SAKI-InternVL3-2B-visual7w-full / trainer_state.json
autoRiver's picture
Add files using upload-large-folder tool
9d7b7cb verified
Raw
History Blame Contribute Delete
122 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9955555555555557,
"eval_steps": 500,
"global_step": 1011,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 9.090909090909091e-07,
"loss": 3.6704,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.8181818181818183e-06,
"loss": 3.8654,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 2.7272727272727272e-06,
"loss": 3.081,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 3.6363636363636366e-06,
"loss": 3.3095,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 4.5454545454545455e-06,
"loss": 3.1735,
"step": 5
},
{
"epoch": 0.02,
"learning_rate": 5.4545454545454545e-06,
"loss": 2.4712,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 6.363636363636364e-06,
"loss": 2.5496,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 7.272727272727273e-06,
"loss": 2.0481,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 8.181818181818183e-06,
"loss": 1.8122,
"step": 9
},
{
"epoch": 0.03,
"learning_rate": 9.090909090909091e-06,
"loss": 1.877,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1e-05,
"loss": 1.543,
"step": 11
},
{
"epoch": 0.04,
"learning_rate": 9.999975326009292e-06,
"loss": 0.8764,
"step": 12
},
{
"epoch": 0.04,
"learning_rate": 9.999901304280686e-06,
"loss": 0.7352,
"step": 13
},
{
"epoch": 0.04,
"learning_rate": 9.99977793554475e-06,
"loss": 1.4572,
"step": 14
},
{
"epoch": 0.04,
"learning_rate": 9.999605221019082e-06,
"loss": 1.1429,
"step": 15
},
{
"epoch": 0.05,
"learning_rate": 9.999383162408303e-06,
"loss": 1.0002,
"step": 16
},
{
"epoch": 0.05,
"learning_rate": 9.999111761904046e-06,
"loss": 0.544,
"step": 17
},
{
"epoch": 0.05,
"learning_rate": 9.998791022184921e-06,
"loss": 0.7894,
"step": 18
},
{
"epoch": 0.06,
"learning_rate": 9.9984209464165e-06,
"loss": 1.4244,
"step": 19
},
{
"epoch": 0.06,
"learning_rate": 9.998001538251283e-06,
"loss": 0.7784,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 9.997532801828659e-06,
"loss": 0.9172,
"step": 21
},
{
"epoch": 0.07,
"learning_rate": 9.997014741774866e-06,
"loss": 0.5907,
"step": 22
},
{
"epoch": 0.07,
"learning_rate": 9.996447363202947e-06,
"loss": 0.7452,
"step": 23
},
{
"epoch": 0.07,
"learning_rate": 9.995830671712701e-06,
"loss": 1.2977,
"step": 24
},
{
"epoch": 0.07,
"learning_rate": 9.995164673390624e-06,
"loss": 0.8772,
"step": 25
},
{
"epoch": 0.08,
"learning_rate": 9.994449374809851e-06,
"loss": 0.6488,
"step": 26
},
{
"epoch": 0.08,
"learning_rate": 9.99368478303009e-06,
"loss": 1.4196,
"step": 27
},
{
"epoch": 0.08,
"learning_rate": 9.992870905597549e-06,
"loss": 0.8043,
"step": 28
},
{
"epoch": 0.09,
"learning_rate": 9.992007750544876e-06,
"loss": 1.5243,
"step": 29
},
{
"epoch": 0.09,
"learning_rate": 9.991095326391061e-06,
"loss": 1.5279,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 9.990133642141359e-06,
"loss": 0.9806,
"step": 31
},
{
"epoch": 0.09,
"learning_rate": 9.98912270728721e-06,
"loss": 0.8041,
"step": 32
},
{
"epoch": 0.1,
"learning_rate": 9.988062531806127e-06,
"loss": 0.3133,
"step": 33
},
{
"epoch": 0.1,
"learning_rate": 9.98695312616162e-06,
"loss": 0.6731,
"step": 34
},
{
"epoch": 0.1,
"learning_rate": 9.98579450130307e-06,
"loss": 1.0422,
"step": 35
},
{
"epoch": 0.11,
"learning_rate": 9.984586668665641e-06,
"loss": 1.1711,
"step": 36
},
{
"epoch": 0.11,
"learning_rate": 9.98332964017015e-06,
"loss": 0.7941,
"step": 37
},
{
"epoch": 0.11,
"learning_rate": 9.982023428222963e-06,
"loss": 0.9158,
"step": 38
},
{
"epoch": 0.12,
"learning_rate": 9.980668045715864e-06,
"loss": 0.5645,
"step": 39
},
{
"epoch": 0.12,
"learning_rate": 9.97926350602593e-06,
"loss": 0.3849,
"step": 40
},
{
"epoch": 0.12,
"learning_rate": 9.9778098230154e-06,
"loss": 0.7624,
"step": 41
},
{
"epoch": 0.12,
"learning_rate": 9.976307011031542e-06,
"loss": 0.601,
"step": 42
},
{
"epoch": 0.13,
"learning_rate": 9.974755084906503e-06,
"loss": 0.7451,
"step": 43
},
{
"epoch": 0.13,
"learning_rate": 9.973154059957162e-06,
"loss": 0.5975,
"step": 44
},
{
"epoch": 0.13,
"learning_rate": 9.971503951984996e-06,
"loss": 0.918,
"step": 45
},
{
"epoch": 0.14,
"learning_rate": 9.9698047772759e-06,
"loss": 1.1868,
"step": 46
},
{
"epoch": 0.14,
"learning_rate": 9.968056552600043e-06,
"loss": 1.0946,
"step": 47
},
{
"epoch": 0.14,
"learning_rate": 9.966259295211698e-06,
"loss": 0.5763,
"step": 48
},
{
"epoch": 0.15,
"learning_rate": 9.964413022849069e-06,
"loss": 1.0594,
"step": 49
},
{
"epoch": 0.15,
"learning_rate": 9.96251775373412e-06,
"loss": 1.0811,
"step": 50
},
{
"epoch": 0.15,
"learning_rate": 9.960573506572391e-06,
"loss": 0.7799,
"step": 51
},
{
"epoch": 0.15,
"learning_rate": 9.958580300552816e-06,
"loss": 1.2399,
"step": 52
},
{
"epoch": 0.16,
"learning_rate": 9.956538155347534e-06,
"loss": 0.9918,
"step": 53
},
{
"epoch": 0.16,
"learning_rate": 9.954447091111695e-06,
"loss": 0.7477,
"step": 54
},
{
"epoch": 0.16,
"learning_rate": 9.952307128483257e-06,
"loss": 0.5662,
"step": 55
},
{
"epoch": 0.17,
"learning_rate": 9.95011828858279e-06,
"loss": 0.794,
"step": 56
},
{
"epoch": 0.17,
"learning_rate": 9.947880593013256e-06,
"loss": 0.7309,
"step": 57
},
{
"epoch": 0.17,
"learning_rate": 9.94559406385981e-06,
"loss": 1.2217,
"step": 58
},
{
"epoch": 0.17,
"learning_rate": 9.94325872368957e-06,
"loss": 0.7106,
"step": 59
},
{
"epoch": 0.18,
"learning_rate": 9.940874595551403e-06,
"loss": 0.6933,
"step": 60
},
{
"epoch": 0.18,
"learning_rate": 9.938441702975689e-06,
"loss": 1.2025,
"step": 61
},
{
"epoch": 0.18,
"learning_rate": 9.935960069974096e-06,
"loss": 0.553,
"step": 62
},
{
"epoch": 0.19,
"learning_rate": 9.93342972103934e-06,
"loss": 1.1507,
"step": 63
},
{
"epoch": 0.19,
"learning_rate": 9.930850681144946e-06,
"loss": 0.7507,
"step": 64
},
{
"epoch": 0.19,
"learning_rate": 9.928222975744992e-06,
"loss": 0.9146,
"step": 65
},
{
"epoch": 0.2,
"learning_rate": 9.92554663077387e-06,
"loss": 1.0014,
"step": 66
},
{
"epoch": 0.2,
"learning_rate": 9.922821672646028e-06,
"loss": 0.5639,
"step": 67
},
{
"epoch": 0.2,
"learning_rate": 9.920048128255699e-06,
"loss": 0.4192,
"step": 68
},
{
"epoch": 0.2,
"learning_rate": 9.91722602497665e-06,
"loss": 0.4047,
"step": 69
},
{
"epoch": 0.21,
