Files
c1_code_10d_4s_10k/trainer_state.json

2773 lines
63 KiB
JSON
Raw Permalink Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.992,
"eval_steps": 500,
"global_step": 390,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0128,
"grad_norm": 6.394782504385327,
"learning_rate": 1.0256410256410257e-06,
"loss": 1.0426,
"step": 1
},
{
"epoch": 0.0256,
"grad_norm": 6.569259765346459,
"learning_rate": 2.0512820512820513e-06,
"loss": 1.0654,
"step": 2
},
{
"epoch": 0.0384,
"grad_norm": 6.230643663019568,
"learning_rate": 3.0769230769230774e-06,
"loss": 1.038,
"step": 3
},
{
"epoch": 0.0512,
"grad_norm": 5.728481734524085,
"learning_rate": 4.102564102564103e-06,
"loss": 1.0223,
"step": 4
},
{
"epoch": 0.064,
"grad_norm": 4.300311825507104,
"learning_rate": 5.128205128205128e-06,
"loss": 1.0198,
"step": 5
},
{
"epoch": 0.0768,
"grad_norm": 2.7305284613406227,
"learning_rate": 6.153846153846155e-06,
"loss": 0.9696,
"step": 6
},
{
"epoch": 0.0896,
"grad_norm": 2.4945340392246735,
"learning_rate": 7.17948717948718e-06,
"loss": 0.9539,
"step": 7
},
{
"epoch": 0.1024,
"grad_norm": 4.445254891417469,
"learning_rate": 8.205128205128205e-06,
"loss": 0.9903,
"step": 8
},
{
"epoch": 0.1152,
"grad_norm": 4.355662620054897,
"learning_rate": 9.230769230769232e-06,
"loss": 0.9684,
"step": 9
},
{
"epoch": 0.128,
"grad_norm": 3.994855636728309,
"learning_rate": 1.0256410256410256e-05,
"loss": 0.9146,
"step": 10
},
{
"epoch": 0.1408,
"grad_norm": 3.600143840830981,
"learning_rate": 1.1282051282051283e-05,
"loss": 0.9168,
"step": 11
},
{
"epoch": 0.1536,
"grad_norm": 2.5072401726977716,
"learning_rate": 1.230769230769231e-05,
"loss": 0.89,
"step": 12
},
{
"epoch": 0.1664,
"grad_norm": 1.5695905535048469,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.8776,
"step": 13
},
{
"epoch": 0.1792,
"grad_norm": 1.476690203321329,
"learning_rate": 1.435897435897436e-05,
"loss": 0.8364,
"step": 14
},
{
"epoch": 0.192,
"grad_norm": 1.2652613857174815,
"learning_rate": 1.5384615384615387e-05,
"loss": 0.8212,
"step": 15
},
{
"epoch": 0.2048,
"grad_norm": 1.0424739509209286,
"learning_rate": 1.641025641025641e-05,
"loss": 0.8171,
"step": 16
},
{
"epoch": 0.2176,
"grad_norm": 1.1502664604164308,
"learning_rate": 1.7435897435897438e-05,
"loss": 0.7992,
"step": 17
},
{
"epoch": 0.2304,
"grad_norm": 1.073919261807942,
"learning_rate": 1.8461538461538465e-05,
"loss": 0.7841,
"step": 18
},
{
"epoch": 0.2432,
"grad_norm": 0.9196218647189659,
"learning_rate": 1.9487179487179488e-05,
"loss": 0.788,
"step": 19
},
{
"epoch": 0.256,
"grad_norm": 1.0253545230334442,
"learning_rate": 2.0512820512820512e-05,
"loss": 0.7642,
"step": 20
},
{
"epoch": 0.2688,
"grad_norm": 1.1328893054322893,
"learning_rate": 2.153846153846154e-05,
"loss": 0.7661,
"step": 21
},
{
"epoch": 0.2816,
"grad_norm": 0.8703556504095974,
"learning_rate": 2.2564102564102566e-05,
"loss": 0.7551,
"step": 22
},
{
"epoch": 0.2944,
"grad_norm": 1.0207044696700907,
"learning_rate": 2.3589743589743593e-05,
"loss": 0.744,
"step": 23
},
{
"epoch": 0.3072,
"grad_norm": 0.7589287619845236,
"learning_rate": 2.461538461538462e-05,
"loss": 0.7488,
"step": 24
},
{
"epoch": 0.32,
"grad_norm": 0.740061535551805,
"learning_rate": 2.5641025641025646e-05,
"loss": 0.7256,
"step": 25
},
{
"epoch": 0.3328,
"grad_norm": 0.8413097464471732,
"learning_rate": 2.6666666666666667e-05,
"loss": 0.737,
"step": 26
},
{
"epoch": 0.3456,
"grad_norm": 0.9536505328257563,
"learning_rate": 2.7692307692307694e-05,
"loss": 0.7581,
"step": 27
},
{
"epoch": 0.3584,
"grad_norm": 1.4675472622777184,
"learning_rate": 2.871794871794872e-05,
"loss": 0.7118,
"step": 28
},
{
"epoch": 0.3712,
"grad_norm": 0.7463919210078415,
"learning_rate": 2.9743589743589747e-05,
"loss": 0.7304,
"step": 29
},
{
"epoch": 0.384,
"grad_norm": 1.5192710771177331,
"learning_rate": 3.0769230769230774e-05,
"loss": 0.7412,
"step": 30
},
{
"epoch": 0.3968,
"grad_norm": 0.9332868316358716,
"learning_rate": 3.1794871794871795e-05,
"loss": 0.7142,
"step": 31
},
{
"epoch": 0.4096,
"grad_norm": 1.03647672191691,
"learning_rate": 3.282051282051282e-05,
"loss": 0.7098,
"step": 32
},
{
"epoch": 0.4224,
"grad_norm": 1.0489540813968616,
"learning_rate": 3.384615384615385e-05,
"loss": 0.718,
"step": 33
},
{
"epoch": 0.4352,
"grad_norm": 1.7415718454005513,
"learning_rate": 3.4871794871794875e-05,
"loss": 0.7281,
"step": 34
},
{
"epoch": 0.448,
"grad_norm": 1.1834722099310977,
"learning_rate": 3.58974358974359e-05,
"loss": 0.7008,
"step": 35
},
{
"epoch": 0.4608,
"grad_norm": 2.474981702846964,
"learning_rate": 3.692307692307693e-05,
"loss": 0.7226,
"step": 36
},
{
"epoch": 0.4736,
"grad_norm": 2.4830307714121873,
"learning_rate": 3.794871794871795e-05,
"loss": 0.7272,
"step": 37
},
{
"epoch": 0.4864,
"grad_norm": 1.1456942753565234,
"learning_rate": 3.8974358974358976e-05,
"loss": 0.7103,
"step": 38
},
{
"epoch": 0.4992,
"grad_norm": 1.6801001628191694,
"learning_rate": 4e-05,
"loss": 0.7066,
"step": 39
},
{
"epoch": 0.512,
"grad_norm": 1.0954177872438302,
"learning_rate": 3.9999198907597046e-05,
"loss": 0.7289,
"step": 40
},
{
"epoch": 0.5248,
"grad_norm": 1.3266879656505657,
"learning_rate": 3.9996795694563096e-05,
"loss": 0.7014,
"step": 41
},
{
"epoch": 0.5376,
"grad_norm": 1.2996108899410874,
"learning_rate": 3.999279055341771e-05,
"loss": 0.6994,
"step": 42
},
{
"epoch": 0.5504,
"grad_norm": 1.2080851674726614,
"learning_rate": 3.998718380500971e-05,
"loss": 0.7036,
"step": 43
},
{
"epoch": 0.5632,
"grad_norm": 1.000152159295558,
"learning_rate": 3.997997589849145e-05,
"loss": 0.7017,
"step": 44
},
{
"epoch": 0.576,
"grad_norm": 0.9903795895140601,
"learning_rate": 3.9971167411282835e-05,
"loss": 0.7039,
"step": 45
},
{
"epoch": 0.5888,
"grad_norm": 1.0346243706022227,
"learning_rate": 3.99607590490251e-05,
"loss": 0.6928,
"step": 46
},
{
"epoch": 0.6016,
"grad_norm": 1.2310623651435908,
"learning_rate": 3.9948751645524235e-05,
"loss": 0.6926,
"step": 47
},
{
"epoch": 0.6144,
"grad_norm": 0.864595484800581,
"learning_rate": 3.9935146162684206e-05,
"loss": 0.6994,
"step": 48
},
{
"epoch": 0.6272,
"grad_norm": 0.8355436411996121,
