learning_rate": 0.0002786932324629116,
+      "loss": 0.0155,
+      "step": 795
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1379247009754181,
+      "learning_rate": 0.0002786402405269329,
+      "loss": 0.0315,
+      "step": 796
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.23714596033096313,
+      "learning_rate": 0.00027858718782586647,
+      "loss": 0.0465,
+      "step": 797
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.12600766122341156,
+      "learning_rate": 0.0002785340743847725,
+      "loss": 0.0359,
+      "step": 798
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13546015322208405,
+      "learning_rate": 0.00027848090022874,
+      "loss": 0.0175,
+      "step": 799
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.07381202280521393,
+      "learning_rate": 0.00027842766538288647,
+      "loss": 0.0302,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.14797933399677277,
+      "learning_rate": 0.0002783743698723582,
+      "loss": 0.0818,
+      "step": 801
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.02270502597093582,
+      "learning_rate": 0.00027832101372233007,
+      "loss": 0.0049,
+      "step": 802
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.322780042886734,
+      "learning_rate": 0.00027826759695800566,
+      "loss": 0.0694,
+      "step": 803
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2222844511270523,
+      "learning_rate": 0.0002782141196046171,
+      "loss": 0.0261,
+      "step": 804
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3285076320171356,
+      "learning_rate": 0.0002781605816874253,
+      "loss": 0.0872,
+      "step": 805
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.08804619312286377,
+      "learning_rate": 0.0002781069832317196,
+      "loss": 0.0578,
+      "step": 806
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.17540492117404938,
+      "learning_rate": 0.00027805332426281793,
+      "loss": 0.0384,
+      "step": 807
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06208420172333717,
+      "learning_rate": 0.00027799960480606706,
+      "loss": 0.0136,
+      "step": 808
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07048339396715164,
+      "learning_rate": 0.0002779458248868421,
+      "loss": 0.026,
+      "step": 809
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.0901297777891159,
+      "learning_rate": 0.00027789198453054666,
+      "loss": 0.0277,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07435130327939987,
+      "learning_rate": 0.0002778380837626132,
+      "loss": 0.0197,
+      "step": 811
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14209306240081787,
+      "learning_rate": 0.00027778412260850234,
+      "loss": 0.0407,
+      "step": 812
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.16662320494651794,
+      "learning_rate": 0.00027773010109370357,
+      "loss": 0.0667,
+      "step": 813
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.031582899391651154,
+      "learning_rate": 0.0002776760192437346,
+      "loss": 0.0104,
+      "step": 814
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.0818646028637886,
+      "learning_rate": 0.00027762187708414195,
+      "loss": 0.0258,
+      "step": 815
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.0822024941444397,
+      "learning_rate": 0.0002775676746405003,
+      "loss": 0.0406,
+      "step": 816
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.11700989305973053,
+      "learning_rate": 0.0002775134119384131,
+      "loss": 0.0335,
+      "step": 817
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.082905612885952,
+      "learning_rate": 0.00027745908900351195,
+      "loss": 0.0161,
+      "step": 818
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.11846023797988892,
+      "learning_rate": 0.00027740470586145726,
+      "loss": 0.0502,
+      "step": 819
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17150120437145233,
+      "learning_rate": 0.00027735026253793756,
+      "loss": 0.0345,
+      "step": 820
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.12443263083696365,
+      "learning_rate": 0.00027729575905867,
+      "loss": 0.0158,
+      "step": 821
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1657358705997467,
+      "learning_rate": 0.0002772411954494001,
+      "loss": 0.0226,
+      "step": 822
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.08840323239564896,
+      "learning_rate": 0.0002771865717359018,
+      "loss": 0.0152,
+      "step": 823
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.08032941073179245,
+      "learning_rate": 0.00027713188794397737,
+      "loss": 0.0129,
+      "step": 824
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1835167407989502,
+      "learning_rate": 0.00027707714409945744,
+      "loss": 0.0569,
+      "step": 825
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.12917861342430115,
+      "learning_rate": 0.0002770223402282012,
+      "loss": 0.0309,
+      "step": 826
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.22104112803936005,
+      "learning_rate": 0.0002769674763560959,
+      "loss": 0.0432,
+      "step": 827
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13979768753051758,
+      "learning_rate": 0.00027691255250905737,
+      "loss": 0.0174,
+      "step": 828
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17627565562725067,
+      "learning_rate": 0.0002768575687130297,
+      "loss": 0.0915,
+      "step": 829
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.486728310585022,
+      "learning_rate": 0.0002768025249939853,
+      "loss": 0.0583,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1259876936674118,
+      "learning_rate": 0.0002767474213779247,
+      "loss": 0.0254,
+      "step": 831
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.17353613674640656,
+      "learning_rate": 0.00027669225789087715,
+      "loss": 0.0238,
+      "step": 832
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.011490284465253353,
+      "learning_rate": 0.00027663703455889973,
+      "loss": 0.0025,
+      "step": 833
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.054609689861536026,
+      "learning_rate": 0.00027658175140807815,
+      "loss": 0.0098,
+      "step": 834
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1213490441441536,
+      "learning_rate": 0.000276526408464526,
+      "loss": 0.0128,
+      "step": 835
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.09482322633266449,
+      "learning_rate": 0.0002764710057543855,
+      "loss": 0.0126,
+      "step": 836
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.057049017399549484,
+      "learning_rate": 0.00027641554330382686,
+      "loss": 0.015,
+      "step": 837
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.18572884798049927,
+      "learning_rate": 0.0002763600211390486,
+      "loss": 0.034,
+      "step": 838
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.09493198245763779,
+      "learning_rate": 0.0002763044392862774,
+      "loss": 0.0408,
+      "step": 839
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2182336002588272,
+      "learning_rate": 0.00027624879777176807,
+      "loss": 0.055,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.08872721344232559,
+      "learning_rate": 0.00027619309662180386,
+      "loss": 0.0383,
+      "step": 841
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.11956200748682022,
+      "learning_rate": 0.0002761373358626959,
+      "loss": 0.0287,
+      "step": 842
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1644572764635086,
+      "learning_rate": 0.0002760815155207837,
+      "loss": 0.0286,
+      "step": 843
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.16476300358772278,
+      "learning_rate": 0.0002760256356224347,
+      "loss": 0.0392,
+      "step": 844
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1026122123003006,
+      "learning_rate": 0.00027596969619404457,
+      "loss": 0.0403,
+      "step": 845
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.17450834810733795,
+      "learning_rate": 0.00027591369726203725,
+      "loss": 0.0586,
+      "step": 846
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.10373177379369736,
+      "learning_rate": 0.0002758576388528645,
+      "loss": 0.0214,
+      "step": 847
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.08164018392562866,
+      "learning_rate": 0.0002758015209930064,
+      "loss": 0.0229,
+      "step": 848
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.07375165820121765,
+      "learning_rate": 0.000275745343708971,
+      "loss": 0.0333,
+      "step": 849
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.09719602763652802,
+      "learning_rate": 0.0002756891070272945,
+      "loss": 0.0214,
+      "step": 850
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5595388412475586,
+      "learning_rate": 0.00027563281097454115,
+      "loss": 0.0657,
+      "step": 851
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.10981204360723495,
+      "learning_rate": 0.0002755764555773031,
+      "loss": 0.0308,
+      "step": 852
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.10418907552957535,
+      "learning_rate": 0.0002755200408622007,
+      "loss": 0.0238,
+      "step": 853
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.0636146143078804,
+      "learning_rate": 0.0002754635668558822,
+      "loss": 0.0143,
+      "step": 854
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12179470807313919,
+      "learning_rate": 0.00027540703358502406,
+      "loss": 0.0393,
+      "step": 855
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.07303999364376068,
+      "learning_rate": 0.00027535044107633046,
+      "loss": 0.0118,
+      "step": 856
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.11226726323366165,
+      "learning_rate": 0.00027529378935653377,
+      "loss": 0.0356,
+      "step": 857
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.16357053816318512,
+      "learning_rate": 0.0002752370784523942,
+      "loss": 0.0378,
+      "step": 858
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.10425914824008942,
+      "learning_rate": 0.0002751803083907,
+      "loss": 0.0423,
+      "step": 859
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.11986647546291351,
+      "learning_rate": 0.0002751234791982674,
+      "loss": 0.054,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14440590143203735,
+      "learning_rate": 0.00027506659090194036,
+      "loss": 0.0418,
+      "step": 861
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.21995751559734344,
+      "learning_rate": 0.0002750096435285909,
+      "loss": 0.0303,
+      "step": 862
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.03415970876812935,
+      "learning_rate": 0.00027495263710511906,
+      "loss": 0.0084,
+      "step": 863
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.052127208560705185,
+      "learning_rate": 0.0002748955716584526,
+      "loss": 0.0124,
+      "step": 864
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.23270320892333984,
+      "learning_rate": 0.0002748384472155472,
+      "loss": 0.0501,
+      "step": 865
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.05627870559692383,
+      "learning_rate": 0.00027478126380338645,
+      "loss": 0.0081,
+      "step": 866
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1844397783279419,
+      "learning_rate": 0.0002747240214489817,
+      "loss": 0.04,
+      "step": 867
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06833455711603165,
+      "learning_rate": 0.0002746667201793722,
+      "loss": 0.0136,
+      "step": 868
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.03551473841071129,
+      "learning_rate": 0.00027460936002162513,
+      "loss": 0.0057,
+      "step": 869
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.0920785516500473,
+      "learning_rate": 0.0002745519410028354,
+      "loss": 0.0103,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1218150407075882,
+      "learning_rate": 0.0002744944631501256,
+      "loss": 0.0427,
+      "step": 871
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3496924042701721,
+      "learning_rate": 0.00027443692649064633,
+      "loss": 0.0686,
+      "step": 872
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3225466310977936,
+      "learning_rate": 0.00027437933105157585,
+      "loss": 0.0518,
+      "step": 873
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.230736643075943,
+      "learning_rate": 0.00027432167686012015,
+      "loss": 0.0468,
+      "step": 874
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.20991326868534088,
+      "learning_rate": 0.00027426396394351313,
+      "loss": 0.0595,
+      "step": 875
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.10641276091337204,
+      "learning_rate": 0.0002742061923290162,
+      "loss": 0.0353,
+      "step": 876
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06472618877887726,
+      "learning_rate": 0.00027414836204391865,
+      "loss": 0.012,
+      "step": 877
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2291422188282013,
+      "learning_rate": 0.0002740904731155375,
+      "loss": 0.0431,
+      "step": 878
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.26647308468818665,
+      "learning_rate": 0.0002740325255712175,
+      "loss": 0.1054,
+      "step": 879
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.08363434672355652,
+      "learning_rate": 0.0002739745194383309,
+      "loss": 0.011,
+      "step": 880
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.10943964123725891,
+      "learning_rate": 0.00027391645474427774,
+      "loss": 0.0331,
+      "step": 881
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2208610624074936,
+      "learning_rate": 0.0002738583315164857,
+      "loss": 0.0499,
+      "step": 882
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.09434379637241364,
+      "learning_rate": 0.00027380014978241026,
+      "loss": 0.0268,
+      "step": 883
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13045388460159302,
+      "learning_rate": 0.0002737419095695343,
+      "loss": 0.0367,
+      "step": 884
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1460418850183487,
+      "learning_rate": 0.00027368361090536844,
+      "loss": 0.0662,
+      "step": 885
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.08823563903570175,
+      "learning_rate": 0.000273625253817451,
+      "loss": 0.0387,
+      "step": 886
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.08193490654230118,
+      "learning_rate": 0.00027356683833334766,
+      "loss": 0.0357,
+      "step": 887
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1274595856666565,