"learning_rate": 9.914355390661897e-06,
"loss": 0.9003,
"step": 70
},
{
"epoch": 0.21,
"learning_rate": 9.911436253643445e-06,
"loss": 0.5146,
"step": 71
},
{
"epoch": 0.21,
"learning_rate": 9.908468642731996e-06,
"loss": 0.472,
"step": 72
},
{
"epoch": 0.22,
"learning_rate": 9.90545258721667e-06,
"loss": 1.1053,
"step": 73
},
{
"epoch": 0.22,
"learning_rate": 9.902388116864723e-06,
"loss": 1.1375,
"step": 74
},
{
"epoch": 0.22,
"learning_rate": 9.899275261921236e-06,
"loss": 0.8971,
"step": 75
},
{
"epoch": 0.23,
"learning_rate": 9.89611405310883e-06,
"loss": 0.5624,
"step": 76
},
{
"epoch": 0.23,
"learning_rate": 9.89290452162736e-06,
"loss": 0.7429,
"step": 77
},
{
"epoch": 0.23,
"learning_rate": 9.88964669915361e-06,
"loss": 0.4981,
"step": 78
},
{
"epoch": 0.23,
"learning_rate": 9.886340617840968e-06,
"loss": 1.0147,
"step": 79
},
{
"epoch": 0.24,
"learning_rate": 9.882986310319124e-06,
"loss": 0.7448,
"step": 80
},
{
"epoch": 0.24,
"learning_rate": 9.879583809693737e-06,
"loss": 1.2266,
"step": 81
},
{
"epoch": 0.24,
"learning_rate": 9.876133149546117e-06,
"loss": 0.4505,
"step": 82
},
{
"epoch": 0.25,
"learning_rate": 9.872634363932887e-06,
"loss": 0.5065,
"step": 83
},
{
"epoch": 0.25,
"learning_rate": 9.869087487385644e-06,
"loss": 0.9039,
"step": 84
},
{
"epoch": 0.25,
"learning_rate": 9.865492554910634e-06,
"loss": 0.9735,
"step": 85
},
{
"epoch": 0.25,
"learning_rate": 9.861849601988384e-06,
"loss": 0.974,
"step": 86
},
{
"epoch": 0.26,
"learning_rate": 9.85815866457337e-06,
"loss": 0.9725,
"step": 87
},
{
"epoch": 0.26,
"learning_rate": 9.854419779093656e-06,
"loss": 0.861,
"step": 88
},
{
"epoch": 0.26,
"learning_rate": 9.85063298245053e-06,
"loss": 0.8065,
"step": 89
},
{
"epoch": 0.27,
"learning_rate": 9.846798312018147e-06,
"loss": 1.1007,
"step": 90
},
{
"epoch": 0.27,
"learning_rate": 9.842915805643156e-06,
"loss": 0.6498,
"step": 91
},
{
"epoch": 0.27,
"learning_rate": 9.838985501644329e-06,
"loss": 0.8587,
"step": 92
},
{
"epoch": 0.28,
"learning_rate": 9.835007438812177e-06,
"loss": 1.3961,
"step": 93
},
{
"epoch": 0.28,
"learning_rate": 9.830981656408575e-06,
"loss": 0.6261,
"step": 94
},
{
"epoch": 0.28,
"learning_rate": 9.82690819416637e-06,
"loss": 0.6485,
"step": 95
},
{
"epoch": 0.28,
"learning_rate": 9.822787092288991e-06,
"loss": 0.4599,
"step": 96
},
{
"epoch": 0.29,
"learning_rate": 9.81861839145005e-06,
"loss": 0.6492,
"step": 97
},
{
"epoch": 0.29,
"learning_rate": 9.814402132792939e-06,
"loss": 0.781,
"step": 98
},
{
"epoch": 0.29,
"learning_rate": 9.81013835793043e-06,
"loss": 1.1889,
"step": 99
},
{
"epoch": 0.3,
"learning_rate": 9.80582710894426e-06,
"loss": 0.7099,
"step": 100
},
{
"epoch": 0.3,
"learning_rate": 9.801468428384716e-06,
"loss": 0.7722,
"step": 101
},
{
"epoch": 0.3,
"learning_rate": 9.797062359270215e-06,
"loss": 0.8005,
"step": 102
},
{
"epoch": 0.31,
"learning_rate": 9.79260894508688e-06,
"loss": 0.7567,
"step": 103
},
{
"epoch": 0.31,
"learning_rate": 9.788108229788111e-06,
"loss": 0.8045,
"step": 104
},
{
"epoch": 0.31,
"learning_rate": 9.783560257794153e-06,
"loss": 0.7741,
"step": 105
},
{
"epoch": 0.31,
"learning_rate": 9.778965073991652e-06,
"loss": 0.7754,
"step": 106
},
{
"epoch": 0.32,
"learning_rate": 9.774322723733216e-06,
"loss": 0.693,
"step": 107
},
{
"epoch": 0.32,
"learning_rate": 9.769633252836969e-06,
"loss": 1.096,
"step": 108
},
{
"epoch": 0.32,
"learning_rate": 9.764896707586095e-06,
"loss": 0.8342,
"step": 109
},
{
"epoch": 0.33,
"learning_rate": 9.760113134728383e-06,
"loss": 0.5432,
"step": 110
},
{
"epoch": 0.33,
"learning_rate": 9.755282581475769e-06,
"loss": 1.1527,
"step": 111
},
{
"epoch": 0.33,
"learning_rate": 9.750405095503859e-06,
"loss": 0.9927,
"step": 112
},
{
"epoch": 0.33,
"learning_rate": 9.745480724951473e-06,
"loss": 0.5546,
"step": 113
},
{
"epoch": 0.34,
"learning_rate": 9.74050951842016e-06,
"loss": 1.0772,
"step": 114
},
{
"epoch": 0.34,
"learning_rate": 9.735491524973723e-06,
"loss": 0.8245,
"step": 115
},
{
"epoch": 0.34,
"learning_rate": 9.730426794137727e-06,
"loss": 0.8091,
"step": 116
},
{
"epoch": 0.35,
"learning_rate": 9.725315375899025e-06,
"loss": 0.984,
"step": 117
},
{
"epoch": 0.35,
"learning_rate": 9.72015732070525e-06,
"loss": 0.7716,
"step": 118
},
{
"epoch": 0.35,
"learning_rate": 9.714952679464324e-06,
"loss": 0.6291,
"step": 119
},
{
"epoch": 0.36,
"learning_rate": 9.709701503543954e-06,
"loss": 1.1444,
"step": 120
},
{
"epoch": 0.36,
"learning_rate": 9.704403844771128e-06,
"loss": 0.8665,
"step": 121
},
{
"epoch": 0.36,
"learning_rate": 9.699059755431599e-06,
"loss": 0.4795,
"step": 122
},
{
"epoch": 0.36,
"learning_rate": 9.693669288269371e-06,
"loss": 0.2765,
"step": 123
},
{
"epoch": 0.37,
"learning_rate": 9.688232496486179e-06,
"loss": 0.7443,
"step": 124
},
{
"epoch": 0.37,
"learning_rate": 9.682749433740963e-06,
"loss": 1.4403,
"step": 125
},
{
"epoch": 0.37,
"learning_rate": 9.677220154149338e-06,
"loss": 0.8673,
"step": 126
},
{
"epoch": 0.38,
"learning_rate": 9.671644712283061e-06,
"loss": 0.9159,
"step": 127
},
{
"epoch": 0.38,
"learning_rate": 9.666023163169493e-06,
"loss": 0.262,
"step": 128
},
{
"epoch": 0.38,
"learning_rate": 9.660355562291055e-06,
"loss": 1.0002,
"step": 129
},
{
"epoch": 0.39,
"learning_rate": 9.65464196558468e-06,
"loss": 0.716,
"step": 130
},
{
"epoch": 0.39,
"learning_rate": 9.648882429441258e-06,
"loss": 0.9188,
"step": 131
},
{
"epoch": 0.39,
"learning_rate": 9.643077010705088e-06,
"loss": 0.696,
"step": 132
},
{
"epoch": 0.39,
"learning_rate": 9.637225766673309e-06,
"loss": 0.5727,
"step": 133
},
{
"epoch": 0.4,
"learning_rate": 9.631328755095334e-06,
"loss": 0.7179,
"step": 134
},
{
"epoch": 0.4,
"learning_rate": 9.62538603417229e-06,
"loss": 1.1148,
"step": 135
},
{
"epoch": 0.4,
"learning_rate": 9.619397662556434e-06,
"loss": 0.6376,
"step": 136
},
{
"epoch": 0.41,
"learning_rate": 9.613363699350575e-06,
"loss": 0.6447,
"step": 137
},
{
"epoch": 0.41,
"learning_rate": 9.607284204107493e-06,
"loss": 0.5552,
"step": 138
},
{
"epoch": 0.41,
"learning_rate": 9.601159236829353e-06,
"loss": 0.6763,
"step": 139
},
{
"epoch": 0.41,
"learning_rate": 9.594988857967107e-06,
"loss": 0.8666,
"step": 140
},
{
"epoch": 0.42,
"learning_rate": 9.588773128419907e-06,
"loss": 0.7681,
"step": 141
},
{
"epoch": 0.42,
"learning_rate": 9.58251210953449e-06,