"learning_rate": 3.9919943690429906e-05,
"loss": 0.6934,
"step": 49
},
{
"epoch": 0.64,
"grad_norm": 0.9997652645558224,
"learning_rate": 3.9903145446619837e-05,
"loss": 0.7097,
"step": 50
},
{
"epoch": 0.6528,
"grad_norm": 1.0298649825320771,
"learning_rate": 3.9884752776948564e-05,
"loss": 0.6851,
"step": 51
},
{
"epoch": 0.6656,
"grad_norm": 0.9113000462888259,
"learning_rate": 3.9864767154838864e-05,
"loss": 0.6924,
"step": 52
},
{
"epoch": 0.6784,
"grad_norm": 0.7614156810052238,
"learning_rate": 3.9843190181323744e-05,
"loss": 0.6795,
"step": 53
},
{
"epoch": 0.6912,
"grad_norm": 0.9253337882696134,
"learning_rate": 3.982002358491817e-05,
"loss": 0.6686,
"step": 54
},
{
"epoch": 0.704,
"grad_norm": 0.8701037905779012,
"learning_rate": 3.979526922148058e-05,
"loss": 0.7001,
"step": 55
},
{
"epoch": 0.7168,
"grad_norm": 0.7437912252700782,
"learning_rate": 3.9768929074064206e-05,
"loss": 0.6792,
"step": 56
},
{
"epoch": 0.7296,
"grad_norm": 0.8094577165731788,
"learning_rate": 3.9741005252758255e-05,
"loss": 0.6821,
"step": 57
},
{
"epoch": 0.7424,
"grad_norm": 0.7509802839033517,
"learning_rate": 3.971149999451886e-05,
"loss": 0.6715,
"step": 58
},
{
"epoch": 0.7552,
"grad_norm": 0.595632383004398,
"learning_rate": 3.9680415662989806e-05,
"loss": 0.6738,
"step": 59
},
{
"epoch": 0.768,
"grad_norm": 0.738106832398729,
"learning_rate": 3.9647754748313294e-05,
"loss": 0.6776,
"step": 60
},
{
"epoch": 0.7808,
"grad_norm": 0.9793451997195339,
"learning_rate": 3.96135198669304e-05,
"loss": 0.6723,
"step": 61
},
{
"epoch": 0.7936,
"grad_norm": 0.9038400518704891,
"learning_rate": 3.957771376137144e-05,
"loss": 0.6804,
"step": 62
},
{
"epoch": 0.8064,
"grad_norm": 0.8940165413819113,
"learning_rate": 3.954033930003634e-05,
"loss": 0.6866,
"step": 63
},
{
"epoch": 0.8192,
"grad_norm": 0.6916586084691009,
"learning_rate": 3.9501399476964806e-05,
"loss": 0.6812,
"step": 64
},
{
"epoch": 0.832,
"grad_norm": 0.5621936446481481,
"learning_rate": 3.946089741159648e-05,
"loss": 0.6734,
"step": 65
},
{
"epoch": 0.8448,
"grad_norm": 0.586800751526733,
"learning_rate": 3.9418836348521045e-05,
"loss": 0.6962,
"step": 66
},
{
"epoch": 0.8576,
"grad_norm": 0.6760075379935138,
"learning_rate": 3.937521965721831e-05,
"loss": 0.6791,
"step": 67
},
{
"epoch": 0.8704,
"grad_norm": 0.7234942389255747,
"learning_rate": 3.933005083178828e-05,
"loss": 0.672,
"step": 68
},
{
"epoch": 0.8832,
"grad_norm": 0.609112508064292,
"learning_rate": 3.928333349067125e-05,
"loss": 0.69,
"step": 69
},
{
"epoch": 0.896,
"grad_norm": 0.80856892769616,
"learning_rate": 3.923507137635792e-05,
"loss": 0.6679,
"step": 70
},
{
"epoch": 0.9088,
"grad_norm": 0.8135651692515927,
"learning_rate": 3.9185268355089606e-05,
"loss": 0.6576,
"step": 71
},
{
"epoch": 0.9216,
"grad_norm": 0.8569740956861722,
"learning_rate": 3.913392841654851e-05,
"loss": 0.6752,
"step": 72
},
{
"epoch": 0.9344,
"grad_norm": 0.9734106522191275,
"learning_rate": 3.9081055673538093e-05,
"loss": 0.666,
"step": 73
},
{
"epoch": 0.9472,
"grad_norm": 0.9734454565935635,
"learning_rate": 3.902665436165364e-05,
"loss": 0.6604,
"step": 74
},
{
"epoch": 0.96,
"grad_norm": 0.8611505130270999,
"learning_rate": 3.897072883894291e-05,
"loss": 0.6701,
"step": 75
},
{
"epoch": 0.9728,
"grad_norm": 0.8768484096515019,
"learning_rate": 3.8913283585557054e-05,
"loss": 0.6736,
"step": 76
},
{
"epoch": 0.9856,
"grad_norm": 0.6899441671900481,
"learning_rate": 3.885432320339167e-05,
"loss": 0.6857,
"step": 77
},
{
"epoch": 0.9984,
"grad_norm": 0.619076353872035,
"learning_rate": 3.879385241571817e-05,
"loss": 0.657,
"step": 78
},
{
"epoch": 1.0112,
"grad_norm": 0.6689305394134695,
"learning_rate": 3.873187606680543e-05,
"loss": 0.6226,
"step": 79
},
{
"epoch": 1.024,
"grad_norm": 0.5388474756672614,
"learning_rate": 3.866839912153168e-05,
"loss": 0.5975,
"step": 80
},
{
"epoch": 1.0368,
"grad_norm": 0.6898767568816427,
"learning_rate": 3.860342666498677e-05,
"loss": 0.6125,
"step": 81
},
{
"epoch": 1.0496,
"grad_norm": 0.8180878679575876,
"learning_rate": 3.853696390206484e-05,
"loss": 0.6142,
"step": 82
},
{
"epoch": 1.0624,
"grad_norm": 1.0725542408684852,
"learning_rate": 3.846901615704734e-05,
"loss": 0.6042,
"step": 83
},
{
"epoch": 1.0752,
"grad_norm": 0.92505515366403,
"learning_rate": 3.839958887317649e-05,
"loss": 0.6097,
"step": 84
},
{
"epoch": 1.088,
"grad_norm": 0.7689887829776089,
"learning_rate": 3.832868761221926e-05,
"loss": 0.6127,
"step": 85
},
{
"epoch": 1.1008,
"grad_norm": 0.9516898192549433,
"learning_rate": 3.825631805402182e-05,
"loss": 0.6127,
"step": 86
},
{
"epoch": 1.1136,
"grad_norm": 0.7950966316561936,
"learning_rate": 3.818248599605448e-05,
"loss": 0.6091,
"step": 87
},
{
"epoch": 1.1264,
"grad_norm": 0.6665255828836638,
"learning_rate": 3.810719735294731e-05,
"loss": 0.6156,
"step": 88
},
{
"epoch": 1.1392,
"grad_norm": 0.7540335740571178,
"learning_rate": 3.8030458156016326e-05,
"loss": 0.6051,
"step": 89
},
{
"epoch": 1.152,
"grad_norm": 0.8629803559587889,
"learning_rate": 3.795227455278029e-05,
"loss": 0.6066,
"step": 90
},
{
"epoch": 1.1648,
"grad_norm": 0.6079801774673296,
"learning_rate": 3.787265280646825e-05,
"loss": 0.5973,
"step": 91
},
{
"epoch": 1.1776,
"grad_norm": 0.7646030455094784,
"learning_rate": 3.7791599295517825e-05,
"loss": 0.609,
"step": 92
},
{
"epoch": 1.1904,
"grad_norm": 0.6181653509538091,
"learning_rate": 3.7709120513064196e-05,
"loss": 0.6243,
"step": 93
},
{
"epoch": 1.2032,
"grad_norm": 0.6819268210545444,
"learning_rate": 3.762522306641998e-05,
"loss": 0.6006,
"step": 94
},
{
"epoch": 1.216,
"grad_norm": 0.7024059883235171,
"learning_rate": 3.7539913676545874e-05,
"loss": 0.6086,
"step": 95
},
{
"epoch": 1.2288000000000001,
"grad_norm": 0.575753668683066,
"learning_rate": 3.745319917751229e-05,
"loss": 0.611,
"step": 96
},
{
"epoch": 1.2416,
"grad_norm": 0.8041432384680723,
"learning_rate": 3.736508651595188e-05,
"loss": 0.5997,
"step": 97
},
{
"epoch": 1.2544,
"grad_norm": 0.44174254522297657,
"learning_rate": 3.727558275050301e-05,
"loss": 0.6057,
"step": 98
},
{
"epoch": 1.2671999999999999,
"grad_norm": 0.6028624132341728,