+      "learning_rate": 0.00027350836448065193,
+      "loss": 0.0346,
+      "step": 888
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.061123717576265335,
+      "learning_rate": 0.0002734498322869847,
+      "loss": 0.0388,
+      "step": 889
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12708084285259247,
+      "learning_rate": 0.0002733912417799945,
+      "loss": 0.0276,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.055733684450387955,
+      "learning_rate": 0.00027333259298735756,
+      "loss": 0.0139,
+      "step": 891
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.049776624888181686,
+      "learning_rate": 0.00027327388593677727,
+      "loss": 0.0141,
+      "step": 892
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1466546654701233,
+      "learning_rate": 0.000273215120655985,
+      "loss": 0.0424,
+      "step": 893
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.04927024617791176,
+      "learning_rate": 0.00027315629717273915,
+      "loss": 0.0121,
+      "step": 894
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14217214286327362,
+      "learning_rate": 0.0002730974155148259,
+      "loss": 0.0365,
+      "step": 895
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06162632629275322,
+      "learning_rate": 0.00027303847571005904,
+      "loss": 0.0185,
+      "step": 896
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.09187627583742142,
+      "learning_rate": 0.00027297947778627947,
+      "loss": 0.024,
+      "step": 897
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.08694395422935486,
+      "learning_rate": 0.00027292042177135575,
+      "loss": 0.016,
+      "step": 898
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2407931238412857,
+      "learning_rate": 0.0002728613076931838,
+      "loss": 0.0895,
+      "step": 899
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.11447851359844208,
+      "learning_rate": 0.0002728021355796871,
+      "loss": 0.0156,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17052114009857178,
+      "learning_rate": 0.0002727429054588165,
+      "loss": 0.0686,
+      "step": 901
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.11735350638628006,
+      "learning_rate": 0.0002726836173585501,
+      "loss": 0.0458,
+      "step": 902
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1015033945441246,
+      "learning_rate": 0.0002726242713068935,
+      "loss": 0.0396,
+      "step": 903
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.09442136436700821,
+      "learning_rate": 0.00027256486733187975,
+      "loss": 0.0354,
+      "step": 904
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.051811713725328445,
+      "learning_rate": 0.0002725054054615691,
+      "loss": 0.0103,
+      "step": 905
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.09581268578767776,
+      "learning_rate": 0.00027244588572404924,
+      "loss": 0.0346,
+      "step": 906
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1265789121389389,
+      "learning_rate": 0.00027238630814743525,
+      "loss": 0.0296,
+      "step": 907
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.11578807979822159,
+      "learning_rate": 0.0002723266727598694,
+      "loss": 0.0374,
+      "step": 908
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.0634288564324379,
+      "learning_rate": 0.0002722669795895214,
+      "loss": 0.0211,
+      "step": 909
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10002614557743073,
+      "learning_rate": 0.0002722072286645881,
+      "loss": 0.0217,
+      "step": 910
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10582344233989716,
+      "learning_rate": 0.0002721474200132937,
+      "loss": 0.0262,
+      "step": 911
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.20417608320713043,
+      "learning_rate": 0.0002720875536638898,
+      "loss": 0.0303,
+      "step": 912
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06233491376042366,
+      "learning_rate": 0.00027202762964465514,
+      "loss": 0.0179,
+      "step": 913
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10917846113443375,
+      "learning_rate": 0.00027196764798389557,
+      "loss": 0.0238,
+      "step": 914
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.20902927219867706,
+      "learning_rate": 0.0002719076087099444,
+      "loss": 0.0744,
+      "step": 915
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.07525712251663208,
+      "learning_rate": 0.000271847511851162,
+      "loss": 0.0145,
+      "step": 916
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13625741004943848,
+      "learning_rate": 0.0002717873574359361,
+      "loss": 0.0557,
+      "step": 917
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10275349766016006,
+      "learning_rate": 0.00027172714549268136,
+      "loss": 0.0156,
+      "step": 918
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.07689966261386871,
+      "learning_rate": 0.0002716668760498399,
+      "loss": 0.0285,
+      "step": 919
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.051624033600091934,
+      "learning_rate": 0.00027160654913588073,
+      "loss": 0.0109,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1263073831796646,
+      "learning_rate": 0.0002715461647793003,
+      "loss": 0.03,
+      "step": 921
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.03605236858129501,
+      "learning_rate": 0.0002714857230086219,
+      "loss": 0.008,
+      "step": 922
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.09554066509008408,
+      "learning_rate": 0.0002714252238523962,
+      "loss": 0.0276,
+      "step": 923
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12727093696594238,
+      "learning_rate": 0.0002713646673392008,
+      "loss": 0.0365,
+      "step": 924
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.21029303967952728,
+      "learning_rate": 0.00027130405349764044,
+      "loss": 0.0554,
+      "step": 925
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.10958801954984665,
+      "learning_rate": 0.00027124338235634695,
+      "loss": 0.032,
+      "step": 926
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06557829678058624,
+      "learning_rate": 0.0002711826539439792,
+      "loss": 0.0145,
+      "step": 927
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.0530441552400589,
+      "learning_rate": 0.0002711218682892232,
+      "loss": 0.014,
+      "step": 928
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.11874904483556747,
+      "learning_rate": 0.00027106102542079195,
+      "loss": 0.0144,
+      "step": 929
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.07747121155261993,
+      "learning_rate": 0.0002710001253674254,
+      "loss": 0.0136,
+      "step": 930
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.055583804845809937,
+      "learning_rate": 0.0002709391681578906,
+      "loss": 0.013,
+      "step": 931
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06069410964846611,
+      "learning_rate": 0.0002708781538209815,
+      "loss": 0.0076,
+      "step": 932
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.019891362637281418,
+      "learning_rate": 0.00027081708238551927,
+      "loss": 0.0038,
+      "step": 933
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1343265175819397,
+      "learning_rate": 0.00027075595388035173,
+      "loss": 0.0307,
+      "step": 934
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.04620016738772392,
+      "learning_rate": 0.00027069476833435397,
+      "loss": 0.0048,
+      "step": 935
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1706463247537613,
+      "learning_rate": 0.00027063352577642776,
+      "loss": 0.0643,
+      "step": 936
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.058014389127492905,
+      "learning_rate": 0.0002705722262355019,
+      "loss": 0.0081,
+      "step": 937
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.11744493991136551,
+      "learning_rate": 0.0002705108697405322,
+      "loss": 0.0308,
+      "step": 938
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.08099761605262756,
+      "learning_rate": 0.00027044945632050127,
+      "loss": 0.0052,
+      "step": 939
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.29563236236572266,
+      "learning_rate": 0.00027038798600441865,
+      "loss": 0.0529,
+      "step": 940
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.043802157044410706,
+      "learning_rate": 0.0002703264588213206,
+      "loss": 0.0071,
+      "step": 941
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12684734165668488,
+      "learning_rate": 0.00027026487480027057,
+      "loss": 0.0433,
+      "step": 942
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.21014286577701569,
+      "learning_rate": 0.00027020323397035855,
+      "loss": 0.028,
+      "step": 943
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.11645261198282242,
+      "learning_rate": 0.00027014153636070157,
+      "loss": 0.0178,
+      "step": 944
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.16726157069206238,
+      "learning_rate": 0.00027007978200044324,
+      "loss": 0.0508,
+      "step": 945
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.10064594447612762,
+      "learning_rate": 0.0002700179709187543,
+      "loss": 0.0239,
+      "step": 946
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.060703571885824203,
+      "learning_rate": 0.00026995610314483205,
+      "loss": 0.0103,
+      "step": 947
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.0527808852493763,
+      "learning_rate": 0.0002698941787079006,
+      "loss": 0.0178,
+      "step": 948
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.08081556856632233,
+      "learning_rate": 0.00026983219763721086,
+      "loss": 0.0157,
+      "step": 949
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12985916435718536,
+      "learning_rate": 0.00026977015996204054,
+      "loss": 0.0575,
+      "step": 950
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15043164789676666,
+      "learning_rate": 0.00026970806571169397,
+      "loss": 0.0302,
+      "step": 951
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.024910060688853264,
+      "learning_rate": 0.00026964591491550235,
+      "loss": 0.0045,
+      "step": 952
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.10944465547800064,
+      "learning_rate": 0.00026958370760282345,
+      "loss": 0.0574,
+      "step": 953
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.114822618663311,
+      "learning_rate": 0.0002695214438030418,
+      "loss": 0.0262,
+      "step": 954
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.15373332798480988,
+      "learning_rate": 0.0002694591235455687,
+      "loss": 0.0206,
+      "step": 955
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14427144825458527,
+      "learning_rate": 0.0002693967468598419,
+      "loss": 0.0508,
+      "step": 956
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.0668393075466156,
+      "learning_rate": 0.000269334313775326,
+      "loss": 0.0195,
+      "step": 957
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06797386705875397,
+      "learning_rate": 0.00026927182432151216,
+      "loss": 0.0081,
+      "step": 958
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.21059945225715637,
+      "learning_rate": 0.00026920927852791825,
+      "loss": 0.1075,
+      "step": 959
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.10499881953001022,
+      "learning_rate": 0.0002691466764240886,
+      "loss": 0.0111,
+      "step": 960
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.033115822821855545,
+      "learning_rate": 0.00026908401803959423,
+      "loss": 0.0054,
+      "step": 961
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2655697464942932,
+      "learning_rate": 0.0002690213034040328,
+      "loss": 0.0455,
+      "step": 962
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1976163387298584,
+      "learning_rate": 0.0002689585325470284,
+      "loss": 0.0454,
+      "step": 963
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.05260282754898071,
+      "learning_rate": 0.00026889570549823184,
+      "loss": 0.0275,
+      "step": 964
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1485443115234375,
+      "learning_rate": 0.0002688328222873203,
+      "loss": 0.0191,
+      "step": 965
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.0436883270740509,
+      "learning_rate": 0.0002687698829439977,
+      "loss": 0.0099,
+      "step": 966
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12818527221679688,
+      "learning_rate": 0.00026870688749799416,
+      "loss": 0.0323,
+      "step": 967
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14603693783283234,
+      "learning_rate": 0.0002686438359790667,
+      "loss": 0.0541,
+      "step": 968
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.09324526786804199,
+      "learning_rate": 0.00026858072841699847,
+      "loss": 0.0272,
+      "step": 969
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.26789504289627075,
+      "learning_rate": 0.0002685175648415994,
+      "loss": 0.0503,
+      "step": 970
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.059855278581380844,
+      "learning_rate": 0.0002684543452827056,
+      "loss": 0.0136,
+      "step": 971
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.08910810202360153,
+      "learning_rate": 0.00026839106977017974,
+      "loss": 0.016,
+      "step": 972
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.09903378039598465,
+      "learning_rate": 0.000268327738333911,
+      "loss": 0.0307,
+      "step": 973
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.16080208122730255,
+      "learning_rate": 0.00026826435100381487,
+      "loss": 0.0318,
+      "step": 974
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.09495270997285843,
+      "learning_rate": 0.0002682009078098333,
+      "loss": 0.0591,
+      "step": 975
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.11322695016860962,
+      "learning_rate": 0.00026813740878193457,
+      "loss": 0.047,
+      "step": 976
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06805938482284546,
+      "learning_rate": 0.0002680738539501134,
+      "loss": 0.0337,
+      "step": 977
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.18398675322532654,
+      "learning_rate": 0.00026801024334439076,
+      "loss": 0.0653,
+      "step": 978
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.09730216860771179,
+      "learning_rate": 0.00026794657699481415,
+      "loss": 0.0463,
+      "step": 979
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.0954691618680954,
+      "learning_rate": 0.0002678828549314573,
+      "loss": 0.0199,