"loss": 1.0483,
"step": 142
},
{
"epoch": 0.42,
"learning_rate": 9.576205863104588e-06,
"loss": 0.5776,
"step": 143
},
{
"epoch": 0.43,
"learning_rate": 9.569854451370306e-06,
"loss": 0.7225,
"step": 144
},
{
"epoch": 0.43,
"learning_rate": 9.563457937017514e-06,
"loss": 0.819,
"step": 145
},
{
"epoch": 0.43,
"learning_rate": 9.557016383177226e-06,
"loss": 0.5488,
"step": 146
},
{
"epoch": 0.44,
"learning_rate": 9.550529853424979e-06,
"loss": 0.7717,
"step": 147
},
{
"epoch": 0.44,
"learning_rate": 9.543998411780202e-06,
"loss": 0.3269,
"step": 148
},
{
"epoch": 0.44,
"learning_rate": 9.537422122705585e-06,
"loss": 0.3406,
"step": 149
},
{
"epoch": 0.44,
"learning_rate": 9.530801051106449e-06,
"loss": 0.4942,
"step": 150
},
{
"epoch": 0.45,
"learning_rate": 9.524135262330098e-06,
"loss": 0.5427,
"step": 151
},
{
"epoch": 0.45,
"learning_rate": 9.517424822165175e-06,
"loss": 0.4134,
"step": 152
},
{
"epoch": 0.45,
"learning_rate": 9.510669796841014e-06,
"loss": 0.5074,
"step": 153
},
{
"epoch": 0.46,
"learning_rate": 9.503870253026992e-06,
"loss": 0.68,
"step": 154
},
{
"epoch": 0.46,
"learning_rate": 9.497026257831856e-06,
"loss": 1.4278,
"step": 155
},
{
"epoch": 0.46,
"learning_rate": 9.490137878803078e-06,
"loss": 0.7842,
"step": 156
},
{
"epoch": 0.47,
"learning_rate": 9.48320518392618e-06,
"loss": 0.7716,
"step": 157
},
{
"epoch": 0.47,
"learning_rate": 9.476228241624059e-06,
"loss": 1.2787,
"step": 158
},
{
"epoch": 0.47,
"learning_rate": 9.46920712075632e-06,
"loss": 1.4852,
"step": 159
},
{
"epoch": 0.47,
"learning_rate": 9.46214189061859e-06,
"loss": 0.8951,
"step": 160
},
{
"epoch": 0.48,
"learning_rate": 9.45503262094184e-06,
"loss": 0.8581,
"step": 161
},
{
"epoch": 0.48,
"learning_rate": 9.447879381891691e-06,
"loss": 0.5185,
"step": 162
},
{
"epoch": 0.48,
"learning_rate": 9.440682244067724e-06,
"loss": 0.9121,
"step": 163
},
{
"epoch": 0.49,
"learning_rate": 9.433441278502784e-06,
"loss": 1.0564,
"step": 164
},
{
"epoch": 0.49,
"learning_rate": 9.426156556662276e-06,
"loss": 0.6961,
"step": 165
},
{
"epoch": 0.49,
"learning_rate": 9.418828150443469e-06,
"loss": 0.6104,
"step": 166
},
{
"epoch": 0.49,
"learning_rate": 9.411456132174768e-06,
"loss": 1.2485,
"step": 167
},
{
"epoch": 0.5,
"learning_rate": 9.404040574615018e-06,
"loss": 0.6443,
"step": 168
},
{
"epoch": 0.5,
"learning_rate": 9.396581550952781e-06,
"loss": 0.5873,
"step": 169
},
{
"epoch": 0.5,
"learning_rate": 9.38907913480561e-06,
"loss": 0.8001,
"step": 170
},
{
"epoch": 0.51,
"learning_rate": 9.381533400219319e-06,
"loss": 0.4015,
"step": 171
},
{
"epoch": 0.51,
"learning_rate": 9.373944421667264e-06,
"loss": 0.8988,
"step": 172
},
{
"epoch": 0.51,
"learning_rate": 9.366312274049602e-06,
"loss": 1.101,
"step": 173
},
{
"epoch": 0.52,
"learning_rate": 9.358637032692546e-06,
"loss": 0.9774,
"step": 174
},
{
"epoch": 0.52,
"learning_rate": 9.35091877334763e-06,
"loss": 0.813,
"step": 175
},
{
"epoch": 0.52,
"learning_rate": 9.343157572190957e-06,
"loss": 0.705,
"step": 176
},
{
"epoch": 0.52,
"learning_rate": 9.33535350582245e-06,
"loss": 0.8088,
"step": 177
},
{
"epoch": 0.53,
"learning_rate": 9.327506651265096e-06,
"loss": 0.5402,
"step": 178
},
{
"epoch": 0.53,
"learning_rate": 9.319617085964177e-06,
"loss": 0.887,
"step": 179
},
{
"epoch": 0.53,
"learning_rate": 9.31168488778652e-06,
"loss": 0.512,
"step": 180
},
{
"epoch": 0.54,
"learning_rate": 9.30371013501972e-06,
"loss": 0.4161,
"step": 181
},
{
"epoch": 0.54,
"learning_rate": 9.295692906371362e-06,
"loss": 0.7829,
"step": 182
},
{
"epoch": 0.54,
"learning_rate": 9.287633280968263e-06,
"loss": 0.8998,
"step": 183
},
{
"epoch": 0.55,
"learning_rate": 9.279531338355666e-06,
"loss": 0.5485,
"step": 184
},
{
"epoch": 0.55,
"learning_rate": 9.271387158496477e-06,
"loss": 1.0056,
"step": 185
},
{
"epoch": 0.55,
"learning_rate": 9.263200821770462e-06,
"loss": 1.1618,
"step": 186
},
{
"epoch": 0.55,
"learning_rate": 9.25497240897346e-06,
"loss": 1.1717,
"step": 187
},
{
"epoch": 0.56,
"learning_rate": 9.246702001316584e-06,
"loss": 0.9905,
"step": 188
},
{
"epoch": 0.56,
"learning_rate": 9.238389680425417e-06,
"loss": 0.641,
"step": 189
},
{
"epoch": 0.56,
"learning_rate": 9.230035528339212e-06,
"loss": 0.7812,
"step": 190
},
{
"epoch": 0.57,
"learning_rate": 9.221639627510076e-06,
"loss": 0.4575,
"step": 191
},
{
"epoch": 0.57,
"learning_rate": 9.213202060802162e-06,
"loss": 0.5168,
"step": 192
},
{
"epoch": 0.57,
"learning_rate": 9.204722911490847e-06,
"loss": 1.2884,
"step": 193
},
{
"epoch": 0.57,
"learning_rate": 9.196202263261908e-06,
"loss": 0.8092,
"step": 194
},
{
"epoch": 0.58,
"learning_rate": 9.18764020021071e-06,
"loss": 1.1165,
"step": 195
},
{
"epoch": 0.58,
"learning_rate": 9.179036806841352e-06,
"loss": 0.9346,
"step": 196
},
{
"epoch": 0.58,
"learning_rate": 9.170392168065858e-06,
"loss": 0.9443,
"step": 197
},
{
"epoch": 0.59,
"learning_rate": 9.161706369203319e-06,
"loss": 1.1309,
"step": 198
},
{
"epoch": 0.59,
"learning_rate": 9.152979495979064e-06,
"loss": 1.3126,
"step": 199
},
{
"epoch": 0.59,
"learning_rate": 9.14421163452381e-06,
"loss": 0.5749,
"step": 200
},
{
"epoch": 0.6,
"learning_rate": 9.13540287137281e-06,
"loss": 0.7429,
"step": 201
},
{
"epoch": 0.6,
"learning_rate": 9.126553293465e-06,
"loss": 0.5262,
"step": 202
},
{
"epoch": 0.6,
"learning_rate": 9.117662988142138e-06,
"loss": 0.9797,
"step": 203
},
{
"epoch": 0.6,
"learning_rate": 9.108732043147952e-06,
"loss": 0.703,
"step": 204
},
{
"epoch": 0.61,
"learning_rate": 9.099760546627262e-06,
"loss": 0.7192,
"step": 205
},
{
"epoch": 0.61,
"learning_rate": 9.090748587125118e-06,
"loss": 1.1918,
"step": 206
},
{
"epoch": 0.61,
"learning_rate": 9.08169625358592e-06,
"loss": 0.6773,
"step": 207
},
{
"epoch": 0.62,
"learning_rate": 9.072603635352548e-06,
"loss": 0.4512,
"step": 208
},
{
"epoch": 0.62,
"learning_rate": 9.06347082216547e-06,
"loss": 0.8288,
"step": 209
},
{
"epoch": 0.62,
"learning_rate": 9.054297904161868e-06,
"loss": 0.7037,
"step": 210
},
{
"epoch": 0.63,
"learning_rate": 9.045084971874738e-06,
"loss": 0.9553,
"step": 211
},
{
"epoch": 0.63,
"learning_rate": 9.035832116232002e-06,
"loss": 0.6199,
"step": 212
},
{
"epoch": 0.63,
"learning_rate": 9.026539428555609e-06,
"loss": 0.7928,
"step": 213
},
{
"epoch": 0.63,
"learning_rate": 9.017207000560639e-06,
"loss": 1.0496,
