ShapeLLM_7B_gapartnet_v1.0 / trainer_state.json
alexhafnerd's picture
upload checkpoints
9bb948a
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 242,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2.5e-06,
"loss": 2.2266,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 5e-06,
"loss": 2.2949,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 7.500000000000001e-06,
"loss": 2.1211,
"step": 3
},
{
"epoch": 0.02,
"learning_rate": 1e-05,
"loss": 1.748,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 1.25e-05,
"loss": 1.4082,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 1.5000000000000002e-05,
"loss": 1.3105,
"step": 6
},
{
"epoch": 0.03,
"learning_rate": 1.7500000000000002e-05,
"loss": 1.1182,
"step": 7
},
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 1.1958,
"step": 8
},
{
"epoch": 0.04,
"learning_rate": 1.999909877856721e-05,
"loss": 1.0928,
"step": 9
},
{
"epoch": 0.04,
"learning_rate": 1.9996395276708856e-05,
"loss": 0.979,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 1.9991889981715696e-05,
"loss": 0.9541,
"step": 11
},
{
"epoch": 0.05,
"learning_rate": 1.9985583705641418e-05,
"loss": 0.9253,
"step": 12
},
{
"epoch": 0.05,
"learning_rate": 1.9977477585156252e-05,
"loss": 0.936,
"step": 13
},
{
"epoch": 0.06,
"learning_rate": 1.9967573081342103e-05,
"loss": 0.8491,
"step": 14
},
{
"epoch": 0.06,
"learning_rate": 1.9955871979429188e-05,
"loss": 0.8838,
"step": 15
},
{
"epoch": 0.07,
"learning_rate": 1.9942376388474282e-05,
"loss": 0.7446,
"step": 16
},
{
"epoch": 0.07,
"learning_rate": 1.992708874098054e-05,
"loss": 0.7017,
"step": 17
},
{
"epoch": 0.07,
"learning_rate": 1.9910011792459086e-05,
"loss": 0.8101,
"step": 18
},
{
"epoch": 0.08,
"learning_rate": 1.989114862093232e-05,
"loss": 0.7563,
"step": 19
},
{
"epoch": 0.08,
"learning_rate": 1.9870502626379127e-05,
"loss": 0.7939,
"step": 20
},
{
"epoch": 0.09,
"learning_rate": 1.9848077530122083e-05,
"loss": 0.7065,
"step": 21
},
{
"epoch": 0.09,
"learning_rate": 1.9823877374156647e-05,
"loss": 0.7466,
"step": 22
},
{
"epoch": 0.1,
"learning_rate": 1.979790652042268e-05,
"loss": 0.7607,
"step": 23
},
{
"epoch": 0.1,
"learning_rate": 1.977016965001817e-05,
"loss": 0.689,
"step": 24
},
{
"epoch": 0.1,
"learning_rate": 1.9740671762355548e-05,
"loss": 0.8286,
"step": 25
},
{
"epoch": 0.11,
"learning_rate": 1.9709418174260523e-05,
"loss": 0.6978,
"step": 26
},
{
"epoch": 0.11,
"learning_rate": 1.9676414519013782e-05,
"loss": 0.6284,
"step": 27
},
{
"epoch": 0.12,
"learning_rate": 1.9641666745335626e-05,
"loss": 0.7383,
"step": 28
},
{
"epoch": 0.12,
"learning_rate": 1.9605181116313725e-05,
"loss": 0.6973,
"step": 29
},
{
"epoch": 0.12,
"learning_rate": 1.9566964208274254e-05,