"learning_rate": 3.718469505124434e-05,
"loss": 0.5939,
"step": 99
},
{
"epoch": 1.28,
"grad_norm": 0.4607737998378977,
"learning_rate": 3.709243069912041e-05,
"loss": 0.5954,
"step": 100
},
{
"epoch": 1.2928,
"grad_norm": 0.5735717821893057,
"learning_rate": 3.699879708535838e-05,
"loss": 0.6158,
"step": 101
},
{
"epoch": 1.3056,
"grad_norm": 0.6925796120242844,
"learning_rate": 3.69038017108759e-05,
"loss": 0.5978,
"step": 102
},
{
"epoch": 1.3184,
"grad_norm": 0.7422327396906829,
"learning_rate": 3.680745218568026e-05,
"loss": 0.5922,
"step": 103
},
{
"epoch": 1.3312,
"grad_norm": 0.563332719790965,
"learning_rate": 3.6709756228258735e-05,
"loss": 0.6167,
"step": 104
},
{
"epoch": 1.3439999999999999,
"grad_norm": 0.5609951130215706,
"learning_rate": 3.6610721664960236e-05,
"loss": 0.6129,
"step": 105
},
{
"epoch": 1.3568,
"grad_norm": 0.7534019038629993,
"learning_rate": 3.65103564293684e-05,
"loss": 0.596,
"step": 106
},
{
"epoch": 1.3696,
"grad_norm": 0.5055194387609822,
"learning_rate": 3.640866856166601e-05,
"loss": 0.5904,
"step": 107
},
{
"epoch": 1.3824,
"grad_norm": 0.5558736753816589,
"learning_rate": 3.6305666207990886e-05,
"loss": 0.6202,
"step": 108
},
{
"epoch": 1.3952,
"grad_norm": 0.4617367226090242,
"learning_rate": 3.6201357619783336e-05,
"loss": 0.6051,
"step": 109
},
{
"epoch": 1.408,
"grad_norm": 0.5344658374625233,
"learning_rate": 3.609575115312511e-05,
"loss": 0.6219,
"step": 110
},
{
"epoch": 1.4208,
"grad_norm": 0.5782265860951042,
"learning_rate": 3.598885526807003e-05,
"loss": 0.6054,
"step": 111
},
{
"epoch": 1.4336,
"grad_norm": 0.42442597924731185,
"learning_rate": 3.5880678527966224e-05,
"loss": 0.5959,
"step": 112
},
{
"epoch": 1.4464000000000001,
"grad_norm": 0.6114986758491804,
"learning_rate": 3.577122959877017e-05,
"loss": 0.599,
"step": 113
},
{
"epoch": 1.4592,
"grad_norm": 0.46032848870907744,
"learning_rate": 3.566051724835245e-05,
"loss": 0.6114,
"step": 114
},
{
"epoch": 1.472,
"grad_norm": 0.5009551581292482,
"learning_rate": 3.554855034579532e-05,
"loss": 0.5967,
"step": 115
},
{
"epoch": 1.4848,
"grad_norm": 0.499487518866959,
"learning_rate": 3.5435337860682304e-05,
"loss": 0.6005,
"step": 116
},
{
"epoch": 1.4976,
"grad_norm": 0.4637149314493999,
"learning_rate": 3.532088886237956e-05,
"loss": 0.5997,
"step": 117
},
{
"epoch": 1.5104,
"grad_norm": 0.5997649370066328,
"learning_rate": 3.520521251930941e-05,
"loss": 0.6009,
"step": 118
},
{
"epoch": 1.5232,
"grad_norm": 0.49394183241808043,
"learning_rate": 3.5088318098215805e-05,
"loss": 0.5893,
"step": 119
},
{
"epoch": 1.536,
"grad_norm": 0.47388103629083245,
"learning_rate": 3.497021496342203e-05,
"loss": 0.6036,
"step": 120
},
{
"epoch": 1.5488,
"grad_norm": 0.542093278623551,
"learning_rate": 3.485091257608047e-05,
"loss": 0.6002,
"step": 121
},
{
"epoch": 1.5615999999999999,
"grad_norm": 0.6431580839857001,
"learning_rate": 3.473042049341474e-05,
"loss": 0.6137,
"step": 122
},
{
"epoch": 1.5744,
"grad_norm": 0.5003369599376405,
"learning_rate": 3.4608748367954064e-05,
"loss": 0.6043,
"step": 123
},
{
"epoch": 1.5872000000000002,
"grad_norm": 0.6569135072252106,
"learning_rate": 3.4485905946759965e-05,
"loss": 0.5843,
"step": 124
},
{
"epoch": 1.6,
"grad_norm": 0.5280937820334869,
"learning_rate": 3.4361903070645484e-05,
"loss": 0.5968,
"step": 125
},
{
"epoch": 1.6128,
"grad_norm": 0.47560785703947545,
"learning_rate": 3.423674967338681e-05,
"loss": 0.5962,
"step": 126
},
{
"epoch": 1.6256,
"grad_norm": 0.6169951917734525,
"learning_rate": 3.411045578092754e-05,
"loss": 0.5949,
"step": 127
},
{
"epoch": 1.6383999999999999,
"grad_norm": 0.4562836643712705,
"learning_rate": 3.398303151057543e-05,
"loss": 0.5983,
"step": 128
},
{
"epoch": 1.6512,
"grad_norm": 0.49312549921360643,
"learning_rate": 3.385448707019199e-05,
"loss": 0.5918,
"step": 129
},
{
"epoch": 1.6640000000000001,
"grad_norm": 0.5860158062199761,
"learning_rate": 3.372483275737468e-05,
"loss": 0.606,
"step": 130
},
{
"epoch": 1.6768,
"grad_norm": 0.5490285575430652,
"learning_rate": 3.359407895863199e-05,
"loss": 0.6001,
"step": 131
},
{
"epoch": 1.6896,
"grad_norm": 0.6348820038195376,
"learning_rate": 3.34622361485514e-05,
"loss": 0.6024,
"step": 132
},
{
"epoch": 1.7024,
"grad_norm": 0.42576475419719334,
"learning_rate": 3.332931488896029e-05,
"loss": 0.5933,
"step": 133
},
{
"epoch": 1.7151999999999998,
"grad_norm": 0.5900041503972492,
"learning_rate": 3.319532582807977e-05,
"loss": 0.5932,
"step": 134
},
{
"epoch": 1.728,
"grad_norm": 0.39127515320743367,
"learning_rate": 3.30602796996717e-05,
"loss": 0.6069,
"step": 135
},
{
"epoch": 1.7408000000000001,
"grad_norm": 0.4498915823225618,
"learning_rate": 3.2924187322178865e-05,
"loss": 0.5996,
"step": 136
},
{
"epoch": 1.7536,
"grad_norm": 0.42155081455650345,
"learning_rate": 3.278705959785821e-05,
"loss": 0.609,
"step": 137
},
{
"epoch": 1.7664,
"grad_norm": 0.4812165543426556,
"learning_rate": 3.2648907511907544e-05,
"loss": 0.5936,
"step": 138
},
{
"epoch": 1.7792,
"grad_norm": 0.39696305993195713,
"learning_rate": 3.250974213158555e-05,
"loss": 0.5788,
"step": 139
},
{
"epoch": 1.792,
"grad_norm": 0.4365984108187007,
"learning_rate": 3.23695746053251e-05,
"loss": 0.5921,
"step": 140
},
{
"epoch": 1.8048,
"grad_norm": 0.5408967696030647,
"learning_rate": 3.222841616184025e-05,
"loss": 0.6161,
"step": 141
},
{
"epoch": 1.8176,
"grad_norm": 0.4144438614201316,
"learning_rate": 3.208627810922665e-05,
"loss": 0.5922,
"step": 142
},
{
"epoch": 1.8304,
"grad_norm": 0.5737519474487879,
"learning_rate": 3.194317183405573e-05,
"loss": 0.5768,
"step": 143
},
{
"epoch": 1.8432,
"grad_norm": 0.47055855879022757,
"learning_rate": 3.1799108800462466e-05,
"loss": 0.5881,
"step": 144
},
{
"epoch": 1.8559999999999999,
"grad_norm": 0.5077652807841551,
"learning_rate": 3.1654100549227024e-05,
"loss": 0.5979,
"step": 145
},
{
"epoch": 1.8688,
"grad_norm": 0.5925352581007367,
"learning_rate": 3.1508158696850275e-05,
"loss": 0.592,
"step": 146
},
{
"epoch": 1.8816000000000002,
"grad_norm": 0.43648166295010277,
"learning_rate": 3.136129493462312e-05,
"loss": 0.5953,
"step": 147
},
{
"epoch": 1.8944,