"step": 214
},
{
"epoch": 0.64,
"learning_rate": 9.007834924354384e-06,
"loss": 0.6966,
"step": 215
},
{
"epoch": 0.64,
"learning_rate": 8.998423292435455e-06,
"loss": 1.0694,
"step": 216
},
{
"epoch": 0.64,
"learning_rate": 8.988972197692857e-06,
"loss": 0.6842,
"step": 217
},
{
"epoch": 0.65,
"learning_rate": 8.97948173340508e-06,
"loss": 0.819,
"step": 218
},
{
"epoch": 0.65,
"learning_rate": 8.969951993239177e-06,
"loss": 1.1949,
"step": 219
},
{
"epoch": 0.65,
"learning_rate": 8.960383071249837e-06,
"loss": 0.5491,
"step": 220
},
{
"epoch": 0.65,
"learning_rate": 8.950775061878453e-06,
"loss": 0.915,
"step": 221
},
{
"epoch": 0.66,
"learning_rate": 8.9411280599522e-06,
"loss": 0.818,
"step": 222
},
{
"epoch": 0.66,
"learning_rate": 8.931442160683094e-06,
"loss": 0.6872,
"step": 223
},
{
"epoch": 0.66,
"learning_rate": 8.921717459667052e-06,
"loss": 0.5837,
"step": 224
},
{
"epoch": 0.67,
"learning_rate": 8.911954052882941e-06,
"loss": 0.4762,
"step": 225
},
{
"epoch": 0.67,
"learning_rate": 8.902152036691649e-06,
"loss": 0.7791,
"step": 226
},
{
"epoch": 0.67,
"learning_rate": 8.892311507835118e-06,
"loss": 0.5179,
"step": 227
},
{
"epoch": 0.68,
"learning_rate": 8.882432563435394e-06,
"loss": 0.5505,
"step": 228
},
{
"epoch": 0.68,
"learning_rate": 8.872515300993669e-06,
"loss": 0.4806,
"step": 229
},
{
"epoch": 0.68,
"learning_rate": 8.862559818389322e-06,
"loss": 1.3794,
"step": 230
},
{
"epoch": 0.68,
"learning_rate": 8.852566213878947e-06,
"loss": 0.5778,
"step": 231
},
{
"epoch": 0.69,
"learning_rate": 8.842534586095383e-06,
"loss": 0.9853,
"step": 232
},
{
"epoch": 0.69,
"learning_rate": 8.83246503404675e-06,
"loss": 0.5953,
"step": 233
},
{
"epoch": 0.69,
"learning_rate": 8.82235765711546e-06,
"loss": 0.4332,
"step": 234
},
{
"epoch": 0.7,
"learning_rate": 8.81221255505724e-06,
"loss": 0.4086,
"step": 235
},
{
"epoch": 0.7,
"learning_rate": 8.802029828000157e-06,
"loss": 0.5621,
"step": 236
},
{
"epoch": 0.7,
"learning_rate": 8.791809576443611e-06,
"loss": 0.584,
"step": 237
},
{
"epoch": 0.71,
"learning_rate": 8.78155190125736e-06,
"loss": 0.4177,
"step": 238
},
{
"epoch": 0.71,
"learning_rate": 8.77125690368052e-06,
"loss": 0.9932,
"step": 239
},
{
"epoch": 0.71,
"learning_rate": 8.760924685320558e-06,
"loss": 1.1094,
"step": 240
},
{
"epoch": 0.71,
"learning_rate": 8.750555348152299e-06,
"loss": 0.5549,
"step": 241
},
{
"epoch": 0.72,
"learning_rate": 8.740148994516912e-06,
"loss": 0.5891,
"step": 242
},
{
"epoch": 0.72,
"learning_rate": 8.729705727120911e-06,
"loss": 0.5701,
"step": 243
},
{
"epoch": 0.72,
"learning_rate": 8.719225649035126e-06,
"loss": 0.896,
"step": 244
},
{
"epoch": 0.73,
"learning_rate": 8.708708863693696e-06,
"loss": 0.5006,
"step": 245
},
{
"epoch": 0.73,
"learning_rate": 8.69815547489305e-06,
"loss": 0.6423,
"step": 246
},
{
"epoch": 0.73,
"learning_rate": 8.68756558679087e-06,
"loss": 0.5675,
"step": 247
},
{
"epoch": 0.73,
"learning_rate": 8.67693930390508e-06,
"loss": 1.0767,
"step": 248
},
{
"epoch": 0.74,
"learning_rate": 8.666276731112802e-06,
"loss": 0.8987,
"step": 249
},
{
"epoch": 0.74,
"learning_rate": 8.655577973649322e-06,
"loss": 0.851,
"step": 250
},
{
"epoch": 0.74,
"learning_rate": 8.644843137107058e-06,
"loss": 0.9187,
"step": 251
},
{
"epoch": 0.75,
"learning_rate": 8.634072327434515e-06,
"loss": 0.9224,
"step": 252
},
{
"epoch": 0.75,
"learning_rate": 8.623265650935233e-06,
"loss": 0.2461,
"step": 253
},
{
"epoch": 0.75,
"learning_rate": 8.612423214266749e-06,
"loss": 0.9108,
"step": 254
},
{
"epoch": 0.76,
"learning_rate": 8.601545124439535e-06,
"loss": 0.2528,
"step": 255
},
{
"epoch": 0.76,
"learning_rate": 8.590631488815945e-06,
"loss": 1.0557,
"step": 256
},
{
"epoch": 0.76,
"learning_rate": 8.579682415109156e-06,
"loss": 1.2658,
"step": 257
},
{
"epoch": 0.76,
"learning_rate": 8.568698011382108e-06,
"loss": 0.6026,
"step": 258
},
{
"epoch": 0.77,
"learning_rate": 8.557678386046429e-06,
"loss": 0.7813,
"step": 259
},
{
"epoch": 0.77,
"learning_rate": 8.54662364786137e-06,
"loss": 0.4103,
"step": 260
},
{
"epoch": 0.77,
"learning_rate": 8.535533905932739e-06,
"loss": 0.4603,
"step": 261
},
{
"epoch": 0.78,
"learning_rate": 8.524409269711808e-06,
"loss": 1.3112,
"step": 262
},
{
"epoch": 0.78,
"learning_rate": 8.513249848994248e-06,
"loss": 0.3503,
"step": 263
},
{
"epoch": 0.78,
"learning_rate": 8.502055753919033e-06,
"loss": 0.6314,
"step": 264
},
{
"epoch": 0.79,
"learning_rate": 8.490827094967364e-06,
"loss": 0.5087,
"step": 265
},
{
"epoch": 0.79,
"learning_rate": 8.479563982961572e-06,
"loss": 0.6134,
"step": 266
},
{
"epoch": 0.79,
"learning_rate": 8.468266529064025e-06,
"loss": 0.3755,
"step": 267
},
{
"epoch": 0.79,
"learning_rate": 8.456934844776033e-06,
"loss": 0.897,
"step": 268
},
{
"epoch": 0.8,
"learning_rate": 8.445569041936743e-06,
"loss": 0.8859,
"step": 269
},
{
"epoch": 0.8,
"learning_rate": 8.434169232722043e-06,
"loss": 0.7047,
"step": 270
},
{
"epoch": 0.8,
"learning_rate": 8.422735529643445e-06,
"loss": 0.8778,
"step": 271
},
{
"epoch": 0.81,
"learning_rate": 8.411268045546984e-06,
"loss": 0.2597,
"step": 272
},
{
"epoch": 0.81,
"learning_rate": 8.399766893612096e-06,
"loss": 0.9406,
"step": 273
},
{
"epoch": 0.81,
"learning_rate": 8.388232187350513e-06,
"loss": 0.7187,
"step": 274
},
{
"epoch": 0.81,
"learning_rate": 8.376664040605122e-06,
"loss": 1.1675,
"step": 275
},
{
"epoch": 0.82,
"learning_rate": 8.365062567548868e-06,
"loss": 0.9677,
"step": 276
},
{
"epoch": 0.82,
"learning_rate": 8.353427882683601e-06,
"loss": 0.4612,
"step": 277
},
{
"epoch": 0.82,
"learning_rate": 8.341760100838967e-06,
"loss": 0.8105,
"step": 278
},
{
"epoch": 0.83,
"learning_rate": 8.33005933717126e-06,
"loss": 0.9147,
"step": 279
},
{
"epoch": 0.83,
"learning_rate": 8.318325707162293e-06,
"loss": 0.6995,
"step": 280
},
{
"epoch": 0.83,
"learning_rate": 8.30655932661826e-06,
"loss": 0.798,
"step": 281
},
{
"epoch": 0.84,
"learning_rate": 8.294760311668586e-06,
"loss": 1.1807,
"step": 282
},
{
"epoch": 0.84,
"learning_rate": 8.282928778764783e-06,
"loss": 1.0107,
"step": 283
},
{
"epoch": 0.84,
"learning_rate": 8.271064844679306e-06,
"loss": 1.0432,
"step": 284
},
{
"epoch": 0.84,
"learning_rate": 8.259168626504395e-06,
"loss": 1.0358,
"step": 285
},
{
"epoch": 0.85,
"learning_rate": 8.247240241650918e-06,
"loss": 0.72,
"step": 286
},