"loss": 0.7092,
"step": 30
},
{
"epoch": 0.13,
"learning_rate": 1.9527022909596537e-05,
"loss": 0.7104,
"step": 31
},
{
"epoch": 0.13,
"learning_rate": 1.9485364419471454e-05,
"loss": 0.6423,
"step": 32
},
{
"epoch": 0.14,
"learning_rate": 1.9441996246603848e-05,
"loss": 0.6541,
"step": 33
},
{
"epoch": 0.14,
"learning_rate": 1.9396926207859085e-05,
"loss": 0.6978,
"step": 34
},
{
"epoch": 0.14,
"learning_rate": 1.9350162426854152e-05,
"loss": 0.7002,
"step": 35
},
{
"epoch": 0.15,
"learning_rate": 1.9301713332493386e-05,
"loss": 0.6743,
"step": 36
},
{
"epoch": 0.15,
"learning_rate": 1.925158765744924e-05,
"loss": 0.6343,
"step": 37
},
{
"epoch": 0.16,
"learning_rate": 1.9199794436588244e-05,
"loss": 0.6274,
"step": 38
},
{
"epoch": 0.16,
"learning_rate": 1.9146343005342546e-05,
"loss": 0.7134,
"step": 39
},
{
"epoch": 0.17,
"learning_rate": 1.909124299802724e-05,
"loss": 0.646,
"step": 40
},
{
"epoch": 0.17,
"learning_rate": 1.9034504346103825e-05,
"loss": 0.6392,
"step": 41
},
{
"epoch": 0.17,
"learning_rate": 1.8976137276390145e-05,
"loss": 0.5862,
"step": 42
},
{
"epoch": 0.18,
"learning_rate": 1.891615230921703e-05,
"loss": 0.6609,
"step": 43
},
{
"epoch": 0.18,
"learning_rate": 1.8854560256532098e-05,
"loss": 0.6292,
"step": 44
},
{
"epoch": 0.19,
"learning_rate": 1.879137221995095e-05,
"loss": 0.5913,
"step": 45
},
{
"epoch": 0.19,
"learning_rate": 1.8726599588756144e-05,
"loss": 0.6235,
"step": 46
},
{
"epoch": 0.19,
"learning_rate": 1.866025403784439e-05,
"loss": 0.7061,
"step": 47
},
{
"epoch": 0.2,
"learning_rate": 1.859234752562217e-05,
"loss": 0.6748,
"step": 48
},
{
"epoch": 0.2,
"learning_rate": 1.8522892291850335e-05,
"loss": 0.5935,
"step": 49
},
{
"epoch": 0.21,
"learning_rate": 1.845190085543795e-05,
"loss": 0.687,
"step": 50
},
{
"epoch": 0.21,
"learning_rate": 1.8379386012185813e-05,
"loss": 0.6694,
"step": 51
},
{
"epoch": 0.21,
"learning_rate": 1.8305360832480118e-05,
"loss": 0.5725,
"step": 52
},
{
"epoch": 0.22,
"learning_rate": 1.8229838658936566e-05,
"loss": 0.6375,
"step": 53
},
{
"epoch": 0.22,
"learning_rate": 1.8152833103995443e-05,
"loss": 0.6128,
"step": 54
},
{
"epoch": 0.23,
"learning_rate": 1.807435804746807e-05,
"loss": 0.6118,
"step": 55
},
{
"epoch": 0.23,
"learning_rate": 1.7994427634035016e-05,
"loss": 0.6218,
"step": 56
},
{
"epoch": 0.24,
"learning_rate": 1.791305627069662e-05,
"loss": 0.6094,
"step": 57
},
{
"epoch": 0.24,
"learning_rate": 1.7830258624176224e-05,
"loss": 0.6504,
"step": 58
},
{
"epoch": 0.24,
"learning_rate": 1.7746049618276545e-05,
"loss": 0.5811,
"step": 59
},
{
"epoch": 0.25,
"learning_rate": 1.766044443118978e-05,
"loss": 0.5911,
"step": 60
},
{
"epoch": 0.25,
"learning_rate": 1.7573458492761802e-05,