+      "step": 980
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.15214982628822327,
+      "learning_rate": 0.00026781907718442013,
+      "loss": 0.0606,
+      "step": 981
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.07308922708034515,
+      "learning_rate": 0.00026775524378382906,
+      "loss": 0.0229,
+      "step": 982
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1865328997373581,
+      "learning_rate": 0.00026769135475983676,
+      "loss": 0.0617,
+      "step": 983
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.0670800730586052,
+      "learning_rate": 0.0002676274101426221,
+      "loss": 0.0213,
+      "step": 984
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.09108185768127441,
+      "learning_rate": 0.0002675634099623903,
+      "loss": 0.0163,
+      "step": 985
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.09892558306455612,
+      "learning_rate": 0.0002674993542493727,
+      "loss": 0.0398,
+      "step": 986
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.18465696275234222,
+      "learning_rate": 0.00026743524303382695,
+      "loss": 0.0456,
+      "step": 987
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.14701491594314575,
+      "learning_rate": 0.000267371076346037,
+      "loss": 0.0217,
+      "step": 988
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.22119949758052826,
+      "learning_rate": 0.0002673068542163128,
+      "loss": 0.0337,
+      "step": 989
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07329166680574417,
+      "learning_rate": 0.0002672425766749907,
+      "loss": 0.0077,
+      "step": 990
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.08214308321475983,
+      "learning_rate": 0.0002671782437524331,
+      "loss": 0.0086,
+      "step": 991
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.16395068168640137,
+      "learning_rate": 0.0002671138554790286,
+      "loss": 0.0511,
+      "step": 992
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07903768122196198,
+      "learning_rate": 0.0002670494118851919,
+      "loss": 0.0227,
+      "step": 993
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.044391512870788574,
+      "learning_rate": 0.0002669849130013639,
+      "loss": 0.0062,
+      "step": 994
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.11790774017572403,
+      "learning_rate": 0.0002669203588580116,
+      "loss": 0.0586,
+      "step": 995
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.023213036358356476,
+      "learning_rate": 0.000266855749485628,
+      "loss": 0.004,
+      "step": 996
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1801631897687912,
+      "learning_rate": 0.0002667910849147324,
+      "loss": 0.0273,
+      "step": 997
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3998229205608368,
+      "learning_rate": 0.00026672636517587,
+      "loss": 0.0479,
+      "step": 998
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.08344905078411102,
+      "learning_rate": 0.0002666615902996121,
+      "loss": 0.0066,
+      "step": 999
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4904734194278717,
+      "learning_rate": 0.00026659676031655605,
+      "loss": 0.107,
+      "step": 1000
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.14752142131328583,
+      "learning_rate": 0.00026653187525732525,
+      "loss": 0.0567,
+      "step": 1001
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.09572061896324158,
+      "learning_rate": 0.0002664669351525691,
+      "loss": 0.045,
+      "step": 1002
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1489264965057373,
+      "learning_rate": 0.00026640194003296297,
+      "loss": 0.0181,
+      "step": 1003
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.06828869134187698,
+      "learning_rate": 0.00026633688992920833,
+      "loss": 0.0204,
+      "step": 1004
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.08580945432186127,
+      "learning_rate": 0.00026627178487203244,
+      "loss": 0.0275,
+      "step": 1005
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2796219289302826,
+      "learning_rate": 0.00026620662489218867,
+      "loss": 0.06,
+      "step": 1006
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.19413504004478455,
+      "learning_rate": 0.0002661414100204563,
+      "loss": 0.048,
+      "step": 1007
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.0517372228205204,
+      "learning_rate": 0.0002660761402876405,
+      "loss": 0.0192,
+      "step": 1008
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.12586665153503418,
+      "learning_rate": 0.0002660108157245724,
+      "loss": 0.064,
+      "step": 1009
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.055950991809368134,
+      "learning_rate": 0.000265945436362109,
+      "loss": 0.0128,
+      "step": 1010
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.03763822093605995,
+      "learning_rate": 0.00026588000223113316,
+      "loss": 0.0107,
+      "step": 1011
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.20842203497886658,
+      "learning_rate": 0.00026581451336255365,
+      "loss": 0.0668,
+      "step": 1012
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.077543243765831,
+      "learning_rate": 0.00026574896978730515,
+      "loss": 0.0218,
+      "step": 1013
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13783104717731476,
+      "learning_rate": 0.0002656833715363481,
+      "loss": 0.0431,
+      "step": 1014
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.049275536090135574,
+      "learning_rate": 0.0002656177186406687,
+      "loss": 0.012,
+      "step": 1015
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.10721635073423386,
+      "learning_rate": 0.00026555201113127907,
+      "loss": 0.0392,
+      "step": 1016
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1177641823887825,
+      "learning_rate": 0.0002654862490392172,
+      "loss": 0.0416,
+      "step": 1017
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1034293919801712,
+      "learning_rate": 0.00026542043239554677,
+      "loss": 0.0262,
+      "step": 1018
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.05769471079111099,
+      "learning_rate": 0.0002653545612313571,
+      "loss": 0.0088,
+      "step": 1019
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2152629792690277,
+      "learning_rate": 0.0002652886355777635,
+      "loss": 0.0709,
+      "step": 1020
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.0717998817563057,
+      "learning_rate": 0.0002652226554659069,
+      "loss": 0.0135,
+      "step": 1021
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.24547475576400757,
+      "learning_rate": 0.0002651566209269539,
+      "loss": 0.0627,
+      "step": 1022
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.17455288767814636,
+      "learning_rate": 0.00026509053199209697,
+      "loss": 0.0466,
+      "step": 1023
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.08559072017669678,
+      "learning_rate": 0.0002650243886925541,
+      "loss": 0.0306,
+      "step": 1024
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.16568362712860107,
+      "learning_rate": 0.0002649581910595691,
+      "loss": 0.0272,
+      "step": 1025
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.14109772443771362,
+      "learning_rate": 0.00026489193912441133,
+      "loss": 0.0241,
+      "step": 1026
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.12116571515798569,
+      "learning_rate": 0.00026482563291837586,
+      "loss": 0.0216,
+      "step": 1027
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1847831755876541,
+      "learning_rate": 0.0002647592724727835,
+      "loss": 0.046,
+      "step": 1028
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1964387595653534,
+      "learning_rate": 0.0002646928578189803,
+      "loss": 0.0223,
+      "step": 1029
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.17670650780200958,
+      "learning_rate": 0.0002646263889883385,
+      "loss": 0.0392,
+      "step": 1030
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3018537759780884,
+      "learning_rate": 0.00026455986601225544,
+      "loss": 0.0601,
+      "step": 1031
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.16954761743545532,
+      "learning_rate": 0.0002644932889221543,
+      "loss": 0.0568,
+      "step": 1032
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.07362630218267441,
+      "learning_rate": 0.0002644266577494837,
+      "loss": 0.0173,
+      "step": 1033
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.22263336181640625,
+      "learning_rate": 0.0002643599725257178,
+      "loss": 0.0528,
+      "step": 1034
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1654106080532074,
+      "learning_rate": 0.00026429323328235635,
+      "loss": 0.0264,
+      "step": 1035
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.16433385014533997,
+      "learning_rate": 0.0002642264400509247,
+      "loss": 0.0403,
+      "step": 1036
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.16119657456874847,
+      "learning_rate": 0.0002641595928629735,
+      "loss": 0.0517,
+      "step": 1037
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06720812618732452,
+      "learning_rate": 0.00026409269175007904,
+      "loss": 0.0275,
+      "step": 1038
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.08320458233356476,
+      "learning_rate": 0.000264025736743843,
+      "loss": 0.0254,
+      "step": 1039
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.10702455043792725,
+      "learning_rate": 0.00026395872787589254,
+      "loss": 0.0173,
+      "step": 1040
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1805281639099121,
+      "learning_rate": 0.0002638916651778803,
+      "loss": 0.0526,
+      "step": 1041
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1021476462483406,
+      "learning_rate": 0.0002638245486814843,
+      "loss": 0.0206,
+      "step": 1042
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.0951414480805397,
+      "learning_rate": 0.00026375737841840803,
+      "loss": 0.0165,
+      "step": 1043
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.07957201451063156,
+      "learning_rate": 0.0002636901544203804,
+      "loss": 0.0205,
+      "step": 1044
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1612643599510193,
+      "learning_rate": 0.0002636228767191555,
+      "loss": 0.0426,
+      "step": 1045
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06410415470600128,
+      "learning_rate": 0.00026355554534651296,
+      "loss": 0.0138,
+      "step": 1046
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.06664423644542694,
+      "learning_rate": 0.0002634881603342578,
+      "loss": 0.0158,
+      "step": 1047
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07890690118074417,
+      "learning_rate": 0.0002634207217142203,
+      "loss": 0.0582,
+      "step": 1048
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14806897938251495,
+      "learning_rate": 0.000263353229518256,
+      "loss": 0.059,
+      "step": 1049
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.08011514693498611,
+      "learning_rate": 0.00026328568377824587,
+      "loss": 0.0114,
+      "step": 1050
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2250976264476776,
+      "learning_rate": 0.00026321808452609615,
+      "loss": 0.0563,
+      "step": 1051
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12238743901252747,
+      "learning_rate": 0.0002631504317937383,
+      "loss": 0.027,
+      "step": 1052
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.21183420717716217,
+      "learning_rate": 0.00026308272561312903,
+      "loss": 0.0975,
+      "step": 1053
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.03879234194755554,
+      "learning_rate": 0.0002630149660162505,
+      "loss": 0.0079,
+      "step": 1054
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.0885310247540474,
+      "learning_rate": 0.0002629471530351097,
+      "loss": 0.0345,
+      "step": 1055
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15572543442249298,
+      "learning_rate": 0.0002628792867017392,
+      "loss": 0.0418,
+      "step": 1056
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.05571586638689041,
+      "learning_rate": 0.00026281136704819674,
+      "loss": 0.0148,
+      "step": 1057
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15888281166553497,
+      "learning_rate": 0.000262743394106565,
+      "loss": 0.0423,
+      "step": 1058
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.06525658816099167,
+      "learning_rate": 0.0002626753679089521,
+      "loss": 0.0179,
+      "step": 1059
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12204741686582565,
+      "learning_rate": 0.0002626072884874911,
+      "loss": 0.025,
+      "step": 1060
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.08302486687898636,
+      "learning_rate": 0.00026253915587434035,
+      "loss": 0.0346,
+      "step": 1061
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1839776635169983,
+      "learning_rate": 0.0002624709701016833,
+      "loss": 0.0328,
+      "step": 1062
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.07696244865655899,
+      "learning_rate": 0.0002624027312017285,
+      "loss": 0.0133,
+      "step": 1063
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.13141821324825287,
+      "learning_rate": 0.0002623344392067096,
+      "loss": 0.0776,
+      "step": 1064
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.10801652073860168,
+      "learning_rate": 0.00026226609414888523,
+      "loss": 0.0308,
+      "step": 1065
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.11759611964225769,
+      "learning_rate": 0.00026219769606053927,
+      "loss": 0.0555,
+      "step": 1066
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.27231448888778687,
+      "learning_rate": 0.00026212924497398044,
+      "loss": 0.1241,
+      "step": 1067
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.05362692102789879,
+      "learning_rate": 0.00026206074092154276,
+      "loss": 0.0345,
+      "step": 1068
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.09251823276281357,
+      "learning_rate": 0.0002619921839355849,
+      "loss": 0.0423,
+      "step": 1069
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.09250043332576752,
+      "learning_rate": 0.000261923574048491,
+      "loss": 0.0433,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.04056562855839729,
+      "learning_rate": 0.0002618549112926698,
+      "loss": 0.0134,
+      "step": 1071
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.04624701663851738,
+      "learning_rate": 0.0002617861957005551,