"grad_norm": 0.4711597990350436,
"learning_rate": 3.121352102768998e-05,
"loss": 0.5995,
"step": 148
},
{
"epoch": 1.9072,
"grad_norm": 0.4476044182976627,
"learning_rate": 3.106484881410628e-05,
"loss": 0.5979,
"step": 149
},
{
"epoch": 1.92,
"grad_norm": 0.38283511609840415,
"learning_rate": 3.091529020389009e-05,
"loss": 0.5863,
"step": 150
},
{
"epoch": 1.9327999999999999,
"grad_norm": 0.5371783397006327,
"learning_rate": 3.076485717806808e-05,
"loss": 0.5816,
"step": 151
},
{
"epoch": 1.9456,
"grad_norm": 0.45170660642774,
"learning_rate": 3.061356178771564e-05,
"loss": 0.5815,
"step": 152
},
{
"epoch": 1.9584000000000001,
"grad_norm": 0.4753466564865614,
"learning_rate": 3.0461416152991555e-05,
"loss": 0.5893,
"step": 153
},
{
"epoch": 1.9712,
"grad_norm": 0.597902042974497,
"learning_rate": 3.0308432462167045e-05,
"loss": 0.5974,
"step": 154
},
{
"epoch": 1.984,
"grad_norm": 0.4247429930070502,
"learning_rate": 3.015462297064936e-05,
"loss": 0.5916,
"step": 155
},
{
"epoch": 1.9968,
"grad_norm": 0.5700650883732693,
"learning_rate": 3.0000000000000004e-05,
"loss": 0.5932,
"step": 156
},
{
"epoch": 2.0096,
"grad_norm": 0.46544682012163896,
"learning_rate": 2.98445759369477e-05,
"loss": 0.5451,
"step": 157
},
{
"epoch": 2.0224,
"grad_norm": 0.45971846284651463,
"learning_rate": 2.9688363232396056e-05,
"loss": 0.5179,
"step": 158
},
{
"epoch": 2.0352,
"grad_norm": 0.6999433468335587,
"learning_rate": 2.9531374400426158e-05,
"loss": 0.5069,
"step": 159
},
{
"epoch": 2.048,
"grad_norm": 0.6367425413782538,
"learning_rate": 2.9373622017294075e-05,
"loss": 0.5256,
"step": 160
},
{
"epoch": 2.0608,
"grad_norm": 0.6608868802831678,
"learning_rate": 2.9215118720423375e-05,
"loss": 0.5169,
"step": 161
},
{
"epoch": 2.0736,
"grad_norm": 0.48788367916038633,
"learning_rate": 2.9055877207392752e-05,
"loss": 0.5095,
"step": 162
},
{
"epoch": 2.0864,
"grad_norm": 0.6797298089521543,
"learning_rate": 2.8895910234918828e-05,
"loss": 0.5171,
"step": 163
},
{
"epoch": 2.0992,
"grad_norm": 0.6895143600985311,
"learning_rate": 2.873523061783426e-05,
"loss": 0.506,
"step": 164
},
{
"epoch": 2.112,
"grad_norm": 0.5092110549661606,
"learning_rate": 2.8573851228061084e-05,
"loss": 0.511,
"step": 165
},
{
"epoch": 2.1248,
"grad_norm": 0.6408741928897569,
"learning_rate": 2.8411784993579633e-05,
"loss": 0.5179,
"step": 166
},
{
"epoch": 2.1376,
"grad_norm": 0.5003463226866105,
"learning_rate": 2.8249044897392814e-05,
"loss": 0.5217,
"step": 167
},
{
"epoch": 2.1504,
"grad_norm": 0.46482236893081896,
"learning_rate": 2.80856439764861e-05,
"loss": 0.5162,
"step": 168
},
{
"epoch": 2.1632,
"grad_norm": 0.5535382177980431,
"learning_rate": 2.792159532078314e-05,
"loss": 0.5089,
"step": 169
},
{
"epoch": 2.176,
"grad_norm": 0.5069367991468174,
"learning_rate": 2.77569120720971e-05,
"loss": 0.517,
"step": 170
},
{
"epoch": 2.1888,
"grad_norm": 0.4895756496702128,
"learning_rate": 2.7591607423077932e-05,
"loss": 0.5204,
"step": 171
},
{
"epoch": 2.2016,
"grad_norm": 0.4484537242040031,
"learning_rate": 2.7425694616155474e-05,
"loss": 0.5006,
"step": 172
},
{
"epoch": 2.2144,
"grad_norm": 0.39926621412366664,
"learning_rate": 2.7259186942478656e-05,
"loss": 0.5018,
"step": 173
},
{
"epoch": 2.2272,
"grad_norm": 0.46792399045602356,
"learning_rate": 2.7092097740850712e-05,
"loss": 0.5061,
"step": 174
},
{
"epoch": 2.24,
"grad_norm": 0.4440563705223137,
"learning_rate": 2.692444039666066e-05,
"loss": 0.5019,
"step": 175
},
{
"epoch": 2.2528,
"grad_norm": 0.46549376329458764,
"learning_rate": 2.6756228340810946e-05,
"loss": 0.5035,
"step": 176
},
{
"epoch": 2.2656,
"grad_norm": 0.38653389290400647,
"learning_rate": 2.6587475048641596e-05,
"loss": 0.5084,
"step": 177
},
{
"epoch": 2.2784,
"grad_norm": 0.48544129475234504,
"learning_rate": 2.6418194038850634e-05,
"loss": 0.5094,
"step": 178
},
{
"epoch": 2.2912,
"grad_norm": 0.3900570960351546,
"learning_rate": 2.624839887241115e-05,
"loss": 0.5108,
"step": 179
},
{
"epoch": 2.304,
"grad_norm": 0.5069783257446727,
"learning_rate": 2.607810315148494e-05,
"loss": 0.5041,
"step": 180
},
{
"epoch": 2.3168,
"grad_norm": 0.3945371522760007,
"learning_rate": 2.5907320518332827e-05,
"loss": 0.5099,
"step": 181
},
{
"epoch": 2.3296,
"grad_norm": 0.562788336543834,
"learning_rate": 2.5736064654221808e-05,
"loss": 0.5174,
"step": 182
},
{
"epoch": 2.3424,
"grad_norm": 0.4545310242783742,
"learning_rate": 2.5564349278329056e-05,
"loss": 0.5046,
"step": 183
},
{
"epoch": 2.3552,
"grad_norm": 0.3875509270098935,
"learning_rate": 2.539218814664288e-05,
"loss": 0.5145,
"step": 184
},
{
"epoch": 2.368,
"grad_norm": 0.46461018413371363,
"learning_rate": 2.521959505086075e-05,
"loss": 0.4889,
"step": 185
},
{
"epoch": 2.3808,
"grad_norm": 0.3464919112988565,
"learning_rate": 2.5046583817284437e-05,
"loss": 0.5229,
"step": 186
},
{
"epoch": 2.3936,
"grad_norm": 0.4739230855888146,
"learning_rate": 2.487316830571244e-05,
"loss": 0.5045,
"step": 187
},
{
"epoch": 2.4064,
"grad_norm": 0.4527569882340709,
"learning_rate": 2.4699362408329646e-05,
"loss": 0.5007,
"step": 188
},
{
"epoch": 2.4192,
"grad_norm": 0.43278250897184295,
"learning_rate": 2.4525180048594452e-05,
"loss": 0.4951,
"step": 189
},
{
"epoch": 2.432,
"grad_norm": 0.41300510138939983,
"learning_rate": 2.435063518012335e-05,
"loss": 0.5203,
"step": 190
},
{
"epoch": 2.4448,
"grad_norm": 0.37260966158544223,
"learning_rate": 2.4175741785573177e-05,
"loss": 0.5117,
"step": 191
},
{
"epoch": 2.4576000000000002,
"grad_norm": 0.36382874275957855,
"learning_rate": 2.4000513875520892e-05,
"loss": 0.503,
"step": 192
},
{
"epoch": 2.4704,
"grad_norm": 0.3986579409856395,
"learning_rate": 2.3824965487341247e-05,
"loss": 0.5193,
"step": 193
},
{
"epoch": 2.4832,
"grad_norm": 0.34853392834188407,
"learning_rate": 2.3649110684082258e-05,
"loss": 0.5247,
"step": 194
},
{
"epoch": 2.496,
"grad_norm": 0.342609875000865,
"learning_rate": 2.3472963553338614e-05,
"loss": 0.5066,
"step": 195
},
{
"epoch": 2.5088,
"grad_norm": 0.36160657632661647,
"learning_rate": 2.3296538206123134e-05,
"loss": 0.5154,
"step": 196
},
{
"epoch": 2.5216,
"grad_norm": 0.3463567044676736,