"loss": 0.5942,
"step": 61
},
{
"epoch": 0.26,
"learning_rate": 1.7485107481711014e-05,
"loss": 0.6296,
"step": 62
},
{
"epoch": 0.26,
"learning_rate": 1.7395407322802374e-05,
"loss": 0.6377,
"step": 63
},
{
"epoch": 0.26,
"learning_rate": 1.7304374183977032e-05,
"loss": 0.5991,
"step": 64
},
{
"epoch": 0.27,
"learning_rate": 1.7212024473438145e-05,
"loss": 0.5847,
"step": 65
},
{
"epoch": 0.27,
"learning_rate": 1.7118374836693407e-05,
"loss": 0.5864,
"step": 66
},
{
"epoch": 0.28,
"learning_rate": 1.7023442153554776e-05,
"loss": 0.5608,
"step": 67
},
{
"epoch": 0.28,
"learning_rate": 1.6927243535095995e-05,
"loss": 0.603,
"step": 68
},
{
"epoch": 0.29,
"learning_rate": 1.6829796320568416e-05,
"loss": 0.5874,
"step": 69
},
{
"epoch": 0.29,
"learning_rate": 1.67311180742757e-05,
"loss": 0.5879,
"step": 70
},
{
"epoch": 0.29,
"learning_rate": 1.6631226582407954e-05,
"loss": 0.6262,
"step": 71
},
{
"epoch": 0.3,
"learning_rate": 1.653013984983585e-05,
"loss": 0.5554,
"step": 72
},
{
"epoch": 0.3,
"learning_rate": 1.6427876096865394e-05,
"loss": 0.5928,
"step": 73
},
{
"epoch": 0.31,
"learning_rate": 1.6324453755953772e-05,
"loss": 0.6152,
"step": 74
},
{
"epoch": 0.31,
"learning_rate": 1.621989146838704e-05,
"loss": 0.6365,
"step": 75
},
{
"epoch": 0.31,
"learning_rate": 1.6114208080920125e-05,
"loss": 0.6069,
"step": 76
},
{
"epoch": 0.32,
"learning_rate": 1.600742264237979e-05,
"loss": 0.5688,
"step": 77
},
{
"epoch": 0.32,
"learning_rate": 1.5899554400231233e-05,
"loss": 0.7437,
"step": 78
},
{
"epoch": 0.33,
"learning_rate": 1.579062279710879e-05,
"loss": 0.6782,
"step": 79
},
{
"epoch": 0.33,
"learning_rate": 1.568064746731156e-05,
"loss": 0.689,
"step": 80
},
{
"epoch": 0.33,
"learning_rate": 1.5569648233264395e-05,
"loss": 0.6775,
"step": 81
},
{
"epoch": 0.34,
"learning_rate": 1.5457645101945046e-05,
"loss": 0.5874,
"step": 82
},
{
"epoch": 0.34,
"learning_rate": 1.5344658261278013e-05,
"loss": 0.5696,
"step": 83
},
{
"epoch": 0.35,
"learning_rate": 1.5230708076495777e-05,
"loss": 0.5891,
"step": 84
},
{
"epoch": 0.35,
"learning_rate": 1.5115815086468103e-05,
"loss": 0.562,
"step": 85
},
{
"epoch": 0.36,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.5867,
"step": 86
},
{
"epoch": 0.36,
"learning_rate": 1.4883283692099114e-05,
"loss": 0.6345,
"step": 87
},
{
"epoch": 0.36,
"learning_rate": 1.4765687200213079e-05,
"loss": 0.6353,
"step": 88
},
{
"epoch": 0.37,
"learning_rate": 1.4647231720437687e-05,
"loss": 0.5674,
"step": 89
},
{
"epoch": 0.37,
"learning_rate": 1.4527938603696376e-05,
"loss": 0.5137,
"step": 90
},
{
"epoch": 0.38,
"learning_rate": 1.4407829351891858e-05,
"loss": 0.6074,
"step": 91
},
{
"epoch": 0.38,
"learning_rate": 1.4286925614030542e-05,