+      "loss": 0.0204,
+      "step": 1072
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.09492779523134232,
+      "learning_rate": 0.00026171742730460583,
+      "loss": 0.0252,
+      "step": 1073
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.07661382853984833,
+      "learning_rate": 0.00026164860613730567,
+      "loss": 0.0164,
+      "step": 1074
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2870001196861267,
+      "learning_rate": 0.0002615797322311633,
+      "loss": 0.0362,
+      "step": 1075
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.0941600501537323,
+      "learning_rate": 0.0002615108056187123,
+      "loss": 0.0277,
+      "step": 1076
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.09941410273313522,
+      "learning_rate": 0.00026144182633251127,
+      "loss": 0.0271,
+      "step": 1077
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.06893230229616165,
+      "learning_rate": 0.0002613727944051434,
+      "loss": 0.0264,
+      "step": 1078
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.09225239604711533,
+      "learning_rate": 0.00026130370986921707,
+      "loss": 0.0124,
+      "step": 1079
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13335295021533966,
+      "learning_rate": 0.0002612345727573653,
+      "loss": 0.0658,
+      "step": 1080
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.08353302627801895,
+      "learning_rate": 0.000261165383102246,
+      "loss": 0.0168,
+      "step": 1081
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.16986088454723358,
+      "learning_rate": 0.00026109614093654195,
+      "loss": 0.0857,
+      "step": 1082
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.07607953995466232,
+      "learning_rate": 0.00026102684629296065,
+      "loss": 0.01,
+      "step": 1083
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1080528199672699,
+      "learning_rate": 0.00026095749920423446,
+      "loss": 0.0605,
+      "step": 1084
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14226533472537994,
+      "learning_rate": 0.0002608880997031205,
+      "loss": 0.0323,
+      "step": 1085
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.0267617329955101,
+      "learning_rate": 0.0002608186478224006,
+      "loss": 0.0046,
+      "step": 1086
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.05727904289960861,
+      "learning_rate": 0.00026074914359488143,
+      "loss": 0.0111,
+      "step": 1087
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.08158308267593384,
+      "learning_rate": 0.0002606795870533942,
+      "loss": 0.0227,
+      "step": 1088
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.17422080039978027,
+      "learning_rate": 0.00026060997823079506,
+      "loss": 0.0583,
+      "step": 1089
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.19084464013576508,
+      "learning_rate": 0.0002605403171599647,
+      "loss": 0.0736,
+      "step": 1090
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.10208334028720856,
+      "learning_rate": 0.00026047060387380855,
+      "loss": 0.021,
+      "step": 1091
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13515685498714447,
+      "learning_rate": 0.0002604008384052568,
+      "loss": 0.0319,
+      "step": 1092
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.13729439675807953,
+      "learning_rate": 0.00026033102078726393,
+      "loss": 0.0292,
+      "step": 1093
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.10295616090297699,
+      "learning_rate": 0.0002602611510528095,
+      "loss": 0.0133,
+      "step": 1094
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14003846049308777,
+      "learning_rate": 0.0002601912292348975,
+      "loss": 0.0413,
+      "step": 1095
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.22413985431194305,
+      "learning_rate": 0.0002601212553665564,
+      "loss": 0.0242,
+      "step": 1096
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.13832725584506989,
+      "learning_rate": 0.0002600512294808395,
+      "loss": 0.0353,
+      "step": 1097
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.29502683877944946,
+      "learning_rate": 0.0002599811516108245,
+      "loss": 0.0362,
+      "step": 1098
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.09490124136209488,
+      "learning_rate": 0.00025991102178961366,
+      "loss": 0.014,
+      "step": 1099
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1145247370004654,
+      "learning_rate": 0.0002598408400503339,
+      "loss": 0.0294,
+      "step": 1100
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38977229595184326,
+      "learning_rate": 0.00025977060642613645,
+      "loss": 0.0827,
+      "step": 1101
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.10398557782173157,
+      "learning_rate": 0.0002597003209501973,
+      "loss": 0.0176,
+      "step": 1102
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.13759955763816833,
+      "learning_rate": 0.0002596299836557168,
+      "loss": 0.0428,
+      "step": 1103
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.05294102802872658,
+      "learning_rate": 0.0002595595945759198,
+      "loss": 0.013,
+      "step": 1104
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2116420418024063,
+      "learning_rate": 0.0002594891537440556,
+      "loss": 0.0416,
+      "step": 1105
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.0850871354341507,
+      "learning_rate": 0.00025941866119339786,
+      "loss": 0.0264,
+      "step": 1106
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.04429350420832634,
+      "learning_rate": 0.00025934811695724484,
+      "loss": 0.0088,
+      "step": 1107
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.0578470379114151,
+      "learning_rate": 0.0002592775210689192,
+      "loss": 0.0295,
+      "step": 1108
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1103309616446495,
+      "learning_rate": 0.00025920687356176784,
+      "loss": 0.0154,
+      "step": 1109
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.09454017877578735,
+      "learning_rate": 0.0002591361744691622,
+      "loss": 0.025,
+      "step": 1110
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.19059227406978607,
+      "learning_rate": 0.0002590654238244979,
+      "loss": 0.0599,
+      "step": 1111
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.08629673719406128,
+      "learning_rate": 0.0002589946216611952,
+      "loss": 0.0151,
+      "step": 1112
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.18637306988239288,
+      "learning_rate": 0.0002589237680126984,
+      "loss": 0.0496,
+      "step": 1113
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.12386718392372131,
+      "learning_rate": 0.00025885286291247634,
+      "loss": 0.0269,
+      "step": 1114
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.18383803963661194,
+      "learning_rate": 0.00025878190639402204,
+      "loss": 0.0408,
+      "step": 1115
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24928437173366547,
+      "learning_rate": 0.0002587108984908528,
+      "loss": 0.0254,
+      "step": 1116
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.023719167336821556,
+      "learning_rate": 0.00025863983923651027,
+      "loss": 0.0037,
+      "step": 1117
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.16337376832962036,
+      "learning_rate": 0.00025856872866456037,
+      "loss": 0.0529,
+      "step": 1118
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.11658964306116104,
+      "learning_rate": 0.00025849756680859317,
+      "loss": 0.063,
+      "step": 1119
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.20387554168701172,
+      "learning_rate": 0.000258426353702223,
+      "loss": 0.0605,
+      "step": 1120
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2778151035308838,
+      "learning_rate": 0.0002583550893790885,
+      "loss": 0.0476,
+      "step": 1121
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.11449744552373886,
+      "learning_rate": 0.0002582837738728522,
+      "loss": 0.0315,
+      "step": 1122
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.10286298394203186,
+      "learning_rate": 0.00025821240721720116,
+      "loss": 0.041,
+      "step": 1123
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.11522707343101501,
+      "learning_rate": 0.00025814098944584645,
+      "loss": 0.0414,
+      "step": 1124
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06536536663770676,
+      "learning_rate": 0.0002580695205925233,
+      "loss": 0.0216,
+      "step": 1125
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.0686458870768547,
+      "learning_rate": 0.00025799800069099105,
+      "loss": 0.0667,
+      "step": 1126
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07378174364566803,
+      "learning_rate": 0.0002579264297750331,
+      "loss": 0.018,
+      "step": 1127
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.05744575336575508,
+      "learning_rate": 0.0002578548078784571,
+      "loss": 0.0328,
+      "step": 1128
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1781056821346283,
+      "learning_rate": 0.0002577831350350947,
+      "loss": 0.056,
+      "step": 1129
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.11974502354860306,
+      "learning_rate": 0.0002577114112788016,
+      "loss": 0.0411,
+      "step": 1130
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07625679671764374,
+      "learning_rate": 0.00025763963664345745,
+      "loss": 0.0332,
+      "step": 1131
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07967997342348099,
+      "learning_rate": 0.00025756781116296617,
+      "loss": 0.0431,
+      "step": 1132
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14101997017860413,
+      "learning_rate": 0.0002574959348712555,
+      "loss": 0.0322,
+      "step": 1133
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12365719676017761,
+      "learning_rate": 0.00025742400780227724,
+      "loss": 0.0205,
+      "step": 1134
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14429523050785065,
+      "learning_rate": 0.0002573520299900073,
+      "loss": 0.069,
+      "step": 1135
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.021441614255309105,
+      "learning_rate": 0.0002572800014684453,
+      "loss": 0.0054,
+      "step": 1136
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.08611132204532623,
+      "learning_rate": 0.0002572079222716151,
+      "loss": 0.0442,
+      "step": 1137
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.09402936697006226,
+      "learning_rate": 0.0002571357924335642,
+      "loss": 0.0352,
+      "step": 1138
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.08581096678972244,
+      "learning_rate": 0.00025706361198836437,
+      "loss": 0.0149,
+      "step": 1139
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.0612567737698555,
+      "learning_rate": 0.0002569913809701109,
+      "loss": 0.014,
+      "step": 1140
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.10282464325428009,
+      "learning_rate": 0.0002569190994129233,
+      "loss": 0.0254,
+      "step": 1141
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07298202067613602,
+      "learning_rate": 0.00025684676735094475,
+      "loss": 0.033,
+      "step": 1142
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.06616336852312088,
+      "learning_rate": 0.0002567743848183423,
+      "loss": 0.0127,
+      "step": 1143
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.09016578644514084,
+      "learning_rate": 0.000256701951849307,
+      "loss": 0.0248,
+      "step": 1144
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.09605623781681061,
+      "learning_rate": 0.0002566294684780536,
+      "loss": 0.0554,
+      "step": 1145
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.13209934532642365,
+      "learning_rate": 0.0002565569347388206,
+      "loss": 0.0437,
+      "step": 1146
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.030348777770996094,
+      "eval_runtime": 39.9058,
+      "eval_samples_per_second": 32.251,
+      "eval_steps_per_second": 8.069,
+      "step": 1146
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.13489413261413574,
+      "learning_rate": 0.0002564843506658704,
+      "loss": 0.0214,
+      "step": 1147
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.036875851452350616,
+      "learning_rate": 0.00025641171629348916,
+      "loss": 0.0075,
+      "step": 1148
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.04911373555660248,
+      "learning_rate": 0.0002563390316559868,
+      "loss": 0.0331,
+      "step": 1149
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.02945212461054325,
+      "learning_rate": 0.0002562662967876969,
+      "loss": 0.0044,
+      "step": 1150
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.09545271843671799,
+      "learning_rate": 0.00025619351172297686,
+      "loss": 0.0342,
+      "step": 1151
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.034161727875471115,
+      "learning_rate": 0.0002561206764962079,
+      "loss": 0.0064,
+      "step": 1152
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.17162153124809265,
+      "learning_rate": 0.00025604779114179457,
+      "loss": 0.0305,
+      "step": 1153
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.10241468250751495,
+      "learning_rate": 0.0002559748556941654,
+      "loss": 0.0143,
+      "step": 1154
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19089680910110474,
+      "learning_rate": 0.0002559018701877726,
+      "loss": 0.0192,
+      "step": 1155
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.19189144670963287,
+      "learning_rate": 0.0002558288346570918,
+      "loss": 0.0385,
+      "step": 1156
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.023649632930755615,
+      "learning_rate": 0.00025575574913662256,
+      "loss": 0.0043,
+      "step": 1157
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.20011720061302185,
+      "learning_rate": 0.0002556826136608877,
+      "loss": 0.0361,
+      "step": 1158
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3903810679912567,
+      "learning_rate": 0.00025560942826443396,
+      "loss": 0.1086,
+      "step": 1159
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.0918634682893753,
+      "learning_rate": 0.0002555361929818315,
+      "loss": 0.0237,
+      "step": 1160
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.11210468411445618,
+      "learning_rate": 0.00025546290784767407,
+      "loss": 0.0432,
+      "step": 1161
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.10598167777061462,
+      "learning_rate": 0.000255389572896579,
+      "loss": 0.0304,
+      "step": 1162