"learning_rate": 2.311984877573636e-05,
"loss": 0.5199,
"step": 197
},
{
"epoch": 2.5343999999999998,
"grad_norm": 0.4058131993726706,
"learning_rate": 2.2942909416634326e-05,
"loss": 0.5186,
"step": 198
},
{
"epoch": 2.5472,
"grad_norm": 0.4057321265113332,
"learning_rate": 2.2765734303294666e-05,
"loss": 0.5143,
"step": 199
},
{
"epoch": 2.56,
"grad_norm": 0.3936214661542984,
"learning_rate": 2.2588337629081107e-05,
"loss": 0.5132,
"step": 200
},
{
"epoch": 2.5728,
"grad_norm": 0.370528958570038,
"learning_rate": 2.2410733605106462e-05,
"loss": 0.5244,
"step": 201
},
{
"epoch": 2.5856,
"grad_norm": 0.4214268915896029,
"learning_rate": 2.2232936459094158e-05,
"loss": 0.4999,
"step": 202
},
{
"epoch": 2.5984,
"grad_norm": 0.34631700186838116,
"learning_rate": 2.205496043423849e-05,
"loss": 0.5164,
"step": 203
},
{
"epoch": 2.6112,
"grad_norm": 0.40937456281219886,
"learning_rate": 2.1876819788063586e-05,
"loss": 0.5052,
"step": 204
},
{
"epoch": 2.624,
"grad_norm": 0.4574105884530299,
"learning_rate": 2.16985287912813e-05,
"loss": 0.5121,
"step": 205
},
{
"epoch": 2.6368,
"grad_norm": 0.3923599307357846,
"learning_rate": 2.1520101726647922e-05,
"loss": 0.5216,
"step": 206
},
{
"epoch": 2.6496,
"grad_norm": 0.46225848050812934,
"learning_rate": 2.1341552887820048e-05,
"loss": 0.52,
"step": 207
},
{
"epoch": 2.6624,
"grad_norm": 0.3813625635848944,
"learning_rate": 2.1162896578209517e-05,
"loss": 0.5233,
"step": 208
},
{
"epoch": 2.6752000000000002,
"grad_norm": 0.36829375003858267,
"learning_rate": 2.0984147109837564e-05,
"loss": 0.5107,
"step": 209
},
{
"epoch": 2.6879999999999997,
"grad_norm": 0.40407451080108014,
"learning_rate": 2.0805318802188307e-05,
"loss": 0.5211,
"step": 210
},
{
"epoch": 2.7008,
"grad_norm": 0.330097221116534,
"learning_rate": 2.0626425981061608e-05,
"loss": 0.5184,
"step": 211
},
{
"epoch": 2.7136,
"grad_norm": 0.4569268466278724,
"learning_rate": 2.0447482977425465e-05,
"loss": 0.4971,
"step": 212
},
{
"epoch": 2.7264,
"grad_norm": 0.32210414711988283,
"learning_rate": 2.0268504126267952e-05,
"loss": 0.5034,
"step": 213
},
{
"epoch": 2.7392,
"grad_norm": 0.4593913215028703,
"learning_rate": 2.008950376544887e-05,
"loss": 0.5055,
"step": 214
},
{
"epoch": 2.752,
"grad_norm": 0.32532356865740414,
"learning_rate": 1.9910496234551132e-05,
"loss": 0.5129,
"step": 215
},
{
"epoch": 2.7648,
"grad_norm": 0.45486772372147316,
"learning_rate": 1.9731495873732055e-05,
"loss": 0.5142,
"step": 216
},
{
"epoch": 2.7776,
"grad_norm": 0.3298842662400382,
"learning_rate": 1.9552517022574542e-05,
"loss": 0.5199,
"step": 217
},
{
"epoch": 2.7904,
"grad_norm": 0.4043174303350555,
"learning_rate": 1.93735740189384e-05,
"loss": 0.5118,
"step": 218
},
{
"epoch": 2.8032,
"grad_norm": 0.31749811135674905,
"learning_rate": 1.9194681197811703e-05,
"loss": 0.5202,
"step": 219
},
{
"epoch": 2.816,
"grad_norm": 0.412159163435511,
"learning_rate": 1.901585289016244e-05,
"loss": 0.5204,
"step": 220
},
{
"epoch": 2.8288,
"grad_norm": 0.3107510340812756,
"learning_rate": 1.8837103421790486e-05,
"loss": 0.5242,
"step": 221
},
{
"epoch": 2.8416,
"grad_norm": 0.390287835259627,
"learning_rate": 1.8658447112179952e-05,
"loss": 0.5121,
"step": 222
},
{
"epoch": 2.8544,
"grad_norm": 0.3287344109428031,
"learning_rate": 1.8479898273352084e-05,
"loss": 0.5097,
"step": 223
},
{
"epoch": 2.8672,
"grad_norm": 0.3374507455352951,
"learning_rate": 1.83014712087187e-05,
"loss": 0.5095,
"step": 224
},
{
"epoch": 2.88,
"grad_norm": 0.33332837158025963,
"learning_rate": 1.8123180211936417e-05,
"loss": 0.5051,
"step": 225
},
{
"epoch": 2.8928000000000003,
"grad_norm": 0.3117179339183091,
"learning_rate": 1.794503956576152e-05,
"loss": 0.5182,
"step": 226
},
{
"epoch": 2.9055999999999997,
"grad_norm": 0.3515290686368801,
"learning_rate": 1.776706354090585e-05,
"loss": 0.5086,
"step": 227
},
{
"epoch": 2.9184,
"grad_norm": 0.29412008878166757,
"learning_rate": 1.758926639489354e-05,
"loss": 0.524,
"step": 228
},
{
"epoch": 2.9312,
"grad_norm": 0.3298778138199897,
"learning_rate": 1.7411662370918893e-05,
"loss": 0.5144,
"step": 229
},
{
"epoch": 2.944,
"grad_norm": 0.2821012239713584,
"learning_rate": 1.7234265696705344e-05,
"loss": 0.5072,
"step": 230
},
{
"epoch": 2.9568,
"grad_norm": 0.3012660720135622,
"learning_rate": 1.7057090583365678e-05,
"loss": 0.5145,
"step": 231
},
{
"epoch": 2.9696,
"grad_norm": 0.28948238104965945,
"learning_rate": 1.6880151224263646e-05,
"loss": 0.5029,
"step": 232
},
{
"epoch": 2.9824,
"grad_norm": 0.32162146380244094,
"learning_rate": 1.6703461793876876e-05,
"loss": 0.5167,
"step": 233
},
{
"epoch": 2.9952,
"grad_norm": 0.3034401446317415,
"learning_rate": 1.6527036446661396e-05,
"loss": 0.5004,
"step": 234
},
{
"epoch": 3.008,
"grad_norm": 0.41046645936186843,
"learning_rate": 1.635088931591775e-05,
"loss": 0.463,
"step": 235
},
{
"epoch": 3.0208,
"grad_norm": 0.39836569087962515,
"learning_rate": 1.6175034512658753e-05,
"loss": 0.4378,
"step": 236
},
{
"epoch": 3.0336,
"grad_norm": 0.6901150993329158,
"learning_rate": 1.5999486124479115e-05,
"loss": 0.4371,
"step": 237
},
{
"epoch": 3.0464,
"grad_norm": 0.41203511413224175,
"learning_rate": 1.5824258214426833e-05,
"loss": 0.4426,
"step": 238
},
{
"epoch": 3.0592,
"grad_norm": 0.5062977186349449,
"learning_rate": 1.5649364819876655e-05,
"loss": 0.4351,
"step": 239
},
{
"epoch": 3.072,
"grad_norm": 0.3979325140755632,
"learning_rate": 1.547481995140556e-05,
"loss": 0.4314,
"step": 240
},
{
"epoch": 3.0848,
"grad_norm": 0.4498545218503761,
"learning_rate": 1.5300637591670357e-05,
"loss": 0.4401,
"step": 241
},
{
"epoch": 3.0976,
"grad_norm": 0.42355565040806803,
"learning_rate": 1.5126831694287564e-05,
"loss": 0.4337,
"step": 242
},
{
"epoch": 3.1104,
"grad_norm": 0.44462201972834053,
"learning_rate": 1.4953416182715566e-05,
"loss": 0.4282,
"step": 243
},
{
"epoch": 3.1232,
"grad_norm": 0.40849174738433486,
"learning_rate": 1.478040494913926e-05,
"loss": 0.4537,
"step": 244
},
{
"epoch": 3.136,
"grad_norm": 0.45302408598349836,
"learning_rate": 1.460781185335713e-05,
"loss": 0.4488,
"step": 245
},
{
"epoch": 3.1488,
"grad_norm": 0.38161356592555723,