"loss": 0.6169,
"step": 92
},
{
"epoch": 0.38,
"learning_rate": 1.4165249182320401e-05,
"loss": 0.5901,
"step": 93
},
{
"epoch": 0.39,
"learning_rate": 1.404282198824305e-05,
"loss": 0.5825,
"step": 94
},
{
"epoch": 0.39,
"learning_rate": 1.3919666098600753e-05,
"loss": 0.6836,
"step": 95
},
{
"epoch": 0.4,
"learning_rate": 1.3795803711538966e-05,
"loss": 0.5596,
"step": 96
},
{
"epoch": 0.4,
"learning_rate": 1.3671257152545277e-05,
"loss": 0.5923,
"step": 97
},
{
"epoch": 0.4,
"learning_rate": 1.3546048870425356e-05,
"loss": 0.6089,
"step": 98
},
{
"epoch": 0.41,
"learning_rate": 1.342020143325669e-05,
"loss": 0.5906,
"step": 99
},
{
"epoch": 0.41,
"learning_rate": 1.3293737524320798e-05,
"loss": 0.5283,
"step": 100
},
{
"epoch": 0.42,
"learning_rate": 1.3166679938014728e-05,
"loss": 0.6548,
"step": 101
},
{
"epoch": 0.42,
"learning_rate": 1.303905157574247e-05,
"loss": 0.5793,
"step": 102
},
{
"epoch": 0.43,
"learning_rate": 1.291087544178713e-05,
"loss": 0.5471,
"step": 103
},
{
"epoch": 0.43,
"learning_rate": 1.2782174639164528e-05,
"loss": 0.6052,
"step": 104
},
{
"epoch": 0.43,
"learning_rate": 1.2652972365459008e-05,
"loss": 0.6357,
"step": 105
},
{
"epoch": 0.44,
"learning_rate": 1.2523291908642219e-05,
"loss": 0.5308,
"step": 106
},
{
"epoch": 0.44,
"learning_rate": 1.2393156642875579e-05,
"loss": 0.6238,
"step": 107
},
{
"epoch": 0.45,
"learning_rate": 1.2262590024297226e-05,
"loss": 0.6016,
"step": 108
},
{
"epoch": 0.45,
"learning_rate": 1.2131615586794162e-05,
"loss": 0.5947,
"step": 109
},
{
"epoch": 0.45,
"learning_rate": 1.2000256937760446e-05,
"loss": 0.626,
"step": 110
},
{
"epoch": 0.46,
"learning_rate": 1.1868537753842052e-05,
"loss": 0.5552,
"step": 111
},
{
"epoch": 0.46,
"learning_rate": 1.1736481776669307e-05,
"loss": 0.6091,
"step": 112
},
{
"epoch": 0.47,
"learning_rate": 1.1604112808577603e-05,
"loss": 0.6089,
"step": 113
},
{
"epoch": 0.47,
"learning_rate": 1.1471454708317163e-05,
"loss": 0.5085,
"step": 114
},
{
"epoch": 0.48,
"learning_rate": 1.1338531386752618e-05,
"loss": 0.574,
"step": 115
},
{
"epoch": 0.48,
"learning_rate": 1.1205366802553231e-05,
"loss": 0.5652,
"step": 116
},
{
"epoch": 0.48,
"learning_rate": 1.107198495787448e-05,
"loss": 0.5852,
"step": 117
},
{
"epoch": 0.49,
"learning_rate": 1.0938409894031793e-05,
"loss": 0.5049,
"step": 118
},
{
"epoch": 0.49,
"learning_rate": 1.0804665687167262e-05,
"loss": 0.6001,
"step": 119
},
{
"epoch": 0.5,
"learning_rate": 1.0670776443910024e-05,
"loss": 0.6045,
"step": 120
},
{
"epoch": 0.5,
"learning_rate": 1.0536766297031216e-05,
"loss": 0.574,
"step": 121
},
{
"epoch": 0.5,
"learning_rate": 1.0402659401094154e-05,
"loss": 0.5769,
"step": 122
},
{
"epoch": 0.51,