{
"epoch": 0.85,
"learning_rate": 8.235279807847223e-06,
"loss": 0.8982,
"step": 287
},
{
"epoch": 0.85,
"learning_rate": 8.223287443137957e-06,
"loss": 0.5655,
"step": 288
},
{
"epoch": 0.86,
"learning_rate": 8.211263265882923e-06,
"loss": 0.7632,
"step": 289
},
{
"epoch": 0.86,
"learning_rate": 8.199207394755892e-06,
"loss": 0.9297,
"step": 290
},
{
"epoch": 0.86,
"learning_rate": 8.18711994874345e-06,
"loss": 0.5562,
"step": 291
},
{
"epoch": 0.87,
"learning_rate": 8.175001047143804e-06,
"loss": 0.9494,
"step": 292
},
{
"epoch": 0.87,
"learning_rate": 8.162850809565623e-06,
"loss": 0.5714,
"step": 293
},
{
"epoch": 0.87,
"learning_rate": 8.150669355926848e-06,
"loss": 1.2367,
"step": 294
},
{
"epoch": 0.87,
"learning_rate": 8.138456806453503e-06,
"loss": 0.7605,
"step": 295
},
{
"epoch": 0.88,
"learning_rate": 8.126213281678527e-06,
"loss": 0.8142,
"step": 296
},
{
"epoch": 0.88,
"learning_rate": 8.113938902440563e-06,
"loss": 0.6616,
"step": 297
},
{
"epoch": 0.88,
"learning_rate": 8.101633789882781e-06,
"loss": 0.7682,
"step": 298
},
{
"epoch": 0.89,
"learning_rate": 8.089298065451673e-06,
"loss": 0.9119,
"step": 299
},
{
"epoch": 0.89,
"learning_rate": 8.076931850895858e-06,
"loss": 1.1528,
"step": 300
},
{
"epoch": 0.89,
"learning_rate": 8.064535268264883e-06,
"loss": 0.5227,
"step": 301
},
{
"epoch": 0.89,
"learning_rate": 8.052108439908014e-06,
"loss": 0.5716,
"step": 302
},
{
"epoch": 0.9,
"learning_rate": 8.039651488473028e-06,
"loss": 0.8335,
"step": 303
},
{
"epoch": 0.9,
"learning_rate": 8.027164536905008e-06,
"loss": 0.4539,
"step": 304
},
{
"epoch": 0.9,
"learning_rate": 8.014647708445124e-06,
"loss": 0.81,
"step": 305
},
{
"epoch": 0.91,
"learning_rate": 8.002101126629422e-06,
"loss": 1.2715,
"step": 306
},
{
"epoch": 0.91,
"learning_rate": 7.989524915287595e-06,
"loss": 1.0417,
"step": 307
},
{
"epoch": 0.91,
"learning_rate": 7.976919198541775e-06,
"loss": 0.7917,
"step": 308
},
{
"epoch": 0.92,
"learning_rate": 7.964284100805297e-06,
"loss": 0.6394,
"step": 309
},
{
"epoch": 0.92,
"learning_rate": 7.951619746781474e-06,
"loss": 0.9316,
"step": 310
},
{
"epoch": 0.92,
"learning_rate": 7.938926261462366e-06,
"loss": 0.8156,
"step": 311
},
{
"epoch": 0.92,
"learning_rate": 7.926203770127552e-06,
"loss": 0.5974,
"step": 312
},
{
"epoch": 0.93,
"learning_rate": 7.913452398342882e-06,
"loss": 1.0814,
"step": 313
},
{
"epoch": 0.93,
"learning_rate": 7.900672271959247e-06,
"loss": 1.1107,
"step": 314
},
{
"epoch": 0.93,
"learning_rate": 7.887863517111337e-06,
"loss": 0.9725,
"step": 315
},
{
"epoch": 0.94,
"learning_rate": 7.875026260216395e-06,
"loss": 1.1188,
"step": 316
},
{
"epoch": 0.94,
"learning_rate": 7.862160627972956e-06,
"loss": 0.4008,
"step": 317
},
{
"epoch": 0.94,
"learning_rate": 7.849266747359619e-06,
"loss": 1.064,
"step": 318
},
{
"epoch": 0.95,
"learning_rate": 7.836344745633785e-06,
"loss": 0.8253,
"step": 319
},
{
"epoch": 0.95,
"learning_rate": 7.823394750330386e-06,
"loss": 0.8087,
"step": 320
},
{
"epoch": 0.95,
"learning_rate": 7.810416889260653e-06,
"loss": 1.0052,
"step": 321
},
{
"epoch": 0.95,
"learning_rate": 7.797411290510836e-06,
"loss": 0.7134,
"step": 322
},
{
"epoch": 0.96,
"learning_rate": 7.78437808244094e-06,
"loss": 1.0733,
"step": 323
},
{
"epoch": 0.96,
"learning_rate": 7.771317393683471e-06,
"loss": 1.2127,
"step": 324
},
{
"epoch": 0.96,
"learning_rate": 7.758229353142153e-06,
"loss": 0.4901,
"step": 325
},
{
"epoch": 0.97,
"learning_rate": 7.74511408999066e-06,
"loss": 0.4975,
"step": 326
},
{
"epoch": 0.97,
"learning_rate": 7.731971733671347e-06,
"loss": 0.8529,
"step": 327
},
{
"epoch": 0.97,
"learning_rate": 7.718802413893963e-06,
"loss": 0.7797,
"step": 328
},
{
"epoch": 0.97,
"learning_rate": 7.70560626063438e-06,
"loss": 1.0622,
"step": 329
},
{
"epoch": 0.98,
"learning_rate": 7.692383404133302e-06,
"loss": 0.8248,
"step": 330
},
{
"epoch": 0.98,
"learning_rate": 7.679133974894984e-06,
"loss": 0.3405,
"step": 331
},
{
"epoch": 0.98,
"learning_rate": 7.665858103685944e-06,
"loss": 0.5791,
"step": 332
},
{
"epoch": 0.99,
"learning_rate": 7.652555921533671e-06,
"loss": 0.4466,
"step": 333
},
{
"epoch": 0.99,
"learning_rate": 7.639227559725333e-06,
"loss": 0.6526,
"step": 334
},
{
"epoch": 0.99,
"learning_rate": 7.6258731498064796e-06,
"loss": 0.3516,
"step": 335
},
{
"epoch": 1.0,
"learning_rate": 7.612492823579744e-06,
"loss": 0.4791,
"step": 336
},
{
"epoch": 1.0,
"learning_rate": 7.5990867131035474e-06,
"loss": 0.4033,
"step": 337
},
{
"epoch": 1.0,
"learning_rate": 7.585654950690786e-06,
"loss": 0.8813,
"step": 338
},
{
"epoch": 1.0,
"learning_rate": 7.572197668907533e-06,
"loss": 0.7984,
"step": 339
},
{
"epoch": 1.01,
"learning_rate": 7.5587150005717256e-06,
"loss": 0.8195,
"step": 340
},
{
"epoch": 1.01,
"learning_rate": 7.545207078751858e-06,
"loss": 1.025,
"step": 341
},
{
"epoch": 1.01,
"learning_rate": 7.531674036765662e-06,
"loss": 0.5973,
"step": 342
},
{
"epoch": 1.02,
"learning_rate": 7.518116008178805e-06,
"loss": 0.3446,
"step": 343
},
{
"epoch": 1.02,
"learning_rate": 7.5045331268035505e-06,
"loss": 0.3659,
"step": 344
},
{
"epoch": 1.02,
"learning_rate": 7.490925526697455e-06,
"loss": 0.4343,
"step": 345
},
{
"epoch": 1.03,
"learning_rate": 7.477293342162038e-06,
"loss": 0.5229,
"step": 346
},
{
"epoch": 1.03,
"learning_rate": 7.463636707741458e-06,
"loss": 0.7872,
"step": 347
},
{
"epoch": 1.03,
"learning_rate": 7.449955758221184e-06,
"loss": 0.5095,
"step": 348
},
{
"epoch": 1.03,
"learning_rate": 7.436250628626662e-06,
"loss": 0.5274,
"step": 349
},
{
"epoch": 1.04,
"learning_rate": 7.42252145422199e-06,
"loss": 0.6428,
"step": 350
},
{
"epoch": 1.04,
"learning_rate": 7.408768370508577e-06,
"loss": 0.5024,
"step": 351
},
{
"epoch": 1.04,
"learning_rate": 7.394991513223806e-06,
"loss": 1.3939,
"step": 352
},
{
"epoch": 1.05,
"learning_rate": 7.381191018339697e-06,
"loss": 0.7416,
"step": 353
},
{
"epoch": 1.05,
"learning_rate": 7.3673670220615615e-06,
"loss": 0.6143,
"step": 354
},
{
"epoch": 1.05,
"learning_rate": 7.353519660826665e-06,
"loss": 0.4654,
"step": 355
},
{
"epoch": 1.05,
"learning_rate": 7.3396490713028674e-06,
"loss": 0.5087,
"step": 356
},
{
"epoch": 1.06,
"learning_rate": 7.325755390387293e-06,
"loss": 0.6051,
"step": 357
},
{
"epoch": 1.06,
"learning_rate": 7.31183875520496e-06,
"loss": 0.4184,
"step": 358