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.03547512739896774,
+      "learning_rate": 0.00025531618816318697,
+      "loss": 0.014,
+      "step": 1163
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.08146083354949951,
+      "learning_rate": 0.00025524275368216245,
+      "loss": 0.0122,
+      "step": 1164
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.046655625104904175,
+      "learning_rate": 0.00025516926948819334,
+      "loss": 0.0151,
+      "step": 1165
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.09417696297168732,
+      "learning_rate": 0.0002550957356159908,
+      "loss": 0.047,
+      "step": 1166
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.08695515990257263,
+      "learning_rate": 0.00025502215210028976,
+      "loss": 0.0363,
+      "step": 1167
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.05286262556910515,
+      "learning_rate": 0.0002549485189758485,
+      "loss": 0.0331,
+      "step": 1168
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1305568516254425,
+      "learning_rate": 0.0002548748362774485,
+      "loss": 0.0552,
+      "step": 1169
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15096144378185272,
+      "learning_rate": 0.000254801104039895,
+      "loss": 0.0341,
+      "step": 1170
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07643090933561325,
+      "learning_rate": 0.0002547273222980165,
+      "loss": 0.0234,
+      "step": 1171
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.052111852914094925,
+      "learning_rate": 0.0002546534910866648,
+      "loss": 0.0278,
+      "step": 1172
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15109075605869293,
+      "learning_rate": 0.00025457961044071523,
+      "loss": 0.039,
+      "step": 1173
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.05562788248062134,
+      "learning_rate": 0.00025450568039506633,
+      "loss": 0.0214,
+      "step": 1174
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1751837581396103,
+      "learning_rate": 0.00025443170098464,
+      "loss": 0.0401,
+      "step": 1175
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.19507139921188354,
+      "learning_rate": 0.0002543576722443816,
+      "loss": 0.0331,
+      "step": 1176
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.10975005477666855,
+      "learning_rate": 0.00025428359420925966,
+      "loss": 0.0155,
+      "step": 1177
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1416396051645279,
+      "learning_rate": 0.00025420946691426586,
+      "loss": 0.0473,
+      "step": 1178
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.03987191617488861,
+      "learning_rate": 0.0002541352903944155,
+      "loss": 0.0069,
+      "step": 1179
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.34085920453071594,
+      "learning_rate": 0.00025406106468474685,
+      "loss": 0.0919,
+      "step": 1180
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.06129152700304985,
+      "learning_rate": 0.0002539867898203215,
+      "loss": 0.0129,
+      "step": 1181
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.08059722930192947,
+      "learning_rate": 0.00025391246583622427,
+      "loss": 0.0172,
+      "step": 1182
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.12509244680404663,
+      "learning_rate": 0.0002538380927675632,
+      "loss": 0.0881,
+      "step": 1183
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.21917979419231415,
+      "learning_rate": 0.00025376367064946945,
+      "loss": 0.0438,
+      "step": 1184
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.05029948800802231,
+      "learning_rate": 0.0002536891995170974,
+      "loss": 0.0102,
+      "step": 1185
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.027424413710832596,
+      "learning_rate": 0.00025361467940562463,
+      "loss": 0.0053,
+      "step": 1186
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.0775713250041008,
+      "learning_rate": 0.0002535401103502517,
+      "loss": 0.0329,
+      "step": 1187
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12953567504882812,
+      "learning_rate": 0.0002534654923862025,
+      "loss": 0.0371,
+      "step": 1188
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.07097966223955154,
+      "learning_rate": 0.00025339082554872377,
+      "loss": 0.0165,
+      "step": 1189
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1304195523262024,
+      "learning_rate": 0.0002533161098730856,
+      "loss": 0.0386,
+      "step": 1190
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.06887423247098923,
+      "learning_rate": 0.00025324134539458096,
+      "loss": 0.0221,
+      "step": 1191
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.08637112379074097,
+      "learning_rate": 0.00025316653214852596,
+      "loss": 0.0341,
+      "step": 1192
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.04632532596588135,
+      "learning_rate": 0.0002530916701702597,
+      "loss": 0.0094,
+      "step": 1193
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.11397617310285568,
+      "learning_rate": 0.00025301675949514435,
+      "loss": 0.0167,
+      "step": 1194
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.04785558953881264,
+      "learning_rate": 0.000252941800158565,
+      "loss": 0.0189,
+      "step": 1195
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.24082554876804352,
+      "learning_rate": 0.00025286679219593,
+      "loss": 0.0472,
+      "step": 1196
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14454412460327148,
+      "learning_rate": 0.00025279173564267014,
+      "loss": 0.0521,
+      "step": 1197
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.16198396682739258,
+      "learning_rate": 0.00025271663053423967,
+      "loss": 0.0606,
+      "step": 1198
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.114061638712883,
+      "learning_rate": 0.0002526414769061155,
+      "loss": 0.012,
+      "step": 1199
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1736219972372055,
+      "learning_rate": 0.00025256627479379755,
+      "loss": 0.0516,
+      "step": 1200
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.04280832037329674,
+      "learning_rate": 0.0002524910242328087,
+      "loss": 0.0073,
+      "step": 1201
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13054266571998596,
+      "learning_rate": 0.0002524157252586946,
+      "loss": 0.0295,
+      "step": 1202
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.24452893435955048,
+      "learning_rate": 0.00025234037790702375,
+      "loss": 0.0856,
+      "step": 1203
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.05776005983352661,
+      "learning_rate": 0.0002522649822133877,
+      "loss": 0.0152,
+      "step": 1204
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.27971917390823364,
+      "learning_rate": 0.0002521895382134006,
+      "loss": 0.1183,
+      "step": 1205
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.04905636981129646,
+      "learning_rate": 0.0002521140459426995,
+      "loss": 0.0126,
+      "step": 1206
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15006506443023682,
+      "learning_rate": 0.0002520385054369444,
+      "loss": 0.0811,
+      "step": 1207
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15131042897701263,
+      "learning_rate": 0.00025196291673181784,
+      "loss": 0.0401,
+      "step": 1208
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1603415459394455,
+      "learning_rate": 0.0002518872798630253,
+      "loss": 0.0448,
+      "step": 1209
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07513672858476639,
+      "learning_rate": 0.0002518115948662949,
+      "loss": 0.0401,
+      "step": 1210
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.11225542426109314,
+      "learning_rate": 0.0002517358617773776,
+      "loss": 0.039,
+      "step": 1211
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.0876198261976242,
+      "learning_rate": 0.000251660080632047,
+      "loss": 0.0239,
+      "step": 1212
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1050589308142662,
+      "learning_rate": 0.0002515842514660994,
+      "loss": 0.0258,
+      "step": 1213
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.0426226444542408,
+      "learning_rate": 0.0002515083743153539,
+      "loss": 0.0111,
+      "step": 1214
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.09025552123785019,
+      "learning_rate": 0.00025143244921565214,
+      "loss": 0.0185,
+      "step": 1215
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12371645122766495,
+      "learning_rate": 0.00025135647620285834,
+      "loss": 0.0326,
+      "step": 1216
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07417233288288116,
+      "learning_rate": 0.0002512804553128596,
+      "loss": 0.0238,
+      "step": 1217
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.10499947518110275,
+      "learning_rate": 0.0002512043865815654,
+      "loss": 0.0464,
+      "step": 1218
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.16344919800758362,
+      "learning_rate": 0.00025112827004490797,
+      "loss": 0.0373,
+      "step": 1219
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.0862027183175087,
+      "learning_rate": 0.00025105210573884203,
+      "loss": 0.0178,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.10541030019521713,
+      "learning_rate": 0.0002509758936993449,
+      "loss": 0.0377,
+      "step": 1221
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.05190376564860344,
+      "learning_rate": 0.00025089963396241643,
+      "loss": 0.0099,
+      "step": 1222
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.09249959141016006,
+      "learning_rate": 0.00025082332656407906,
+      "loss": 0.0157,
+      "step": 1223
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.02348952367901802,
+      "learning_rate": 0.00025074697154037765,
+      "loss": 0.0041,
+      "step": 1224
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12875327467918396,
+      "learning_rate": 0.0002506705689273797,
+      "loss": 0.0173,
+      "step": 1225
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.13971397280693054,
+      "learning_rate": 0.0002505941187611749,
+      "loss": 0.0381,
+      "step": 1226
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.21139316260814667,
+      "learning_rate": 0.00025051762107787583,
+      "loss": 0.0399,
+      "step": 1227
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.10346369445323944,
+      "learning_rate": 0.0002504410759136171,
+      "loss": 0.031,
+      "step": 1228
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.021524077281355858,
+      "learning_rate": 0.00025036448330455603,
+      "loss": 0.0041,
+      "step": 1229
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.21078258752822876,
+      "learning_rate": 0.0002502878432868722,
+      "loss": 0.0291,
+      "step": 1230
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.28720253705978394,
+      "learning_rate": 0.00025021115589676774,
+      "loss": 0.0318,
+      "step": 1231
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2182384580373764,
+      "learning_rate": 0.00025013442117046694,
+      "loss": 0.0407,
+      "step": 1232
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1223733052611351,
+      "learning_rate": 0.0002500576391442166,
+      "loss": 0.0189,
+      "step": 1233
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1699313372373581,
+      "learning_rate": 0.0002499808098542858,
+      "loss": 0.1081,
+      "step": 1234
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.21604309976100922,
+      "learning_rate": 0.00024990393333696603,
+      "loss": 0.0406,
+      "step": 1235
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.11065655201673508,
+      "learning_rate": 0.00024982700962857094,
+      "loss": 0.0274,
+      "step": 1236
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.10013590008020401,
+      "learning_rate": 0.0002497500387654367,
+      "loss": 0.0138,
+      "step": 1237
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.03474019467830658,
+      "learning_rate": 0.0002496730207839215,
+      "loss": 0.0067,
+      "step": 1238
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1373460739850998,
+      "learning_rate": 0.00024959595572040594,
+      "loss": 0.0382,
+      "step": 1239
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1674460619688034,
+      "learning_rate": 0.0002495188436112928,
+      "loss": 0.0187,
+      "step": 1240
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.056852634996175766,
+      "learning_rate": 0.0002494416844930072,
+      "loss": 0.02,
+      "step": 1241
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1567879319190979,
+      "learning_rate": 0.00024936447840199626,
+      "loss": 0.0488,
+      "step": 1242
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.19893474876880646,
+      "learning_rate": 0.0002492872253747294,
+      "loss": 0.0382,
+      "step": 1243
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07066723704338074,
+      "learning_rate": 0.0002492099254476983,
+      "loss": 0.0194,
+      "step": 1244
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.11466959118843079,
+      "learning_rate": 0.00024913257865741663,
+      "loss": 0.0367,
+      "step": 1245
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.08930857479572296,
+      "learning_rate": 0.0002490551850404203,
+      "loss": 0.0186,
+      "step": 1246
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.0905904471874237,
+      "learning_rate": 0.0002489777446332673,
+      "loss": 0.0349,
+      "step": 1247
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.225018709897995,
+      "learning_rate": 0.0002489002574725378,
+      "loss": 0.0579,
+      "step": 1248
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15631456673145294,
+      "learning_rate": 0.0002488227235948339,
+      "loss": 0.0361,
+      "step": 1249
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.06862124055624008,
+      "learning_rate": 0.0002487451430367798,
+      "loss": 0.0351,
+      "step": 1250
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.10271900147199631,
+      "learning_rate": 0.00024866751583502194,
+      "loss": 0.0393,
+      "step": 1251
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12624254822731018,
+      "learning_rate": 0.0002485898420262286,
+      "loss": 0.0309,
+      "step": 1252
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.116575688123703,
+      "learning_rate": 0.00024851212164709013,
+      "loss": 0.058,
+      "step": 1253
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.06756250560283661,
+      "learning_rate": 0.00024843435473431886,