"learning_rate": 1.0268479928100615e-05,
"loss": 0.5615,
"step": 123
},
{
"epoch": 0.51,
"learning_rate": 1.0134252063133976e-05,
"loss": 0.6157,
"step": 124
},
{
"epoch": 0.52,
"learning_rate": 1e-05,
"loss": 0.6018,
"step": 125
},
{
"epoch": 0.52,
"learning_rate": 9.865747936866027e-06,
"loss": 0.5439,
"step": 126
},
{
"epoch": 0.52,
"learning_rate": 9.73152007189939e-06,
"loss": 0.5176,
"step": 127
},
{
"epoch": 0.53,
"learning_rate": 9.597340598905851e-06,
"loss": 0.6396,
"step": 128
},
{
"epoch": 0.53,
"learning_rate": 9.463233702968784e-06,
"loss": 0.626,
"step": 129
},
{
"epoch": 0.54,
"learning_rate": 9.329223556089976e-06,
"loss": 0.6184,
"step": 130
},
{
"epoch": 0.54,
"learning_rate": 9.195334312832742e-06,
"loss": 0.5371,
"step": 131
},
{
"epoch": 0.55,
"learning_rate": 9.061590105968208e-06,
"loss": 0.5627,
"step": 132
},
{
"epoch": 0.55,
"learning_rate": 8.928015042125523e-06,
"loss": 0.5444,
"step": 133
},
{
"epoch": 0.55,
"learning_rate": 8.79463319744677e-06,
"loss": 0.5645,
"step": 134
},
{
"epoch": 0.56,
"learning_rate": 8.661468613247387e-06,
"loss": 0.5764,
"step": 135
},
{
"epoch": 0.56,
"learning_rate": 8.528545291682839e-06,
"loss": 0.5312,
"step": 136
},
{
"epoch": 0.57,
"learning_rate": 8.395887191422397e-06,
"loss": 0.5015,
"step": 137
},
{
"epoch": 0.57,
"learning_rate": 8.263518223330698e-06,
"loss": 0.5667,
"step": 138
},
{
"epoch": 0.57,
"learning_rate": 8.131462246157953e-06,
"loss": 0.5728,
"step": 139
},
{
"epoch": 0.58,
"learning_rate": 7.999743062239557e-06,
"loss": 0.573,
"step": 140
},
{
"epoch": 0.58,
"learning_rate": 7.868384413205842e-06,
"loss": 0.5425,
"step": 141
},
{
"epoch": 0.59,
"learning_rate": 7.73740997570278e-06,
"loss": 0.6074,
"step": 142
},
{
"epoch": 0.59,
"learning_rate": 7.606843357124426e-06,
"loss": 0.5059,
"step": 143
},
{
"epoch": 0.6,
"learning_rate": 7.476708091357783e-06,
"loss": 0.5574,
"step": 144
},
{
"epoch": 0.6,
"learning_rate": 7.347027634540993e-06,
"loss": 0.5779,
"step": 145
},
{
"epoch": 0.6,
"learning_rate": 7.217825360835475e-06,
"loss": 0.5581,
"step": 146
},
{
"epoch": 0.61,
"learning_rate": 7.089124558212872e-06,
"loss": 0.5833,
"step": 147
},
{
"epoch": 0.61,
"learning_rate": 6.960948424257532e-06,
"loss": 0.53,
"step": 148
},
{
"epoch": 0.62,
"learning_rate": 6.833320061985278e-06,
"loss": 0.6011,
"step": 149
},
{
"epoch": 0.62,
"learning_rate": 6.706262475679205e-06,
"loss": 0.5784,
"step": 150
},
{
"epoch": 0.62,
"learning_rate": 6.579798566743314e-06,
"loss": 0.6519,
"step": 151
},
{
"epoch": 0.63,
"learning_rate": 6.453951129574644e-06,
"loss": 0.6189,
"step": 152
},
{
"epoch": 0.63,
"learning_rate": 6.3287428474547256e-06,
"loss": 0.6582,
"step": 153
},
{
"epoch": 0.64,