"learning_rate": 1.443565072167095e-05,
"loss": 0.4334,
"step": 246
},
{
"epoch": 3.1616,
"grad_norm": 0.4477548848994946,
"learning_rate": 1.4263935345778202e-05,
"loss": 0.4443,
"step": 247
},
{
"epoch": 3.1744,
"grad_norm": 0.4083281344276796,
"learning_rate": 1.409267948166718e-05,
"loss": 0.4315,
"step": 248
},
{
"epoch": 3.1872,
"grad_norm": 0.3176643943634673,
"learning_rate": 1.3921896848515064e-05,
"loss": 0.4449,
"step": 249
},
{
"epoch": 3.2,
"grad_norm": 0.40201283506559043,
"learning_rate": 1.3751601127588849e-05,
"loss": 0.441,
"step": 250
},
{
"epoch": 3.2128,
"grad_norm": 0.3383208102497633,
"learning_rate": 1.3581805961149371e-05,
"loss": 0.436,
"step": 251
},
{
"epoch": 3.2256,
"grad_norm": 0.3110589456051368,
"learning_rate": 1.341252495135841e-05,
"loss": 0.4353,
"step": 252
},
{
"epoch": 3.2384,
"grad_norm": 0.3272197166941022,
"learning_rate": 1.324377165918906e-05,
"loss": 0.4343,
"step": 253
},
{
"epoch": 3.2512,
"grad_norm": 0.31641831793934144,
"learning_rate": 1.3075559603339354e-05,
"loss": 0.4315,
"step": 254
},
{
"epoch": 3.2640000000000002,
"grad_norm": 0.32780471432228836,
"learning_rate": 1.2907902259149287e-05,
"loss": 0.4356,
"step": 255
},
{
"epoch": 3.2768,
"grad_norm": 0.32626163102304695,
"learning_rate": 1.274081305752135e-05,
"loss": 0.4401,
"step": 256
},
{
"epoch": 3.2896,
"grad_norm": 0.3169286387534973,
"learning_rate": 1.2574305383844528e-05,
"loss": 0.4434,
"step": 257
},
{
"epoch": 3.3024,
"grad_norm": 0.32721144517777084,
"learning_rate": 1.2408392576922075e-05,
"loss": 0.4414,
"step": 258
},
{
"epoch": 3.3152,
"grad_norm": 0.27217470559211066,
"learning_rate": 1.2243087927902905e-05,
"loss": 0.4475,
"step": 259
},
{
"epoch": 3.328,
"grad_norm": 0.3359690140623319,
"learning_rate": 1.2078404679216864e-05,
"loss": 0.4246,
"step": 260
},
{
"epoch": 3.3407999999999998,
"grad_norm": 0.26838112419162513,
"learning_rate": 1.1914356023513904e-05,
"loss": 0.4303,
"step": 261
},
{
"epoch": 3.3536,
"grad_norm": 0.29178007189259314,
"learning_rate": 1.1750955102607193e-05,
"loss": 0.4366,
"step": 262
},
{
"epoch": 3.3664,
"grad_norm": 0.2791059524097632,
"learning_rate": 1.1588215006420374e-05,
"loss": 0.4396,
"step": 263
},
{
"epoch": 3.3792,
"grad_norm": 0.27798085486262086,
"learning_rate": 1.1426148771938915e-05,
"loss": 0.4372,
"step": 264
},
{
"epoch": 3.392,
"grad_norm": 0.29048854928627454,
"learning_rate": 1.1264769382165748e-05,
"loss": 0.4366,
"step": 265
},
{
"epoch": 3.4048,
"grad_norm": 0.3096664519908522,
"learning_rate": 1.110408976508118e-05,
"loss": 0.4325,
"step": 266
},
{
"epoch": 3.4176,
"grad_norm": 0.3053421610217013,
"learning_rate": 1.094412279260726e-05,
"loss": 0.4227,
"step": 267
},
{
"epoch": 3.4304,
"grad_norm": 0.2703969052430842,
"learning_rate": 1.0784881279576635e-05,
"loss": 0.4413,
"step": 268
},
{
"epoch": 3.4432,
"grad_norm": 0.31743100870335433,
"learning_rate": 1.0626377982705929e-05,
"loss": 0.4485,
"step": 269
},
{
"epoch": 3.456,
"grad_norm": 0.2951598049339725,
"learning_rate": 1.0468625599573842e-05,
"loss": 0.4223,
"step": 270
},
{
"epoch": 3.4688,
"grad_norm": 0.3014289743968507,
"learning_rate": 1.0311636767603952e-05,
"loss": 0.441,
"step": 271
},
{
"epoch": 3.4816,
"grad_norm": 0.2872576850605509,
"learning_rate": 1.0155424063052306e-05,
"loss": 0.4388,
"step": 272
},
{
"epoch": 3.4944,
"grad_norm": 0.3003555194020943,
"learning_rate": 1.0000000000000006e-05,
"loss": 0.4383,
"step": 273
},
{
"epoch": 3.5072,
"grad_norm": 0.27188339642329284,
"learning_rate": 9.84537702935065e-06,
"loss": 0.4391,
"step": 274
},
{
"epoch": 3.52,
"grad_norm": 0.3133892904576033,
"learning_rate": 9.691567537832964e-06,
"loss": 0.4489,
"step": 275
},
{
"epoch": 3.5328,
"grad_norm": 0.27326154899036914,
"learning_rate": 9.538583847008452e-06,
"loss": 0.4188,
"step": 276
},
{
"epoch": 3.5456,
"grad_norm": 0.2659259632741142,
"learning_rate": 9.386438212284372e-06,
"loss": 0.423,
"step": 277
},
{
"epoch": 3.5584,
"grad_norm": 0.28898963859236165,
"learning_rate": 9.235142821931928e-06,
"loss": 0.431,
"step": 278
},
{
"epoch": 3.5712,
"grad_norm": 0.27216817344052996,
"learning_rate": 9.084709796109907e-06,
"loss": 0.4381,
"step": 279
},
{
"epoch": 3.584,
"grad_norm": 0.2762213585544952,
"learning_rate": 8.93515118589373e-06,
"loss": 0.4425,
"step": 280
},
{
"epoch": 3.5968,
"grad_norm": 0.3098087436899742,
"learning_rate": 8.786478972310023e-06,
"loss": 0.4301,
"step": 281
},
{
"epoch": 3.6096,
"grad_norm": 0.2772833601212417,
"learning_rate": 8.638705065376887e-06,
"loss": 0.4289,
"step": 282
},
{
"epoch": 3.6224,
"grad_norm": 0.28264765343950765,
"learning_rate": 8.491841303149728e-06,
"loss": 0.4311,
"step": 283
},
{
"epoch": 3.6352,
"grad_norm": 0.29573961882101746,
"learning_rate": 8.345899450772975e-06,
"loss": 0.4371,
"step": 284
},
{
"epoch": 3.648,
"grad_norm": 0.2507671856832434,
"learning_rate": 8.200891199537549e-06,
"loss": 0.4434,
"step": 285
},
{
"epoch": 3.6608,
"grad_norm": 0.25755157366812154,
"learning_rate": 8.056828165944282e-06,
"loss": 0.4314,
"step": 286
},
{
"epoch": 3.6736,
"grad_norm": 0.26829055240158844,
"learning_rate": 7.913721890773354e-06,
"loss": 0.4393,
"step": 287
},
{
"epoch": 3.6864,
"grad_norm": 0.24653841182739372,
"learning_rate": 7.771583838159756e-06,
"loss": 0.4335,
"step": 288
},
{
"epoch": 3.6992000000000003,
"grad_norm": 0.27273851728987,
"learning_rate": 7.630425394674903e-06,
"loss": 0.4346,
"step": 289
},
{
"epoch": 3.7119999999999997,
"grad_norm": 0.2542223684948622,
"learning_rate": 7.49025786841445e-06,
"loss": 0.4203,
"step": 290
},
{
"epoch": 3.7248,
"grad_norm": 0.26861560889688035,
"learning_rate": 7.3510924880924575e-06,
"loss": 0.4461,
"step": 291
},
{
"epoch": 3.7376,
"grad_norm": 0.2534116220118216,
"learning_rate": 7.212940402141808e-06,
"loss": 0.4325,
"step": 292
},
{
"epoch": 3.7504,
"grad_norm": 0.27795799974837215,
"learning_rate": 7.075812677821145e-06,
"loss": 0.4248,
"step": 293
},
{
"epoch": 3.7632,
"grad_norm": 0.2613087055563045,
"learning_rate": 6.939720300328303e-06,
"loss": 0.455,
"step": 294
},
{
"epoch": 3.776,
"grad_norm": 0.27299344851818386,