+      "loss": 0.0335,
+      "step": 1254
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.20835717022418976,
+      "learning_rate": 0.0002483565413246492,
+      "loss": 0.0389,
+      "step": 1255
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.04360177740454674,
+      "learning_rate": 0.0002482786814548374,
+      "loss": 0.008,
+      "step": 1256
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1068229153752327,
+      "learning_rate": 0.0002482007751616616,
+      "loss": 0.0304,
+      "step": 1257
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.04819338023662567,
+      "learning_rate": 0.0002481228224819221,
+      "loss": 0.0098,
+      "step": 1258
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.48405715823173523,
+      "learning_rate": 0.00024804482345244105,
+      "loss": 0.0348,
+      "step": 1259
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.09796518087387085,
+      "learning_rate": 0.0002479667781100622,
+      "loss": 0.0153,
+      "step": 1260
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.13171538710594177,
+      "learning_rate": 0.0002478886864916516,
+      "loss": 0.0316,
+      "step": 1261
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.0907411128282547,
+      "learning_rate": 0.00024781054863409676,
+      "loss": 0.0169,
+      "step": 1262
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.10159718245267868,
+      "learning_rate": 0.00024773236457430745,
+      "loss": 0.013,
+      "step": 1263
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.10823512077331543,
+      "learning_rate": 0.00024765413434921495,
+      "loss": 0.0252,
+      "step": 1264
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07199376821517944,
+      "learning_rate": 0.0002475758579957724,
+      "loss": 0.0105,
+      "step": 1265
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.11216728389263153,
+      "learning_rate": 0.0002474975355509549,
+      "loss": 0.0339,
+      "step": 1266
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.16655175387859344,
+      "learning_rate": 0.00024741916705175906,
+      "loss": 0.0306,
+      "step": 1267
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.08566506952047348,
+      "learning_rate": 0.00024734075253520345,
+      "loss": 0.0329,
+      "step": 1268
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1542367786169052,
+      "learning_rate": 0.00024726229203832824,
+      "loss": 0.0284,
+      "step": 1269
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1685347855091095,
+      "learning_rate": 0.00024718378559819554,
+      "loss": 0.0385,
+      "step": 1270
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1904221624135971,
+      "learning_rate": 0.00024710523325188885,
+      "loss": 0.0435,
+      "step": 1271
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.10915929824113846,
+      "learning_rate": 0.00024702663503651357,
+      "loss": 0.0129,
+      "step": 1272
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.04411763325333595,
+      "learning_rate": 0.0002469479909891967,
+      "loss": 0.0038,
+      "step": 1273
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.22485259175300598,
+      "learning_rate": 0.0002468693011470869,
+      "loss": 0.0456,
+      "step": 1274
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.10708510875701904,
+      "learning_rate": 0.00024679056554735454,
+      "loss": 0.0192,
+      "step": 1275
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15084552764892578,
+      "learning_rate": 0.00024671178422719137,
+      "loss": 0.0293,
+      "step": 1276
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14543551206588745,
+      "learning_rate": 0.000246632957223811,
+      "loss": 0.0666,
+      "step": 1277
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1648811399936676,
+      "learning_rate": 0.00024655408457444853,
+      "loss": 0.0321,
+      "step": 1278
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16748228669166565,
+      "learning_rate": 0.00024647516631636055,
+      "loss": 0.0373,
+      "step": 1279
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.04038754105567932,
+      "learning_rate": 0.00024639620248682523,
+      "loss": 0.0049,
+      "step": 1280
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1675775945186615,
+      "learning_rate": 0.00024631719312314234,
+      "loss": 0.0517,
+      "step": 1281
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.227004274725914,
+      "learning_rate": 0.00024623813826263303,
+      "loss": 0.0445,
+      "step": 1282
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.05555510148406029,
+      "learning_rate": 0.00024615903794264005,
+      "loss": 0.0096,
+      "step": 1283
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16279524564743042,
+      "learning_rate": 0.00024607989220052766,
+      "loss": 0.0452,
+      "step": 1284
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.22099511325359344,
+      "learning_rate": 0.0002460007010736814,
+      "loss": 0.0484,
+      "step": 1285
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3313157558441162,
+      "learning_rate": 0.00024592146459950835,
+      "loss": 0.0798,
+      "step": 1286
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1560799926519394,
+      "learning_rate": 0.0002458421828154371,
+      "loss": 0.0523,
+      "step": 1287
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.0924949198961258,
+      "learning_rate": 0.0002457628557589174,
+      "loss": 0.0416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.061663124710321426,
+      "learning_rate": 0.0002456834834674207,
+      "loss": 0.0187,
+      "step": 1289
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.04804534092545509,
+      "learning_rate": 0.0002456040659784396,
+      "loss": 0.0236,
+      "step": 1290
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.09753583371639252,
+      "learning_rate": 0.00024552460332948804,
+      "loss": 0.0447,
+      "step": 1291
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.03994222357869148,
+      "learning_rate": 0.0002454450955581015,
+      "loss": 0.0098,
+      "step": 1292
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12844492495059967,
+      "learning_rate": 0.0002453655427018364,
+      "loss": 0.0234,
+      "step": 1293
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12967482209205627,
+      "learning_rate": 0.000245285944798271,
+      "loss": 0.0435,
+      "step": 1294
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.25114384293556213,
+      "learning_rate": 0.00024520630188500423,
+      "loss": 0.0539,
+      "step": 1295
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.25040391087532043,
+      "learning_rate": 0.0002451266139996568,
+      "loss": 0.037,
+      "step": 1296
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.21144863963127136,
+      "learning_rate": 0.0002450468811798703,
+      "loss": 0.0371,
+      "step": 1297
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2176048457622528,
+      "learning_rate": 0.00024496710346330776,
+      "loss": 0.0311,
+      "step": 1298
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.06802531331777573,
+      "learning_rate": 0.0002448872808876533,
+      "loss": 0.0095,
+      "step": 1299
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.11026319861412048,
+      "learning_rate": 0.0002448074134906123,
+      "loss": 0.0132,
+      "step": 1300
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2511361539363861,
+      "learning_rate": 0.00024472750130991126,
+      "loss": 0.0091,
+      "step": 1301
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42377692461013794,
+      "learning_rate": 0.0002446475443832979,
+      "loss": 0.0669,
+      "step": 1302
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.587988555431366,
+      "learning_rate": 0.000244567542748541,
+      "loss": 0.0737,
+      "step": 1303
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1163543239235878,
+      "learning_rate": 0.0002444874964434305,
+      "loss": 0.0151,
+      "step": 1304
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.036374811083078384,
+      "learning_rate": 0.00024440740550577754,
+      "loss": 0.0067,
+      "step": 1305
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07870301604270935,
+      "learning_rate": 0.00024432726997341403,
+      "loss": 0.0191,
+      "step": 1306
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.09554275870323181,
+      "learning_rate": 0.0002442470898841933,
+      "loss": 0.0169,
+      "step": 1307
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.20255301892757416,
+      "learning_rate": 0.0002441668652759896,
+      "loss": 0.0404,
+      "step": 1308
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.015268395654857159,
+      "learning_rate": 0.0002440865961866981,
+      "loss": 0.002,
+      "step": 1309
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.08300946652889252,
+      "learning_rate": 0.0002440062826542351,
+      "loss": 0.0281,
+      "step": 1310
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2025083601474762,
+      "learning_rate": 0.00024392592471653786,
+      "loss": 0.0407,
+      "step": 1311
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.01875820755958557,
+      "learning_rate": 0.0002438455224115647,
+      "loss": 0.0024,
+      "step": 1312
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.16822032630443573,
+      "learning_rate": 0.0002437650757772947,
+      "loss": 0.0356,
+      "step": 1313
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.307230681180954,
+      "learning_rate": 0.0002436845848517281,
+      "loss": 0.0277,
+      "step": 1314
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.21822179853916168,
+      "learning_rate": 0.00024360404967288586,
+      "loss": 0.0153,
+      "step": 1315
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.18913914263248444,
+      "learning_rate": 0.00024352347027881003,
+      "loss": 0.0664,
+      "step": 1316
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.26664435863494873,
+      "learning_rate": 0.0002434428467075634,
+      "loss": 0.0821,
+      "step": 1317
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.15764296054840088,
+      "learning_rate": 0.00024336217899722967,
+      "loss": 0.0663,
+      "step": 1318
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.09952249377965927,
+      "learning_rate": 0.00024328146718591352,
+      "loss": 0.0497,
+      "step": 1319
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.20798932015895844,
+      "learning_rate": 0.00024320071131174022,
+      "loss": 0.0448,
+      "step": 1320
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2187434434890747,
+      "learning_rate": 0.00024311991141285602,
+      "loss": 0.0547,
+      "step": 1321
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.08128710836172104,
+      "learning_rate": 0.00024303906752742797,
+      "loss": 0.0232,
+      "step": 1322
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1579497754573822,
+      "learning_rate": 0.00024295817969364382,
+      "loss": 0.0368,
+      "step": 1323
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3530323803424835,
+      "learning_rate": 0.00024287724794971207,
+      "loss": 0.0543,
+      "step": 1324
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.13028964400291443,
+      "learning_rate": 0.00024279627233386212,
+      "loss": 0.0562,
+      "step": 1325
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.17670606076717377,
+      "learning_rate": 0.00024271525288434385,
+      "loss": 0.033,
+      "step": 1326
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14736194908618927,
+      "learning_rate": 0.00024263418963942808,
+      "loss": 0.0403,
+      "step": 1327
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2033924162387848,
+      "learning_rate": 0.00024255308263740618,
+      "loss": 0.0584,
+      "step": 1328
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.08926638215780258,
+      "learning_rate": 0.00024247193191659016,
+      "loss": 0.0368,
+      "step": 1329
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.09010445326566696,
+      "learning_rate": 0.0002423907375153128,
+      "loss": 0.0313,
+      "step": 1330
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.07403320074081421,
+      "learning_rate": 0.00024230949947192748,
+      "loss": 0.0146,
+      "step": 1331
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.11623091250658035,
+      "learning_rate": 0.00024222821782480812,
+      "loss": 0.0308,
+      "step": 1332
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.20798785984516144,
+      "learning_rate": 0.0002421468926123493,
+      "loss": 0.0447,
+      "step": 1333
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12543538212776184,
+      "learning_rate": 0.00024206552387296621,
+      "loss": 0.0438,
+      "step": 1334
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12966863811016083,
+      "learning_rate": 0.00024198411164509447,
+      "loss": 0.0453,
+      "step": 1335
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.05985172837972641,
+      "learning_rate": 0.00024190265596719043,
+      "loss": 0.0102,
+      "step": 1336
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14263281226158142,
+      "learning_rate": 0.00024182115687773075,
+      "loss": 0.0544,
+      "step": 1337
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.190725639462471,
+      "learning_rate": 0.00024173961441521284,
+      "loss": 0.0265,
+      "step": 1338
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.29231366515159607,
+      "learning_rate": 0.00024165802861815435,
+      "loss": 0.0684,
+      "step": 1339
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.13645826280117035,
+      "learning_rate": 0.00024157639952509356,
+      "loss": 0.0577,
+      "step": 1340
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.15891732275485992,
+      "learning_rate": 0.0002414947271745892,
+      "loss": 0.0455,
+      "step": 1341
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2538587152957916,
+      "learning_rate": 0.00024141301160522037,
+      "loss": 0.0566,
+      "step": 1342
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.08588481694459915,
+      "learning_rate": 0.00024133125285558658,
+      "loss": 0.0265,
+      "step": 1343
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1366318315267563,
+      "learning_rate": 0.00024124945096430775,
+      "loss": 0.0209,
+      "step": 1344
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12919899821281433,
+      "learning_rate": 0.00024116760597002427,
+      "loss": 0.0358,
+      "step": 1345
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1527070701122284,