"learning_rate": 6.204196288461037e-06,
"loss": 0.6318,
"step": 154
},
{
"epoch": 0.64,
"learning_rate": 6.080333901399252e-06,
"loss": 0.542,
"step": 155
},
{
"epoch": 0.64,
"learning_rate": 5.957178011756952e-06,
"loss": 0.5054,
"step": 156
},
{
"epoch": 0.65,
"learning_rate": 5.834750817679606e-06,
"loss": 0.623,
"step": 157
},
{
"epoch": 0.65,
"learning_rate": 5.713074385969457e-06,
"loss": 0.5654,
"step": 158
},
{
"epoch": 0.66,
"learning_rate": 5.5921706481081405e-06,
"loss": 0.5664,
"step": 159
},
{
"epoch": 0.66,
"learning_rate": 5.47206139630363e-06,
"loss": 0.5759,
"step": 160
},
{
"epoch": 0.67,
"learning_rate": 5.352768279562315e-06,
"loss": 0.5417,
"step": 161
},
{
"epoch": 0.67,
"learning_rate": 5.234312799786921e-06,
"loss": 0.5359,
"step": 162
},
{
"epoch": 0.67,
"learning_rate": 5.116716307900893e-06,
"loss": 0.5945,
"step": 163
},
{
"epoch": 0.68,
"learning_rate": 5.000000000000003e-06,
"loss": 0.6191,
"step": 164
},
{
"epoch": 0.68,
"learning_rate": 4.8841849135319015e-06,
"loss": 0.5498,
"step": 165
},
{
"epoch": 0.69,
"learning_rate": 4.769291923504226e-06,
"loss": 0.5535,
"step": 166
},
{
"epoch": 0.69,
"learning_rate": 4.655341738721989e-06,
"loss": 0.5085,
"step": 167
},
{
"epoch": 0.69,
"learning_rate": 4.542354898054953e-06,
"loss": 0.5669,
"step": 168
},
{
"epoch": 0.7,
"learning_rate": 4.430351766735609e-06,
"loss": 0.541,
"step": 169
},
{
"epoch": 0.7,
"learning_rate": 4.319352532688444e-06,
"loss": 0.5391,
"step": 170
},
{
"epoch": 0.71,
"learning_rate": 4.209377202891212e-06,
"loss": 0.543,
"step": 171
},
{
"epoch": 0.71,
"learning_rate": 4.100445599768774e-06,
"loss": 0.5576,
"step": 172
},
{
"epoch": 0.71,
"learning_rate": 3.99257735762021e-06,
"loss": 0.5491,
"step": 173
},
{
"epoch": 0.72,
"learning_rate": 3.885791919079878e-06,
"loss": 0.6331,
"step": 174
},
{
"epoch": 0.72,
"learning_rate": 3.7801085316129615e-06,
"loss": 0.5042,
"step": 175
},
{
"epoch": 0.73,
"learning_rate": 3.6755462440462288e-06,
"loss": 0.5757,
"step": 176
},
{
"epoch": 0.73,
"learning_rate": 3.5721239031346067e-06,
"loss": 0.5269,
"step": 177
},
{
"epoch": 0.74,
"learning_rate": 3.4698601501641517e-06,
"loss": 0.6038,
"step": 178
},
{
"epoch": 0.74,
"learning_rate": 3.3687734175920505e-06,
"loss": 0.5627,
"step": 179
},
{
"epoch": 0.74,
"learning_rate": 3.2688819257242963e-06,
"loss": 0.5598,
"step": 180
},
{
"epoch": 0.75,
"learning_rate": 3.1702036794315837e-06,
"loss": 0.5784,
"step": 181
},
{
"epoch": 0.75,
"learning_rate": 3.0727564649040066e-06,
"loss": 0.5947,
"step": 182
},
{
"epoch": 0.76,
"learning_rate": 2.976557846445225e-06,
"loss": 0.5703,
"step": 183
},
{
"epoch": 0.76,
"learning_rate": 2.8816251633065963e-06,
"loss": 0.5503,
"step": 184
},
{