},
{
"epoch": 1.06,
"learning_rate": 7.297899303107441e-06,
"loss": 0.6711,
"step": 359
},
{
"epoch": 1.07,
"learning_rate": 7.283937171671498e-06,
"loss": 0.6007,
"step": 360
},
{
"epoch": 1.07,
"learning_rate": 7.269952498697734e-06,
"loss": 0.7316,
"step": 361
},
{
"epoch": 1.07,
"learning_rate": 7.2559454222092265e-06,
"loss": 0.8477,
"step": 362
},
{
"epoch": 1.08,
"learning_rate": 7.241916080450163e-06,
"loss": 0.671,
"step": 363
},
{
"epoch": 1.08,
"learning_rate": 7.227864611884483e-06,
"loss": 0.8339,
"step": 364
},
{
"epoch": 1.08,
"learning_rate": 7.21379115519451e-06,
"loss": 0.4638,
"step": 365
},
{
"epoch": 1.08,
"learning_rate": 7.199695849279576e-06,
"loss": 0.5445,
"step": 366
},
{
"epoch": 1.09,
"learning_rate": 7.185578833254665e-06,
"loss": 0.3726,
"step": 367
},
{
"epoch": 1.09,
"learning_rate": 7.171440246449024e-06,
"loss": 0.6089,
"step": 368
},
{
"epoch": 1.09,
"learning_rate": 7.157280228404796e-06,
"loss": 0.2451,
"step": 369
},
{
"epoch": 1.1,
"learning_rate": 7.143098918875643e-06,
"loss": 0.9142,
"step": 370
},
{
"epoch": 1.1,
"learning_rate": 7.128896457825364e-06,
"loss": 0.4222,
"step": 371
},
{
"epoch": 1.1,
"learning_rate": 7.114672985426516e-06,
"loss": 0.8029,
"step": 372
},
{
"epoch": 1.11,
"learning_rate": 7.100428642059033e-06,
"loss": 0.5958,
"step": 373
},
{
"epoch": 1.11,
"learning_rate": 7.086163568308828e-06,
"loss": 0.6472,
"step": 374
},
{
"epoch": 1.11,
"learning_rate": 7.071877904966422e-06,
"loss": 0.6891,
"step": 375
},
{
"epoch": 1.11,
"learning_rate": 7.057571793025545e-06,
"loss": 0.5977,
"step": 376
},
{
"epoch": 1.12,
"learning_rate": 7.043245373681746e-06,
"loss": 0.8907,
"step": 377
},
{
"epoch": 1.12,
"learning_rate": 7.028898788331e-06,
"loss": 0.8414,
"step": 378
},
{
"epoch": 1.12,
"learning_rate": 7.014532178568314e-06,
"loss": 0.6476,
"step": 379
},
{
"epoch": 1.13,
"learning_rate": 7.0001456861863236e-06,
"loss": 0.6617,
"step": 380
},
{
"epoch": 1.13,
"learning_rate": 6.985739453173903e-06,
"loss": 0.7768,
"step": 381
},
{
"epoch": 1.13,
"learning_rate": 6.971313621714756e-06,
"loss": 0.9925,
"step": 382
},
{
"epoch": 1.13,
"learning_rate": 6.9568683341860135e-06,
"loss": 0.8438,
"step": 383
},
{
"epoch": 1.14,
"learning_rate": 6.942403733156832e-06,
"loss": 0.5001,
"step": 384
},
{
"epoch": 1.14,
"learning_rate": 6.927919961386984e-06,
"loss": 0.6054,
"step": 385
},
{
"epoch": 1.14,
"learning_rate": 6.913417161825449e-06,
"loss": 0.4705,
"step": 386
},
{
"epoch": 1.15,
"learning_rate": 6.898895477609007e-06,
"loss": 0.732,
"step": 387
},
{
"epoch": 1.15,
"learning_rate": 6.884355052060814e-06,
"loss": 0.3317,
"step": 388
},
{
"epoch": 1.15,
"learning_rate": 6.869796028689002e-06,
"loss": 0.3195,
"step": 389
},
{
"epoch": 1.16,
"learning_rate": 6.8552185511852555e-06,
"loss": 0.2632,
"step": 390
},
{
"epoch": 1.16,
"learning_rate": 6.840622763423391e-06,
"loss": 0.244,
"step": 391
},
{
"epoch": 1.16,
"learning_rate": 6.82600880945794e-06,
"loss": 0.6587,
"step": 392
},
{
"epoch": 1.16,
"learning_rate": 6.811376833522729e-06,
"loss": 0.8977,
"step": 393
},
{
"epoch": 1.17,
"learning_rate": 6.796726980029454e-06,
"loss": 0.6492,
"step": 394
},
{
"epoch": 1.17,
"learning_rate": 6.782059393566254e-06,
"loss": 0.3927,
"step": 395
},
{
"epoch": 1.17,
"learning_rate": 6.767374218896286e-06,
"loss": 0.7996,
"step": 396
},
{
"epoch": 1.18,
"learning_rate": 6.752671600956295e-06,
"loss": 0.6397,
"step": 397
},
{
"epoch": 1.18,
"learning_rate": 6.737951684855185e-06,
"loss": 0.8445,
"step": 398
},
{
"epoch": 1.18,
"learning_rate": 6.723214615872585e-06,
"loss": 0.5694,
"step": 399
},
{
"epoch": 1.19,
"learning_rate": 6.708460539457418e-06,
"loss": 1.0009,
"step": 400
},
{
"epoch": 1.19,
"learning_rate": 6.693689601226458e-06,
"loss": 0.649,
"step": 401
},
{
"epoch": 1.19,
"learning_rate": 6.6789019469629034e-06,
"loss": 0.3688,
"step": 402
},
{
"epoch": 1.19,
"learning_rate": 6.664097722614934e-06,
"loss": 0.2484,
"step": 403
},
{
"epoch": 1.2,
"learning_rate": 6.649277074294265e-06,
"loss": 0.8437,
"step": 404
},
{
"epoch": 1.2,
"learning_rate": 6.634440148274712e-06,
"loss": 0.7236,
"step": 405
},
{
"epoch": 1.2,
"learning_rate": 6.619587090990748e-06,
"loss": 0.8052,
"step": 406
},
{
"epoch": 1.21,
"learning_rate": 6.604718049036047e-06,
"loss": 0.717,
"step": 407
},
{
"epoch": 1.21,
"learning_rate": 6.589833169162055e-06,
"loss": 0.3613,
"step": 408
},
{
"epoch": 1.21,
"learning_rate": 6.574932598276524e-06,
"loss": 0.5547,
"step": 409
},
{
"epoch": 1.21,
"learning_rate": 6.5600164834420754e-06,
"loss": 0.7964,
"step": 410
},
{
"epoch": 1.22,
"learning_rate": 6.545084971874738e-06,
"loss": 0.9424,
"step": 411
},
{
"epoch": 1.22,
"learning_rate": 6.530138210942505e-06,
"loss": 0.6747,
"step": 412
},
{
"epoch": 1.22,
"learning_rate": 6.5151763481638705e-06,
"loss": 0.3819,
"step": 413
},
{
"epoch": 1.23,
"learning_rate": 6.500199531206381e-06,
"loss": 0.467,
"step": 414
},
{
"epoch": 1.23,
"learning_rate": 6.485207907885175e-06,
"loss": 0.8154,
"step": 415
},
{
"epoch": 1.23,
"learning_rate": 6.47020162616152e-06,
"loss": 1.0398,
"step": 416
},
{
"epoch": 1.24,
"learning_rate": 6.455180834141359e-06,
"loss": 0.546,
"step": 417
},
{
"epoch": 1.24,
"learning_rate": 6.440145680073847e-06,
"loss": 1.1329,
"step": 418
},
{
"epoch": 1.24,
"learning_rate": 6.425096312349881e-06,
"loss": 0.4361,
"step": 419
},
{
"epoch": 1.24,
"learning_rate": 6.410032879500647e-06,
"loss": 1.6479,
"step": 420
},
{
"epoch": 1.25,
"learning_rate": 6.3949555301961474e-06,
"loss": 0.5026,
"step": 421
},
{
"epoch": 1.25,
"learning_rate": 6.3798644132437304e-06,
"loss": 0.418,
"step": 422
},
{
"epoch": 1.25,
"learning_rate": 6.364759677586627e-06,
"loss": 0.4096,
"step": 423
},
{
"epoch": 1.26,
"learning_rate": 6.349641472302484e-06,
"loss": 0.525,
"step": 424
},
{
"epoch": 1.26,
"learning_rate": 6.334509946601879e-06,
"loss": 0.4727,
"step": 425
},
{
"epoch": 1.26,
"learning_rate": 6.3193652498268656e-06,
"loss": 0.3308,
"step": 426
},
{
"epoch": 1.27,
"learning_rate": 6.304207531449486e-06,
"loss": 0.6859,
"step": 427
},
{
"epoch": 1.27,
"learning_rate": 6.2890369410703e-06,
"loss": 0.5276,
"step": 428
},
{
"epoch": 1.27,
"learning_rate": 6.273853628416911e-06,
"loss": 0.4161,
"step": 429
},
{
"epoch": 1.27,
"learning_rate": 6.258657743342486e-06,
"loss": 1.2386,