+      "learning_rate": 0.0002410857179113967,
+      "loss": 0.0584,
+      "step": 1346
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1441652625799179,
+      "learning_rate": 0.00024100378682710618,
+      "loss": 0.026,
+      "step": 1347
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.0560770146548748,
+      "learning_rate": 0.00024092181275585397,
+      "loss": 0.0126,
+      "step": 1348
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.23829127848148346,
+      "learning_rate": 0.00024083979573636172,
+      "loss": 0.0492,
+      "step": 1349
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.22331084311008453,
+      "learning_rate": 0.00024075773580737138,
+      "loss": 0.0374,
+      "step": 1350
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16740433871746063,
+      "learning_rate": 0.0002406756330076452,
+      "loss": 0.033,
+      "step": 1351
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.12624043226242065,
+      "learning_rate": 0.0002405934873759655,
+      "loss": 0.0254,
+      "step": 1352
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2925248444080353,
+      "learning_rate": 0.00024051129895113506,
+      "loss": 0.0966,
+      "step": 1353
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.050702452659606934,
+      "learning_rate": 0.00024042906777197676,
+      "loss": 0.0058,
+      "step": 1354
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.11182265728712082,
+      "learning_rate": 0.00024034679387733367,
+      "loss": 0.0209,
+      "step": 1355
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07762409001588821,
+      "learning_rate": 0.00024026447730606911,
+      "loss": 0.0117,
+      "step": 1356
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.05566919595003128,
+      "learning_rate": 0.00024018211809706652,
+      "loss": 0.012,
+      "step": 1357
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.026816535741090775,
+      "learning_rate": 0.00024009971628922937,
+      "loss": 0.0058,
+      "step": 1358
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.14158879220485687,
+      "learning_rate": 0.0002400172719214814,
+      "loss": 0.0242,
+      "step": 1359
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.10178912431001663,
+      "learning_rate": 0.0002399347850327664,
+      "loss": 0.0144,
+      "step": 1360
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2671686112880707,
+      "learning_rate": 0.00023985225566204834,
+      "loss": 0.1116,
+      "step": 1361
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2026364952325821,
+      "learning_rate": 0.00023976968384831107,
+      "loss": 0.0511,
+      "step": 1362
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.046000707894563675,
+      "learning_rate": 0.0002396870696305586,
+      "loss": 0.0089,
+      "step": 1363
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.243350088596344,
+      "learning_rate": 0.00023960441304781495,
+      "loss": 0.0376,
+      "step": 1364
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.053250979632139206,
+      "learning_rate": 0.0002395217141391242,
+      "loss": 0.008,
+      "step": 1365
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.08489834517240524,
+      "learning_rate": 0.0002394389729435503,
+      "loss": 0.0216,
+      "step": 1366
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.09859486669301987,
+      "learning_rate": 0.00023935618950017738,
+      "loss": 0.0253,
+      "step": 1367
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.11453449726104736,
+      "learning_rate": 0.00023927336384810933,
+      "loss": 0.0414,
+      "step": 1368
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1473090499639511,
+      "learning_rate": 0.00023919049602647005,
+      "loss": 0.0365,
+      "step": 1369
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.12153466045856476,
+      "learning_rate": 0.00023910758607440335,
+      "loss": 0.0314,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.17143134772777557,
+      "learning_rate": 0.000239024634031073,
+      "loss": 0.0928,
+      "step": 1371
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.11081311106681824,
+      "learning_rate": 0.00023894163993566257,
+      "loss": 0.0535,
+      "step": 1372
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13488808274269104,
+      "learning_rate": 0.0002388586038273755,
+      "loss": 0.0321,
+      "step": 1373
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.0592711940407753,
+      "learning_rate": 0.0002387755257454352,
+      "loss": 0.01,
+      "step": 1374
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.09835012257099152,
+      "learning_rate": 0.00023869240572908467,
+      "loss": 0.0295,
+      "step": 1375
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.071134053170681,
+      "learning_rate": 0.000238609243817587,
+      "loss": 0.0243,
+      "step": 1376
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14431652426719666,
+      "learning_rate": 0.0002385260400502248,
+      "loss": 0.0344,
+      "step": 1377
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.10391832143068314,
+      "learning_rate": 0.00023844279446630067,
+      "loss": 0.0231,
+      "step": 1378
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07357161492109299,
+      "learning_rate": 0.00023835950710513677,
+      "loss": 0.0163,
+      "step": 1379
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.16738182306289673,
+      "learning_rate": 0.00023827617800607523,
+      "loss": 0.0423,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07547144591808319,
+      "learning_rate": 0.00023819280720847774,
+      "loss": 0.0273,
+      "step": 1381
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.10503777116537094,
+      "learning_rate": 0.0002381093947517256,
+      "loss": 0.0192,
+      "step": 1382
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.0630551353096962,
+      "learning_rate": 0.00023802594067521998,
+      "loss": 0.0115,
+      "step": 1383
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.02077486738562584,
+      "learning_rate": 0.00023794244501838162,
+      "loss": 0.0045,
+      "step": 1384
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.09841371327638626,
+      "learning_rate": 0.00023785890782065087,
+      "loss": 0.0242,
+      "step": 1385
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.21591459214687347,
+      "learning_rate": 0.00023777532912148781,
+      "loss": 0.0237,
+      "step": 1386
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.03989405184984207,
+      "learning_rate": 0.000237691708960372,
+      "loss": 0.0051,
+      "step": 1387
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12305942177772522,
+      "learning_rate": 0.0002376080473768026,
+      "loss": 0.0264,
+      "step": 1388
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14408881962299347,
+      "learning_rate": 0.00023752434441029848,
+      "loss": 0.0322,
+      "step": 1389
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.04419580101966858,
+      "learning_rate": 0.00023744060010039784,
+      "loss": 0.0073,
+      "step": 1390
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.18515107035636902,
+      "learning_rate": 0.0002373568144866586,
+      "loss": 0.0465,
+      "step": 1391
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.048167865723371506,
+      "learning_rate": 0.00023727298760865812,
+      "loss": 0.0138,
+      "step": 1392
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.08519299328327179,
+      "learning_rate": 0.0002371891195059932,
+      "loss": 0.0095,
+      "step": 1393
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.21691879630088806,
+      "learning_rate": 0.00023710521021828016,
+      "loss": 0.0381,
+      "step": 1394
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.09678614884614944,
+      "learning_rate": 0.00023702125978515478,
+      "loss": 0.0099,
+      "step": 1395
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.08847987651824951,
+      "learning_rate": 0.0002369372682462723,
+      "loss": 0.0165,
+      "step": 1396
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.03246233984827995,
+      "learning_rate": 0.0002368532356413073,
+      "loss": 0.0058,
+      "step": 1397
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07045282423496246,
+      "learning_rate": 0.00023676916200995386,
+      "loss": 0.0164,
+      "step": 1398
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.05581701174378395,
+      "learning_rate": 0.00023668504739192528,
+      "loss": 0.0152,
+      "step": 1399
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15774132311344147,
+      "learning_rate": 0.0002366008918269544,
+      "loss": 0.0243,
+      "step": 1400
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.172657772898674,
+      "learning_rate": 0.00023651669535479334,
+      "loss": 0.0184,
+      "step": 1401
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.08128032833337784,
+      "learning_rate": 0.0002364324580152135,
+      "loss": 0.0186,
+      "step": 1402
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.06358969956636429,
+      "learning_rate": 0.00023634817984800554,
+      "loss": 0.0102,
+      "step": 1403
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31414860486984253,
+      "learning_rate": 0.00023626386089297958,
+      "loss": 0.0514,
+      "step": 1404
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.22831489145755768,
+      "learning_rate": 0.00023617950118996487,
+      "loss": 0.0323,
+      "step": 1405
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.048902370035648346,
+      "learning_rate": 0.00023609510077880996,
+      "loss": 0.0033,
+      "step": 1406
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.03278432413935661,
+      "learning_rate": 0.00023601065969938262,
+      "loss": 0.0031,
+      "step": 1407
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.09343546628952026,
+      "learning_rate": 0.00023592617799156977,
+      "loss": 0.0199,
+      "step": 1408
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.0755714625120163,
+      "learning_rate": 0.00023584165569527757,
+      "loss": 0.0086,
+      "step": 1409
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.28567177057266235,
+      "learning_rate": 0.00023575709285043138,
+      "loss": 0.0256,
+      "step": 1410
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1896996796131134,
+      "learning_rate": 0.0002356724894969757,
+      "loss": 0.0291,
+      "step": 1411
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1428869366645813,
+      "learning_rate": 0.0002355878456748742,
+      "loss": 0.0574,
+      "step": 1412
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.25432294607162476,
+      "learning_rate": 0.0002355031614241095,
+      "loss": 0.0433,
+      "step": 1413
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2577909231185913,
+      "learning_rate": 0.00023541843678468355,
+      "loss": 0.0376,
+      "step": 1414
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1479143500328064,
+      "learning_rate": 0.0002353336717966172,
+      "loss": 0.0248,
+      "step": 1415
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.058144185692071915,
+      "learning_rate": 0.00023524886649995043,
+      "loss": 0.0102,
+      "step": 1416
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.18476702272891998,
+      "learning_rate": 0.00023516402093474225,
+      "loss": 0.0658,
+      "step": 1417
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1367078274488449,
+      "learning_rate": 0.00023507913514107074,
+      "loss": 0.0228,
+      "step": 1418
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.05217135697603226,
+      "learning_rate": 0.00023499420915903293,
+      "loss": 0.0117,
+      "step": 1419
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.311260461807251,
+      "learning_rate": 0.00023490924302874478,
+      "loss": 0.0945,
+      "step": 1420
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.06179346889257431,
+      "learning_rate": 0.00023482423679034134,
+      "loss": 0.0102,
+      "step": 1421
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.0694802924990654,
+      "learning_rate": 0.00023473919048397652,
+      "loss": 0.0187,
+      "step": 1422
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.09105714410543442,
+      "learning_rate": 0.00023465410414982317,
+      "loss": 0.0245,
+      "step": 1423
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.10562916845083237,
+      "learning_rate": 0.0002345689778280731,
+      "loss": 0.0296,
+      "step": 1424
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07471620291471481,
+      "learning_rate": 0.00023448381155893695,
+      "loss": 0.0288,
+      "step": 1425
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.11771635711193085,
+      "learning_rate": 0.0002343986053826442,
+      "loss": 0.0165,
+      "step": 1426
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.056794993579387665,
+      "learning_rate": 0.00023431335933944323,
+      "loss": 0.02,
+      "step": 1427
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.10688856244087219,
+      "learning_rate": 0.00023422807346960131,
+      "loss": 0.037,
+      "step": 1428
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10420051217079163,
+      "learning_rate": 0.00023414274781340442,
+      "loss": 0.0211,
+      "step": 1429
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.09319007396697998,
+      "learning_rate": 0.00023405738241115737,
+      "loss": 0.0324,
+      "step": 1430
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.11446485668420792,
+      "learning_rate": 0.00023397197730318377,
+      "loss": 0.0381,
+      "step": 1431
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10845956206321716,
+      "learning_rate": 0.00023388653252982594,
+      "loss": 0.0171,
+      "step": 1432
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08544383198022842,
+      "learning_rate": 0.000233801048131445,
+      "loss": 0.0239,
+      "step": 1433
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10372909903526306,
+      "learning_rate": 0.0002337155241484207,
+      "loss": 0.0429,
+      "step": 1434
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.24167174100875854,
+      "learning_rate": 0.00023362996062115154,
+      "loss": 0.1291,
+      "step": 1435
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10461205989122391,
+      "learning_rate": 0.00023354435759005473,
+      "loss": 0.0385,
+      "step": 1436
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.14408838748931885,
+      "learning_rate": 0.0002334587150955661,
+      "loss": 0.0377,