"epoch": 0.76,
"learning_rate": 2.7879755265618558e-06,
"loss": 0.5996,
"step": 185
},
{
"epoch": 0.77,
"learning_rate": 2.69562581602297e-06,
"loss": 0.5635,
"step": 186
},
{
"epoch": 0.77,
"learning_rate": 2.6045926771976306e-06,
"loss": 0.564,
"step": 187
},
{
"epoch": 0.78,
"learning_rate": 2.514892518288988e-06,
"loss": 0.5691,
"step": 188
},
{
"epoch": 0.78,
"learning_rate": 2.4265415072382016e-06,
"loss": 0.5339,
"step": 189
},
{
"epoch": 0.79,
"learning_rate": 2.339555568810221e-06,
"loss": 0.5229,
"step": 190
},
{
"epoch": 0.79,
"learning_rate": 2.2539503817234553e-06,
"loss": 0.5537,
"step": 191
},
{
"epoch": 0.79,
"learning_rate": 2.1697413758237785e-06,
"loss": 0.4932,
"step": 192
},
{
"epoch": 0.8,
"learning_rate": 2.0869437293033835e-06,
"loss": 0.6104,
"step": 193
},
{
"epoch": 0.8,
"learning_rate": 2.0055723659649907e-06,
"loss": 0.5398,
"step": 194
},
{
"epoch": 0.81,
"learning_rate": 1.9256419525319316e-06,
"loss": 0.5278,
"step": 195
},
{
"epoch": 0.81,
"learning_rate": 1.8471668960045575e-06,
"loss": 0.5093,
"step": 196
},
{
"epoch": 0.81,
"learning_rate": 1.7701613410634367e-06,
"loss": 0.6057,
"step": 197
},
{
"epoch": 0.82,
"learning_rate": 1.6946391675198838e-06,
"loss": 0.5505,
"step": 198
},
{
"epoch": 0.82,
"learning_rate": 1.620613987814189e-06,
"loss": 0.5613,
"step": 199
},
{
"epoch": 0.83,
"learning_rate": 1.5480991445620541e-06,
"loss": 0.6006,
"step": 200
},
{
"epoch": 0.83,
"learning_rate": 1.4771077081496654e-06,
"loss": 0.5789,
"step": 201
},
{
"epoch": 0.83,
"learning_rate": 1.407652474377832e-06,
"loss": 0.5161,
"step": 202
},
{
"epoch": 0.84,
"learning_rate": 1.339745962155613e-06,
"loss": 0.5525,
"step": 203
},
{
"epoch": 0.84,
"learning_rate": 1.273400411243857e-06,
"loss": 0.5801,
"step": 204
},
{
"epoch": 0.85,
"learning_rate": 1.2086277800490554e-06,
"loss": 0.5269,
"step": 205
},
{
"epoch": 0.85,
"learning_rate": 1.1454397434679022e-06,
"loss": 0.5669,
"step": 206
},
{
"epoch": 0.86,
"learning_rate": 1.083847690782972e-06,
"loss": 0.5742,
"step": 207
},
{
"epoch": 0.86,
"learning_rate": 1.0238627236098619e-06,
"loss": 0.5005,
"step": 208
},
{
"epoch": 0.86,
"learning_rate": 9.65495653896179e-07,
"loss": 0.6187,
"step": 209
},
{
"epoch": 0.87,
"learning_rate": 9.08757001972762e-07,
"loss": 0.5801,
"step": 210
},
{
"epoch": 0.87,
"learning_rate": 8.536569946574546e-07,
"loss": 0.5444,
"step": 211
},
{
"epoch": 0.88,
"learning_rate": 8.002055634117578e-07,
"loss": 0.5105,
"step": 212
},
{
"epoch": 0.88,
"learning_rate": 7.48412342550765e-07,
"loss": 0.5176,
"step": 213
},
{
"epoch": 0.88,
"learning_rate": 6.98286667506618e-07,
"loss": 0.5625,
"step": 214
},
{
"epoch": 0.89,
"learning_rate": 6.498375731458529e-07,