"learning_rate": 6.8046741719202385e-06,
"loss": 0.4485,
"step": 295
},
{
"epoch": 3.7888,
"grad_norm": 0.2781875846770507,
"learning_rate": 6.67068511103971e-06,
"loss": 0.4514,
"step": 296
},
{
"epoch": 3.8016,
"grad_norm": 0.2584225371356067,
"learning_rate": 6.537763851448593e-06,
"loss": 0.4467,
"step": 297
},
{
"epoch": 3.8144,
"grad_norm": 0.3085510141589761,
"learning_rate": 6.4059210413680175e-06,
"loss": 0.4343,
"step": 298
},
{
"epoch": 3.8272,
"grad_norm": 0.2715244321951438,
"learning_rate": 6.275167242625331e-06,
"loss": 0.4271,
"step": 299
},
{
"epoch": 3.84,
"grad_norm": 0.2738849657890685,
"learning_rate": 6.145512929808013e-06,
"loss": 0.4414,
"step": 300
},
{
"epoch": 3.8528000000000002,
"grad_norm": 0.2509332337895888,
"learning_rate": 6.016968489424572e-06,
"loss": 0.4283,
"step": 301
},
{
"epoch": 3.8656,
"grad_norm": 0.2585900899489169,
"learning_rate": 5.889544219072465e-06,
"loss": 0.4249,
"step": 302
},
{
"epoch": 3.8784,
"grad_norm": 0.2504734736778933,
"learning_rate": 5.7632503266131925e-06,
"loss": 0.4274,
"step": 303
},
{
"epoch": 3.8912,
"grad_norm": 0.2343119676208743,
"learning_rate": 5.638096929354522e-06,
"loss": 0.4265,
"step": 304
},
{
"epoch": 3.904,
"grad_norm": 0.23914782192599746,
"learning_rate": 5.514094053240035e-06,
"loss": 0.4274,
"step": 305
},
{
"epoch": 3.9168,
"grad_norm": 0.24779763263825993,
"learning_rate": 5.39125163204594e-06,
"loss": 0.4366,
"step": 306
},
{
"epoch": 3.9295999999999998,
"grad_norm": 0.2387628080118382,
"learning_rate": 5.269579506585259e-06,
"loss": 0.4445,
"step": 307
},
{
"epoch": 3.9424,
"grad_norm": 0.23695884317251484,
"learning_rate": 5.149087423919541e-06,
"loss": 0.4383,
"step": 308
},
{
"epoch": 3.9552,
"grad_norm": 0.23140809524176137,
"learning_rate": 5.029785036577976e-06,
"loss": 0.4311,
"step": 309
},
{
"epoch": 3.968,
"grad_norm": 0.2260624177391602,
"learning_rate": 4.911681901784198e-06,
"loss": 0.4481,
"step": 310
},
{
"epoch": 3.9808,
"grad_norm": 0.2452242981234645,
"learning_rate": 4.794787480690597e-06,
"loss": 0.4391,
"step": 311
},
{
"epoch": 3.9936,
"grad_norm": 0.23650964042923645,
"learning_rate": 4.679111137620442e-06,
"loss": 0.4449,
"step": 312
},
{
"epoch": 4.0064,
"grad_norm": 0.29962007310679284,
"learning_rate": 4.5646621393177e-06,
"loss": 0.4054,
"step": 313
},
{
"epoch": 4.0192,
"grad_norm": 0.3312469740936787,
"learning_rate": 4.451449654204685e-06,
"loss": 0.409,
"step": 314
},
{
"epoch": 4.032,
"grad_norm": 0.27286209439834047,
"learning_rate": 4.339482751647557e-06,
"loss": 0.378,
"step": 315
},
{
"epoch": 4.0448,
"grad_norm": 0.3231896957399875,
"learning_rate": 4.228770401229824e-06,
"loss": 0.4044,
"step": 316
},
{
"epoch": 4.0576,
"grad_norm": 0.4227960209011115,
"learning_rate": 4.119321472033779e-06,
"loss": 0.3691,
"step": 317
},
{
"epoch": 4.0704,
"grad_norm": 0.3590703865605139,
"learning_rate": 4.011144731929981e-06,
"loss": 0.3743,
"step": 318
},
{
"epoch": 4.0832,
"grad_norm": 0.3276315585715536,
"learning_rate": 3.904248846874894e-06,
"loss": 0.4057,
"step": 319
},
{
"epoch": 4.096,
"grad_norm": 0.3204463262223722,
"learning_rate": 3.7986423802166705e-06,
"loss": 0.3841,
"step": 320
},
{
"epoch": 4.1088,
"grad_norm": 0.3572447564220915,
"learning_rate": 3.694333792009115e-06,
"loss": 0.391,
"step": 321
},
{
"epoch": 4.1216,
"grad_norm": 0.30830859475235545,
"learning_rate": 3.5913314383339937e-06,
"loss": 0.391,
"step": 322
},
{
"epoch": 4.1344,
"grad_norm": 0.29639654723525105,
"learning_rate": 3.4896435706316e-06,
"loss": 0.3888,
"step": 323
},
{
"epoch": 4.1472,
"grad_norm": 0.2866219313421289,
"learning_rate": 3.3892783350397675e-06,
"loss": 0.3897,
"step": 324
},
{
"epoch": 4.16,
"grad_norm": 0.2859626125074166,
"learning_rate": 3.290243771741275e-06,
"loss": 0.3844,
"step": 325
},
{
"epoch": 4.1728,
"grad_norm": 0.29987145812710686,
"learning_rate": 3.1925478143197418e-06,
"loss": 0.38,
"step": 326
},
{
"epoch": 4.1856,
"grad_norm": 0.27714350402080357,
"learning_rate": 3.0961982891241083e-06,
"loss": 0.3915,
"step": 327
},
{
"epoch": 4.1984,
"grad_norm": 0.2816921404300732,
"learning_rate": 3.001202914641628e-06,
"loss": 0.3827,
"step": 328
},
{
"epoch": 4.2112,
"grad_norm": 0.2624104774293918,
"learning_rate": 2.907569300879596e-06,
"loss": 0.3802,
"step": 329
},
{
"epoch": 4.224,
"grad_norm": 0.26216341440755286,
"learning_rate": 2.815304948755664e-06,
"loss": 0.3811,
"step": 330
},
{
"epoch": 4.2368,
"grad_norm": 0.2597148815879336,
"learning_rate": 2.7244172494969978e-06,
"loss": 0.3941,
"step": 331
},
{
"epoch": 4.2496,
"grad_norm": 0.2589303491965205,
"learning_rate": 2.6349134840481294e-06,
"loss": 0.3836,
"step": 332
},
{
"epoch": 4.2624,
"grad_norm": 0.25039699970590135,
"learning_rate": 2.546800822487714e-06,
"loss": 0.3956,
"step": 333
},
{
"epoch": 4.2752,
"grad_norm": 0.22867459998419756,
"learning_rate": 2.4600863234541338e-06,
"loss": 0.3984,
"step": 334
},
{
"epoch": 4.288,
"grad_norm": 0.22813975032869171,
"learning_rate": 2.374776933580025e-06,
"loss": 0.3945,
"step": 335
},
{
"epoch": 4.3008,
"grad_norm": 0.24726113875422293,
"learning_rate": 2.2908794869358044e-06,
"loss": 0.3804,
"step": 336
},
{
"epoch": 4.3136,
"grad_norm": 0.23839697981679817,
"learning_rate": 2.2084007044821764e-06,
"loss": 0.3918,
"step": 337
},
{
"epoch": 4.3264,
"grad_norm": 0.24105041292682955,
"learning_rate": 2.127347193531757e-06,
"loss": 0.3926,
"step": 338
},
{
"epoch": 4.3392,
"grad_norm": 0.22817096036617446,
"learning_rate": 2.0477254472197237e-06,
"loss": 0.3864,
"step": 339
},
{
"epoch": 4.352,
"grad_norm": 0.22470269288312994,
"learning_rate": 1.96954184398368e-06,
"loss": 0.3812,
"step": 340
},
{
"epoch": 4.3648,
"grad_norm": 0.22124074384761025,
"learning_rate": 1.8928026470526917e-06,
"loss": 0.3853,
"step": 341
},
{
"epoch": 4.3776,
"grad_norm": 0.22290829954818844,
"learning_rate": 1.817514003945524e-06,
"loss": 0.3892,
"step": 342
},
{
"epoch": 4.3904,
"grad_norm": 0.24016004520116097,
"learning_rate": 1.743681945978184e-06,
"loss": 0.3746,
"step": 343
},
{
"epoch": 4.4032,
"grad_norm": 0.21396083544636266,
"learning_rate": 1.6713123877807413e-06,