+      "step": 1437
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08705660700798035,
+      "learning_rate": 0.0002333730331781401,
+      "loss": 0.0169,
+      "step": 1438
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10698029398918152,
+      "learning_rate": 0.00023328731187824986,
+      "loss": 0.0383,
+      "step": 1439
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.18005134165287018,
+      "learning_rate": 0.0002332015512363871,
+      "loss": 0.0408,
+      "step": 1440
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.11144935339689255,
+      "learning_rate": 0.00023311575129306202,
+      "loss": 0.0434,
+      "step": 1441
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.09303693473339081,
+      "learning_rate": 0.0002330299120888035,
+      "loss": 0.0259,
+      "step": 1442
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.09196025878190994,
+      "learning_rate": 0.00023294403366415904,
+      "loss": 0.0256,
+      "step": 1443
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.09944535046815872,
+      "learning_rate": 0.00023285811605969442,
+      "loss": 0.0691,
+      "step": 1444
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.0766916275024414,
+      "learning_rate": 0.00023277215931599417,
+      "loss": 0.0162,
+      "step": 1445
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.0471719354391098,
+      "learning_rate": 0.00023268616347366114,
+      "loss": 0.0157,
+      "step": 1446
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.0748835876584053,
+      "learning_rate": 0.0002326001285733168,
+      "loss": 0.0162,
+      "step": 1447
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2493734508752823,
+      "learning_rate": 0.0002325140546556009,
+      "loss": 0.0908,
+      "step": 1448
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1908605992794037,
+      "learning_rate": 0.0002324279417611717,
+      "loss": 0.0352,
+      "step": 1449
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.16963645815849304,
+      "learning_rate": 0.00023234178993070595,
+      "loss": 0.0597,
+      "step": 1450
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1448785662651062,
+      "learning_rate": 0.0002322555992048987,
+      "loss": 0.0341,
+      "step": 1451
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.11966606229543686,
+      "learning_rate": 0.00023216936962446334,
+      "loss": 0.0447,
+      "step": 1452
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.06863813102245331,
+      "learning_rate": 0.00023208310123013176,
+      "loss": 0.0184,
+      "step": 1453
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.08081576228141785,
+      "learning_rate": 0.000231996794062654,
+      "loss": 0.0183,
+      "step": 1454
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.04790128767490387,
+      "learning_rate": 0.00023191044816279856,
+      "loss": 0.0159,
+      "step": 1455
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.11623428761959076,
+      "learning_rate": 0.00023182406357135217,
+      "loss": 0.036,
+      "step": 1456
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.19882117211818695,
+      "learning_rate": 0.0002317376403291198,
+      "loss": 0.0356,
+      "step": 1457
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.06410811841487885,
+      "learning_rate": 0.0002316511784769248,
+      "loss": 0.0153,
+      "step": 1458
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1210549846291542,
+      "learning_rate": 0.00023156467805560862,
+      "loss": 0.0254,
+      "step": 1459
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.09589160978794098,
+      "learning_rate": 0.00023147813910603102,
+      "loss": 0.0231,
+      "step": 1460
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.05613451451063156,
+      "learning_rate": 0.00023139156166906993,
+      "loss": 0.008,
+      "step": 1461
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.12222158908843994,
+      "learning_rate": 0.00023130494578562147,
+      "loss": 0.0236,
+      "step": 1462
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.12595443427562714,
+      "learning_rate": 0.00023121829149659988,
+      "loss": 0.0284,
+      "step": 1463
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.05631411075592041,
+      "learning_rate": 0.00023113159884293762,
+      "loss": 0.0083,
+      "step": 1464
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15821842849254608,
+      "learning_rate": 0.00023104486786558516,
+      "loss": 0.0281,
+      "step": 1465
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.28132763504981995,
+      "learning_rate": 0.0002309580986055112,
+      "loss": 0.0744,
+      "step": 1466
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.08583173155784607,
+      "learning_rate": 0.00023087129110370243,
+      "loss": 0.0163,
+      "step": 1467
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1472005695104599,
+      "learning_rate": 0.00023078444540116364,
+      "loss": 0.0342,
+      "step": 1468
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15789683163166046,
+      "learning_rate": 0.0002306975615389177,
+      "loss": 0.0321,
+      "step": 1469
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.0862409770488739,
+      "learning_rate": 0.00023061063955800542,
+      "loss": 0.0337,
+      "step": 1470
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.09513189643621445,
+      "learning_rate": 0.00023052367949948562,
+      "loss": 0.0156,
+      "step": 1471
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.16023319959640503,
+      "learning_rate": 0.00023043668140443522,
+      "loss": 0.0437,
+      "step": 1472
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.28757092356681824,
+      "learning_rate": 0.0002303496453139491,
+      "loss": 0.0526,
+      "step": 1473
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.09820155799388885,
+      "learning_rate": 0.00023026257126913986,
+      "loss": 0.0087,
+      "step": 1474
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.23134587705135345,
+      "learning_rate": 0.00023017545931113822,
+      "loss": 0.0613,
+      "step": 1475
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.06153428182005882,
+      "learning_rate": 0.0002300883094810929,
+      "loss": 0.0086,
+      "step": 1476
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.17993584275245667,
+      "learning_rate": 0.00023000112182017032,
+      "loss": 0.0339,
+      "step": 1477
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.23367144167423248,
+      "learning_rate": 0.00022991389636955483,
+      "loss": 0.0785,
+      "step": 1478
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.057765256613492966,
+      "learning_rate": 0.00022982663317044864,
+      "loss": 0.0077,
+      "step": 1479
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.17549645900726318,
+      "learning_rate": 0.00022973933226407174,
+      "loss": 0.0578,
+      "step": 1480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13486583530902863,
+      "learning_rate": 0.0002296519936916621,
+      "loss": 0.0381,
+      "step": 1481
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.11634548753499985,
+      "learning_rate": 0.00022956461749447528,
+      "loss": 0.0356,
+      "step": 1482
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.03911494463682175,
+      "learning_rate": 0.0002294772037137847,
+      "loss": 0.0082,
+      "step": 1483
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13597272336483002,
+      "learning_rate": 0.0002293897523908816,
+      "loss": 0.037,
+      "step": 1484
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.03297096863389015,
+      "learning_rate": 0.0002293022635670748,
+      "loss": 0.0101,
+      "step": 1485
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1217992827296257,
+      "learning_rate": 0.00022921473728369099,
+      "loss": 0.0488,
+      "step": 1486
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.08392113447189331,
+      "learning_rate": 0.0002291271735820744,
+      "loss": 0.0213,
+      "step": 1487
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.0728277638554573,
+      "learning_rate": 0.00022903957250358707,
+      "loss": 0.0323,
+      "step": 1488
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.19564445316791534,
+      "learning_rate": 0.0002289519340896086,
+      "loss": 0.0362,
+      "step": 1489
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.09455154836177826,
+      "learning_rate": 0.00022886425838153634,
+      "loss": 0.0305,
+      "step": 1490
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.02463528886437416,
+      "learning_rate": 0.00022877654542078515,
+      "loss": 0.0055,
+      "step": 1491
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.10636550933122635,
+      "learning_rate": 0.0002286887952487875,
+      "loss": 0.0254,
+      "step": 1492
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.08179948478937149,
+      "learning_rate": 0.00022860100790699352,
+      "loss": 0.0341,
+      "step": 1493
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.04053513705730438,
+      "learning_rate": 0.00022851318343687074,
+      "loss": 0.0059,
+      "step": 1494
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.06254950165748596,
+      "learning_rate": 0.00022842532187990444,
+      "loss": 0.016,
+      "step": 1495
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.11671124398708344,
+      "learning_rate": 0.00022833742327759722,
+      "loss": 0.0316,
+      "step": 1496
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.05388714000582695,
+      "learning_rate": 0.00022824948767146926,
+      "loss": 0.0114,
+      "step": 1497
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07483407109975815,
+      "learning_rate": 0.00022816151510305824,
+      "loss": 0.0121,
+      "step": 1498
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.08650153130292892,
+      "learning_rate": 0.00022807350561391938,
+      "loss": 0.0518,
+      "step": 1499
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1296052485704422,
+      "learning_rate": 0.00022798545924562508,
+      "loss": 0.0666,
+      "step": 1500
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15292461216449738,
+      "learning_rate": 0.00022789737603976542,
+      "loss": 0.0314,
+      "step": 1501
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2241302728652954,
+      "learning_rate": 0.00022780925603794775,
+      "loss": 0.13,
+      "step": 1502
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07691671699285507,
+      "learning_rate": 0.00022772109928179688,
+      "loss": 0.0303,
+      "step": 1503
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07967071235179901,
+      "learning_rate": 0.0002276329058129548,
+      "loss": 0.0104,
+      "step": 1504
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15211229026317596,
+      "learning_rate": 0.00022754467567308114,
+      "loss": 0.0463,
+      "step": 1505
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1364462524652481,
+      "learning_rate": 0.00022745640890385263,
+      "loss": 0.0333,
+      "step": 1506
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.08477602154016495,
+      "learning_rate": 0.00022736810554696335,
+      "loss": 0.0144,
+      "step": 1507
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.030945677310228348,
+      "learning_rate": 0.0002272797656441247,
+      "loss": 0.0082,
+      "step": 1508
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.0667153000831604,
+      "learning_rate": 0.00022719138923706525,
+      "loss": 0.0285,
+      "step": 1509
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15130023658275604,
+      "learning_rate": 0.00022710297636753096,
+      "loss": 0.0493,
+      "step": 1510
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07945651561021805,
+      "learning_rate": 0.00022701452707728486,
+      "loss": 0.0181,
+      "step": 1511
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1147598847746849,
+      "learning_rate": 0.00022692604140810735,
+      "loss": 0.0377,
+      "step": 1512
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.04304948449134827,
+      "learning_rate": 0.00022683751940179588,
+      "loss": 0.0128,
+      "step": 1513
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.08819684386253357,
+      "learning_rate": 0.00022674896110016503,
+      "loss": 0.0296,
+      "step": 1514
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.06335631757974625,
+      "learning_rate": 0.0002266603665450467,
+      "loss": 0.0188,
+      "step": 1515
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.08433008193969727,
+      "learning_rate": 0.00022657173577828979,
+      "loss": 0.0251,
+      "step": 1516
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.06014099717140198,
+      "learning_rate": 0.00022648306884176034,
+      "loss": 0.0193,
+      "step": 1517
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.05266990885138512,
+      "learning_rate": 0.00022639436577734143,
+      "loss": 0.0112,
+      "step": 1518
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.10652010887861252,
+      "learning_rate": 0.00022630562662693328,
+      "loss": 0.0312,
+      "step": 1519
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.043453726917505264,
+      "learning_rate": 0.00022621685143245308,
+      "loss": 0.009,
+      "step": 1520
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0685136690735817,
+      "learning_rate": 0.00022612804023583515,
+      "loss": 0.0189,
+      "step": 1521
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.14430442452430725,
+      "learning_rate": 0.0002260391930790307,
+      "loss": 0.066,
+      "step": 1522
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0724061131477356,
+      "learning_rate": 0.00022595031000400794,
+      "loss": 0.0129,
+      "step": 1523
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.18257959187030792,
+      "learning_rate": 0.00022586139105275214,
+      "loss": 0.0434,
+      "step": 1524
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.06716416776180267,
+      "learning_rate": 0.00022577243626726548,
+      "loss": 0.0102,
+      "step": 1525
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.05102796107530594,
+      "learning_rate": 0.00022568344568956697,
+      "loss": 0.0094,
+      "step": 1526
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0711396113038063,
+      "learning_rate": 0.0002255944193616927,
+      "loss": 0.0138,
+      "step": 1527
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4581,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1527,
+  "total_flos": 1.4257736449943142e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}