"step": 430
},
{
"epoch": 1.28,
"learning_rate": 6.243449435824276e-06,
"loss": 0.3866,
"step": 431
},
{
"epoch": 1.28,
"learning_rate": 6.228228855962133e-06,
"loss": 0.4733,
"step": 432
},
{
"epoch": 1.28,
"learning_rate": 6.212996153977038e-06,
"loss": 0.6827,
"step": 433
},
{
"epoch": 1.29,
"learning_rate": 6.1977514802096105e-06,
"loss": 0.6511,
"step": 434
},
{
"epoch": 1.29,
"learning_rate": 6.182494985118625e-06,
"loss": 0.6122,
"step": 435
},
{
"epoch": 1.29,
"learning_rate": 6.1672268192795285e-06,
"loss": 0.5142,
"step": 436
},
{
"epoch": 1.29,
"learning_rate": 6.151947133382954e-06,
"loss": 0.6888,
"step": 437
},
{
"epoch": 1.3,
"learning_rate": 6.136656078233233e-06,
"loss": 0.3579,
"step": 438
},
{
"epoch": 1.3,
"learning_rate": 6.121353804746907e-06,
"loss": 0.5189,
"step": 439
},
{
"epoch": 1.3,
"learning_rate": 6.106040463951237e-06,
"loss": 0.5957,
"step": 440
},
{
"epoch": 1.31,
"learning_rate": 6.090716206982714e-06,
"loss": 0.5698,
"step": 441
},
{
"epoch": 1.31,
"learning_rate": 6.075381185085568e-06,
"loss": 0.754,
"step": 442
},
{
"epoch": 1.31,
"learning_rate": 6.060035549610275e-06,
"loss": 0.3864,
"step": 443
},
{
"epoch": 1.32,
"learning_rate": 6.044679452012059e-06,
"loss": 0.323,
"step": 444
},
{
"epoch": 1.32,
"learning_rate": 6.029313043849407e-06,
"loss": 0.5114,
"step": 445
},
{
"epoch": 1.32,
"learning_rate": 6.013936476782563e-06,
"loss": 0.3257,
"step": 446
},
{
"epoch": 1.32,
"learning_rate": 5.9985499025720354e-06,
"loss": 0.6599,
"step": 447
},
{
"epoch": 1.33,
"learning_rate": 5.9831534730771e-06,
"loss": 0.6474,
"step": 448
},
{
"epoch": 1.33,
"learning_rate": 5.967747340254303e-06,
"loss": 0.4902,
"step": 449
},
{
"epoch": 1.33,
"learning_rate": 5.952331656155951e-06,
"loss": 0.3298,
"step": 450
},
{
"epoch": 1.34,
"learning_rate": 5.936906572928625e-06,
"loss": 0.392,
"step": 451
},
{
"epoch": 1.34,
"learning_rate": 5.9214722428116675e-06,
"loss": 0.4271,
"step": 452
},
{
"epoch": 1.34,
"learning_rate": 5.906028818135687e-06,
"loss": 0.8006,
"step": 453
},
{
"epoch": 1.35,
"learning_rate": 5.89057645132105e-06,
"loss": 0.4802,
"step": 454
},
{
"epoch": 1.35,
"learning_rate": 5.8751152948763815e-06,
"loss": 0.7187,
"step": 455
},
{
"epoch": 1.35,
"learning_rate": 5.859645501397048e-06,
"loss": 0.5567,
"step": 456
},
{
"epoch": 1.35,
"learning_rate": 5.844167223563669e-06,
"loss": 1.182,
"step": 457
},
{
"epoch": 1.36,
"learning_rate": 5.828680614140599e-06,
"loss": 0.5452,
"step": 458
},
{
"epoch": 1.36,
"learning_rate": 5.813185825974419e-06,
"loss": 0.4738,
"step": 459
},
{
"epoch": 1.36,
"learning_rate": 5.797683011992432e-06,
"loss": 0.1843,
"step": 460
},
{
"epoch": 1.37,
"learning_rate": 5.782172325201155e-06,
"loss": 0.5573,
"step": 461
},
{
"epoch": 1.37,
"learning_rate": 5.766653918684803e-06,
"loss": 1.2589,
"step": 462
},
{
"epoch": 1.37,
"learning_rate": 5.751127945603786e-06,
"loss": 0.5891,
"step": 463
},
{
"epoch": 1.37,
"learning_rate": 5.735594559193187e-06,
"loss": 0.528,
"step": 464
},
{
"epoch": 1.38,
"learning_rate": 5.720053912761261e-06,
"loss": 0.677,
"step": 465
},
{
"epoch": 1.38,
"learning_rate": 5.704506159687914e-06,
"loss": 1.3406,
"step": 466
},
{
"epoch": 1.38,
"learning_rate": 5.68895145342319e-06,
"loss": 0.6868,
"step": 467
},
{
"epoch": 1.39,
"learning_rate": 5.673389947485763e-06,
"loss": 0.4714,
"step": 468
},
{
"epoch": 1.39,
"learning_rate": 5.657821795461413e-06,
"loss": 0.5036,
"step": 469
},
{
"epoch": 1.39,
"learning_rate": 5.642247151001515e-06,
"loss": 0.8809,
"step": 470
},
{
"epoch": 1.4,
"learning_rate": 5.626666167821522e-06,
"loss": 0.4692,
"step": 471
},
{
"epoch": 1.4,
"learning_rate": 5.611078999699448e-06,
"loss": 0.409,
"step": 472
},
{
"epoch": 1.4,
"learning_rate": 5.59548580047435e-06,
"loss": 0.5143,
"step": 473
},
{
"epoch": 1.4,
"learning_rate": 5.5798867240448075e-06,
"loss": 0.5217,
"step": 474
},
{
"epoch": 1.41,
"learning_rate": 5.5642819243674085e-06,
"loss": 0.8134,
"step": 475
},
{
"epoch": 1.41,
"learning_rate": 5.548671555455226e-06,
"loss": 0.3935,
"step": 476
},
{
"epoch": 1.41,
"learning_rate": 5.5330557713763e-06,
"loss": 0.6119,
"step": 477
},
{
"epoch": 1.42,
"learning_rate": 5.517434726252113e-06,
"loss": 0.7529,
"step": 478
},
{
"epoch": 1.42,
"learning_rate": 5.5018085742560745e-06,
"loss": 0.3934,
"step": 479
},
{
"epoch": 1.42,
"learning_rate": 5.486177469611999e-06,
"loss": 0.8057,
"step": 480
},
{
"epoch": 1.43,
"learning_rate": 5.470541566592573e-06,
"loss": 0.5755,
"step": 481
},
{
"epoch": 1.43,
"learning_rate": 5.454901019517851e-06,
"loss": 0.6497,
"step": 482
},
{
"epoch": 1.43,
"learning_rate": 5.439255982753717e-06,
"loss": 0.9222,
"step": 483
},
{
"epoch": 1.43,
"learning_rate": 5.423606610710368e-06,
"loss": 0.7081,
"step": 484
},
{
"epoch": 1.44,
"learning_rate": 5.4079530578407895e-06,
"loss": 0.9882,
"step": 485
},
{
"epoch": 1.44,
"learning_rate": 5.392295478639226e-06,
"loss": 0.7794,
"step": 486
},
{
"epoch": 1.44,
"learning_rate": 5.376634027639664e-06,
"loss": 0.9232,
"step": 487
},
{
"epoch": 1.45,
"learning_rate": 5.360968859414305e-06,
"loss": 0.9868,
"step": 488
},
{
"epoch": 1.45,
"learning_rate": 5.345300128572031e-06,
"loss": 0.529,
"step": 489
},
{
"epoch": 1.45,
"learning_rate": 5.32962798975689e-06,
"loss": 1.0004,
"step": 490
},
{
"epoch": 1.45,
"learning_rate": 5.3139525976465675e-06,
"loss": 0.4804,
"step": 491
},
{
"epoch": 1.46,
"learning_rate": 5.298274106950855e-06,
"loss": 0.6725,
"step": 492
},
{
"epoch": 1.46,
"learning_rate": 5.282592672410124e-06,
"loss": 1.0645,
"step": 493
},
{
"epoch": 1.46,
"learning_rate": 5.2669084487938025e-06,
"loss": 0.6151,
"step": 494
},
{
"epoch": 1.47,
"learning_rate": 5.251221590898848e-06,
"loss": 0.7598,
"step": 495
},
{
"epoch": 1.47,
"learning_rate": 5.235532253548213e-06,
"loss": 0.624,
"step": 496
},
{
"epoch": 1.47,
"learning_rate": 5.219840591589325e-06,
"loss": 0.7862,
"step": 497
},
{
"epoch": 1.48,
"learning_rate": 5.204146759892551e-06,
"loss": 0.783,
"step": 498
},
{
"epoch": 1.48,
"learning_rate": 5.188450913349674e-06,
"loss": 0.7683,
"step": 499
},
{
"epoch": 1.48,
"learning_rate": 5.172753206872363e-06,
"loss": 0.6639,
"step": 500
},
{
"epoch": 1.48,
"learning_rate": 5.157053795390642e-06,
"loss": 0.9211,
"step": 501
},
{
"epoch": 1.49,
"learning_rate": 5.141352833851367e-06,
"loss": 0.4029,