"loss": 0.5059,
"step": 215
},
{
"epoch": 0.89,
"learning_rate": 6.030737921409169e-07,
"loss": 0.5312,
"step": 216
},
{
"epoch": 0.9,
"learning_rate": 5.580037533961546e-07,
"loss": 0.5488,
"step": 217
},
{
"epoch": 0.9,
"learning_rate": 5.146355805285452e-07,
"loss": 0.481,
"step": 218
},
{
"epoch": 0.9,
"learning_rate": 4.7297709040346474e-07,
"loss": 0.4875,
"step": 219
},
{
"epoch": 0.91,
"learning_rate": 4.3303579172574884e-07,
"loss": 0.5137,
"step": 220
},
{
"epoch": 0.91,
"learning_rate": 3.9481888368627764e-07,
"loss": 0.5737,
"step": 221
},
{
"epoch": 0.92,
"learning_rate": 3.5833325466437697e-07,
"loss": 0.5044,
"step": 222
},
{
"epoch": 0.92,
"learning_rate": 3.235854809862193e-07,
"loss": 0.4939,
"step": 223
},
{
"epoch": 0.93,
"learning_rate": 2.905818257394799e-07,
"loss": 0.5977,
"step": 224
},
{
"epoch": 0.93,
"learning_rate": 2.593282376444539e-07,
"loss": 0.5251,
"step": 225
},
{
"epoch": 0.93,
"learning_rate": 2.2983034998182997e-07,
"loss": 0.4976,
"step": 226
},
{
"epoch": 0.94,
"learning_rate": 2.0209347957732328e-07,
"loss": 0.5818,
"step": 227
},
{
"epoch": 0.94,
"learning_rate": 1.761226258433524e-07,
"loss": 0.5381,
"step": 228
},
{
"epoch": 0.95,
"learning_rate": 1.519224698779198e-07,
"loss": 0.5776,
"step": 229
},
{
"epoch": 0.95,
"learning_rate": 1.2949737362087156e-07,
"loss": 0.5542,
"step": 230
},
{
"epoch": 0.95,
"learning_rate": 1.0885137906768373e-07,
"loss": 0.5691,
"step": 231
},
{
"epoch": 0.96,
"learning_rate": 8.99882075409153e-08,
"loss": 0.5183,
"step": 232
},
{
"epoch": 0.96,
"learning_rate": 7.291125901946027e-08,
"loss": 0.6221,
"step": 233
},
{
"epoch": 0.97,
"learning_rate": 5.7623611525721155e-08,
"loss": 0.6509,
"step": 234
},
{
"epoch": 0.97,
"learning_rate": 4.412802057081278e-08,
"loss": 0.6748,
"step": 235
},
{
"epoch": 0.98,
"learning_rate": 3.242691865790071e-08,
"loss": 0.5635,
"step": 236
},
{
"epoch": 0.98,
"learning_rate": 2.2522414843748618e-08,
"loss": 0.6003,
"step": 237
},
{
"epoch": 0.98,
"learning_rate": 1.4416294358582383e-08,
"loss": 0.5596,
"step": 238
},
{
"epoch": 0.99,
"learning_rate": 8.110018284304132e-09,
"loss": 0.5457,
"step": 239
},
{
"epoch": 0.99,
"learning_rate": 3.6047232911462506e-09,
"loss": 0.5715,
"step": 240
},
{
"epoch": 1.0,
"learning_rate": 9.012214327897006e-10,
"loss": 0.5107,
"step": 241
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.4875,
"step": 242
},
{
"epoch": 1.0,
"step": 242,
"total_flos": 3.991902850132214e+17,
"train_loss": 0.6390532186208677,
"train_runtime": 1897.3254,
"train_samples_per_second": 16.267,
"train_steps_per_second": 0.128
}
],
"max_steps": 242,
"num_train_epochs": 1,
"total_flos": 3.991902850132214e+17,
"trial_name": null,
"trial_params": null
}