"loss": 0.3752,
"step": 344
},
{
"epoch": 4.416,
"grad_norm": 0.21462260986745693,
"learning_rate": 1.6004111268235156e-06,
"loss": 0.3938,
"step": 345
},
{
"epoch": 4.4288,
"grad_norm": 0.21840179140888716,
"learning_rate": 1.5309838429526714e-06,
"loss": 0.381,
"step": 346
},
{
"epoch": 4.4416,
"grad_norm": 0.22612770072690125,
"learning_rate": 1.4630360979351644e-06,
"loss": 0.3767,
"step": 347
},
{
"epoch": 4.4544,
"grad_norm": 0.22616358980175943,
"learning_rate": 1.396573335013236e-06,
"loss": 0.3949,
"step": 348
},
{
"epoch": 4.4672,
"grad_norm": 0.21834471671265632,
"learning_rate": 1.3316008784683265e-06,
"loss": 0.3952,
"step": 349
},
{
"epoch": 4.48,
"grad_norm": 0.22384371553459914,
"learning_rate": 1.2681239331945695e-06,
"loss": 0.4051,
"step": 350
},
{
"epoch": 4.4928,
"grad_norm": 0.2145005953417142,
"learning_rate": 1.2061475842818337e-06,
"loss": 0.3991,
"step": 351
},
{
"epoch": 4.5056,
"grad_norm": 0.2159902095624403,
"learning_rate": 1.1456767966083393e-06,
"loss": 0.3793,
"step": 352
},
{
"epoch": 4.5184,
"grad_norm": 0.21070490480560447,
"learning_rate": 1.086716414442952e-06,
"loss": 0.3937,
"step": 353
},
{
"epoch": 4.5312,
"grad_norm": 0.220079307692125,
"learning_rate": 1.0292711610570904e-06,
"loss": 0.391,
"step": 354
},
{
"epoch": 4.5440000000000005,
"grad_norm": 0.20512973244246804,
"learning_rate": 9.733456383463658e-07,
"loss": 0.3967,
"step": 355
},
{
"epoch": 4.5568,
"grad_norm": 0.21311854141756914,
"learning_rate": 9.189443264619102e-07,
"loss": 0.4129,
"step": 356
},
{
"epoch": 4.5696,
"grad_norm": 0.21805552103822698,
"learning_rate": 8.660715834514977e-07,
"loss": 0.3916,
"step": 357
},
{
"epoch": 4.5824,
"grad_norm": 0.2080489832732226,
"learning_rate": 8.147316449103959e-07,
"loss": 0.4004,
"step": 358
},
{
"epoch": 4.5952,
"grad_norm": 0.20739889265925496,
"learning_rate": 7.649286236420806e-07,
"loss": 0.4153,
"step": 359
},
{
"epoch": 4.608,
"grad_norm": 0.21067276573842075,
"learning_rate": 7.166665093287539e-07,
"loss": 0.4006,
"step": 360
},
{
"epoch": 4.6208,
"grad_norm": 0.22299398200312037,
"learning_rate": 6.69949168211721e-07,
"loss": 0.3749,
"step": 361
},
{
"epoch": 4.6336,
"grad_norm": 0.2037159019231344,
"learning_rate": 6.247803427816945e-07,
"loss": 0.4018,
"step": 362
},
{
"epoch": 4.6464,
"grad_norm": 0.21263928604401106,
"learning_rate": 5.811636514789598e-07,
"loss": 0.3799,
"step": 363
},
{
"epoch": 4.6592,
"grad_norm": 0.2148226409056421,
"learning_rate": 5.391025884035239e-07,
"loss": 0.3806,
"step": 364
},
{
"epoch": 4.672,
"grad_norm": 0.20705283380158873,
"learning_rate": 4.986005230351954e-07,
"loss": 0.3777,
"step": 365
},
{
"epoch": 4.6848,
"grad_norm": 0.21576036694240572,
"learning_rate": 4.5966069996365993e-07,
"loss": 0.3986,
"step": 366
},
{
"epoch": 4.6975999999999996,
"grad_norm": 0.21027087952143309,
"learning_rate": 4.22286238628562e-07,
"loss": 0.3794,
"step": 367
},
{
"epoch": 4.7104,
"grad_norm": 0.21565973905897212,
"learning_rate": 3.8648013306960664e-07,
"loss": 0.3808,
"step": 368
},
{
"epoch": 4.7232,
"grad_norm": 0.2027994309063992,
"learning_rate": 3.522452516867048e-07,
"loss": 0.3912,
"step": 369
},
{
"epoch": 4.736,
"grad_norm": 0.20333862712713335,
"learning_rate": 3.1958433701019697e-07,
"loss": 0.3711,
"step": 370
},
{
"epoch": 4.7488,
"grad_norm": 0.20786917102005878,
"learning_rate": 2.8850000548115155e-07,
"loss": 0.3759,
"step": 371
},
{
"epoch": 4.7616,
"grad_norm": 0.21090283998936593,
"learning_rate": 2.5899474724174313e-07,
"loss": 0.3818,
"step": 372
},
{
"epoch": 4.7744,
"grad_norm": 0.22006312730244917,
"learning_rate": 2.3107092593579905e-07,
"loss": 0.3909,
"step": 373
},
{
"epoch": 4.7872,
"grad_norm": 0.20748609787041353,
"learning_rate": 2.0473077851942858e-07,
"loss": 0.3876,
"step": 374
},
{
"epoch": 4.8,
"grad_norm": 0.21091568244756467,
"learning_rate": 1.799764150818306e-07,
"loss": 0.3926,
"step": 375
},
{
"epoch": 4.8128,
"grad_norm": 0.20541127946636106,
"learning_rate": 1.5680981867625566e-07,
"loss": 0.3902,
"step": 376
},
{
"epoch": 4.8256,
"grad_norm": 0.20811933533876514,
"learning_rate": 1.3523284516113955e-07,
"loss": 0.3818,
"step": 377
},
{
"epoch": 4.8384,
"grad_norm": 0.20569700919140488,
"learning_rate": 1.1524722305144231e-07,
"loss": 0.3852,
"step": 378
},
{
"epoch": 4.8512,
"grad_norm": 0.2069276613444958,
"learning_rate": 9.685455338016347e-08,
"loss": 0.3988,
"step": 379
},
{
"epoch": 4.864,
"grad_norm": 0.20911198227882938,
"learning_rate": 8.005630957010014e-08,
"loss": 0.3851,
"step": 380
},
{
"epoch": 4.8768,
"grad_norm": 0.20532543240595294,
"learning_rate": 6.485383731580142e-08,
"loss": 0.3914,
"step": 381
},
{
"epoch": 4.8896,
"grad_norm": 0.20732465469266587,
"learning_rate": 5.1248354475768034e-08,
"loss": 0.3845,
"step": 382
},
{
"epoch": 4.9024,
"grad_norm": 0.21466224187863875,
"learning_rate": 3.924095097489922e-08,
"loss": 0.3775,
"step": 383
},
{
"epoch": 4.9152000000000005,
"grad_norm": 0.20921729995840035,
"learning_rate": 2.8832588717164766e-08,
"loss": 0.3839,
"step": 384
},
{
"epoch": 4.928,
"grad_norm": 0.20764848352691964,
"learning_rate": 2.0024101508555604e-08,
"loss": 0.3927,
"step": 385
},
{
"epoch": 4.9408,
"grad_norm": 0.2119491015744724,
"learning_rate": 1.281619499029274e-08,
"loss": 0.3819,
"step": 386
},
{
"epoch": 4.9536,
"grad_norm": 0.21028962864646566,
"learning_rate": 7.209446582292501e-09,
"loss": 0.3863,
"step": 387
},
{
"epoch": 4.9664,
"grad_norm": 0.20568189636278641,
"learning_rate": 3.2043054369057523e-09,
"loss": 0.4027,
"step": 388
},
{
"epoch": 4.9792,
"grad_norm": 0.20207762958979356,
"learning_rate": 8.010924029533406e-10,
"loss": 0.3858,
"step": 389
},
{
"epoch": 4.992,
"grad_norm": 0.20602286936096575,
"learning_rate": 0.0,
"loss": 0.3772,
"step": 390
},
{
"epoch": 4.992,
"step": 390,
"total_flos": 1.896730808245289e+18,
"train_loss": 0.5385960905979841,
"train_runtime": 18314.8306,
"train_samples_per_second": 2.73,
"train_steps_per_second": 0.021
}
],
"logging_steps": 1,
"max_steps": 390,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.896730808245289e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}