"step": 502
},
{
"epoch": 1.49,
"learning_rate": 5.1256504772166885e-06,
"loss": 0.7503,
"step": 503
},
{
"epoch": 1.49,
"learning_rate": 5.109946880462526e-06,
"loss": 0.4327,
"step": 504
},
{
"epoch": 1.5,
"learning_rate": 5.0942421985770415e-06,
"loss": 0.4537,
"step": 505
},
{
"epoch": 1.5,
"learning_rate": 5.078536586559104e-06,
"loss": 1.0954,
"step": 506
},
{
"epoch": 1.5,
"learning_rate": 5.062830199416764e-06,
"loss": 0.7649,
"step": 507
},
{
"epoch": 1.51,
"learning_rate": 5.047123192165722e-06,
"loss": 0.4564,
"step": 508
},
{
"epoch": 1.51,
"learning_rate": 5.031415719827796e-06,
"loss": 0.713,
"step": 509
},
{
"epoch": 1.51,
"learning_rate": 5.015707937429398e-06,
"loss": 0.8004,
"step": 510
},
{
"epoch": 1.51,
"learning_rate": 5e-06,
"loss": 0.4723,
"step": 511
},
{
"epoch": 1.52,
"learning_rate": 4.984292062570603e-06,
"loss": 0.679,
"step": 512
},
{
"epoch": 1.52,
"learning_rate": 4.968584280172206e-06,
"loss": 0.6975,
"step": 513
},
{
"epoch": 1.52,
"learning_rate": 4.952876807834281e-06,
"loss": 0.8661,
"step": 514
},
{
"epoch": 1.53,
"learning_rate": 4.937169800583237e-06,
"loss": 0.7603,
"step": 515
},
{
"epoch": 1.53,
"learning_rate": 4.921463413440898e-06,
"loss": 0.506,
"step": 516
},
{
"epoch": 1.53,
"learning_rate": 4.90575780142296e-06,
"loss": 0.777,
"step": 517
},
{
"epoch": 1.53,
"learning_rate": 4.890053119537475e-06,
"loss": 0.306,
"step": 518
},
{
"epoch": 1.54,
"learning_rate": 4.874349522783313e-06,
"loss": 1.0963,
"step": 519
},
{
"epoch": 1.54,
"learning_rate": 4.8586471661486345e-06,
"loss": 0.633,
"step": 520
},
{
"epoch": 1.54,
"learning_rate": 4.842946204609359e-06,
"loss": 0.4427,
"step": 521
},
{
"epoch": 1.55,
"learning_rate": 4.827246793127639e-06,
"loss": 1.1556,
"step": 522
},
{
"epoch": 1.55,
"learning_rate": 4.811549086650327e-06,
"loss": 0.2155,
"step": 523
},
{
"epoch": 1.55,
"learning_rate": 4.79585324010745e-06,
"loss": 0.6892,
"step": 524
},
{
"epoch": 1.56,
"learning_rate": 4.780159408410677e-06,
"loss": 0.8799,
"step": 525
},
{
"epoch": 1.56,
"learning_rate": 4.7644677464517874e-06,
"loss": 0.8894,
"step": 526
},
{
"epoch": 1.56,
"learning_rate": 4.748778409101153e-06,
"loss": 1.3423,
"step": 527
},
{
"epoch": 1.56,
"learning_rate": 4.733091551206198e-06,
"loss": 0.6785,
"step": 528
},
{
"epoch": 1.57,
"learning_rate": 4.717407327589878e-06,
"loss": 0.5295,
"step": 529
},
{
"epoch": 1.57,
"learning_rate": 4.7017258930491474e-06,
"loss": 0.4646,
"step": 530
},
{
"epoch": 1.57,
"learning_rate": 4.686047402353433e-06,
"loss": 0.649,
"step": 531
},
{
"epoch": 1.58,
"learning_rate": 4.670372010243111e-06,
"loss": 0.7047,
"step": 532
},
{
"epoch": 1.58,
"learning_rate": 4.654699871427972e-06,
"loss": 0.3488,
"step": 533
},
{
"epoch": 1.58,
"learning_rate": 4.639031140585697e-06,
"loss": 0.633,
"step": 534
},
{
"epoch": 1.59,
"learning_rate": 4.6233659723603374e-06,
"loss": 0.4718,
"step": 535
},
{
"epoch": 1.59,
"learning_rate": 4.6077045213607765e-06,
"loss": 1.0392,
"step": 536
},
{
"epoch": 1.59,
"learning_rate": 4.592046942159213e-06,
"loss": 0.5922,
"step": 537
},
{
"epoch": 1.59,
"learning_rate": 4.576393389289633e-06,
"loss": 0.7208,
"step": 538
},
{
"epoch": 1.6,
"learning_rate": 4.560744017246284e-06,
"loss": 1.0903,
"step": 539
},
{
"epoch": 1.6,
"learning_rate": 4.545098980482151e-06,
"loss": 0.6144,
"step": 540
},
{
"epoch": 1.6,
"learning_rate": 4.529458433407429e-06,
"loss": 0.2287,
"step": 541
},
{
"epoch": 1.61,
"learning_rate": 4.513822530388004e-06,
"loss": 0.514,
"step": 542
},
{
"epoch": 1.61,
"learning_rate": 4.4981914257439254e-06,
"loss": 0.3978,
"step": 543
},
{
"epoch": 1.61,
"learning_rate": 4.482565273747888e-06,
"loss": 0.5293,
"step": 544
},
{
"epoch": 1.61,
"learning_rate": 4.466944228623701e-06,
"loss": 0.5814,
"step": 545
},
{
"epoch": 1.62,
"learning_rate": 4.451328444544774e-06,
"loss": 0.3387,
"step": 546
},
{
"epoch": 1.62,
"learning_rate": 4.4357180756325915e-06,
"loss": 0.6912,
"step": 547
},
{
"epoch": 1.62,
"learning_rate": 4.420113275955193e-06,
"loss": 0.4387,
"step": 548
},
{
"epoch": 1.63,
"learning_rate": 4.404514199525651e-06,
"loss": 0.6954,
"step": 549
},
{
"epoch": 1.63,
"learning_rate": 4.388921000300553e-06,
"loss": 0.5974,
"step": 550
},
{
"epoch": 1.63,
"learning_rate": 4.373333832178478e-06,
"loss": 0.5862,
"step": 551
},
{
"epoch": 1.64,
"learning_rate": 4.357752848998486e-06,
"loss": 0.3981,
"step": 552
},
{
"epoch": 1.64,
"learning_rate": 4.342178204538588e-06,
"loss": 0.6565,
"step": 553
},
{
"epoch": 1.64,
"learning_rate": 4.326610052514238e-06,
"loss": 0.573,
"step": 554
},
{
"epoch": 1.64,
"learning_rate": 4.31104854657681e-06,
"loss": 0.6382,
"step": 555
},
{
"epoch": 1.65,
"learning_rate": 4.295493840312087e-06,
"loss": 0.9154,
"step": 556
},
{
"epoch": 1.65,
"learning_rate": 4.279946087238739e-06,
"loss": 0.4963,
"step": 557
},
{
"epoch": 1.65,
"learning_rate": 4.264405440806813e-06,
"loss": 0.9118,
"step": 558
},
{
"epoch": 1.66,
"learning_rate": 4.248872054396215e-06,
"loss": 0.9341,
"step": 559
},
{
"epoch": 1.66,
"learning_rate": 4.233346081315197e-06,
"loss": 0.7765,
"step": 560
},
{
"epoch": 1.66,
"learning_rate": 4.217827674798845e-06,
"loss": 0.495,
"step": 561
},
{
"epoch": 1.67,
"learning_rate": 4.202316988007568e-06,
"loss": 0.3652,
"step": 562
},
{
"epoch": 1.67,
"learning_rate": 4.186814174025582e-06,
"loss": 0.6113,
"step": 563
},
{
"epoch": 1.67,
"learning_rate": 4.171319385859402e-06,
"loss": 0.6173,
"step": 564
},
{
"epoch": 1.67,
"learning_rate": 4.155832776436331e-06,
"loss": 0.5766,
"step": 565
},
{
"epoch": 1.68,
"learning_rate": 4.140354498602952e-06,
"loss": 0.5768,
"step": 566
},
{
"epoch": 1.68,
"learning_rate": 4.124884705123619e-06,
"loss": 0.4933,
"step": 567
},
{
"epoch": 1.68,
"learning_rate": 4.109423548678949e-06,
"loss": 0.6811,
"step": 568
},
{
"epoch": 1.69,
"learning_rate": 4.093971181864313e-06,
"loss": 1.0536,
"step": 569
},
{
"epoch": 1.69,
"learning_rate": 4.078527757188333e-06,
"loss": 0.7951,
"step": 570
},
{
"epoch": 1.69,
"learning_rate": 4.063093427071376e-06,
"loss": 0.6613,
"step": 571
},
{
"epoch": 1.69,
"learning_rate": 4.047668343844051e-06,
"loss": 0.4553,
"step": 572
},
{
"epoch": 1.7,
"learning_rate": 4.032252659745699e-06,
"loss": 0.7864,
"step": 573
},
{
"epoch": 1.7,
"learning_rate": 4.016846526922901e-06,
"loss": 0.6082,
"step": 574
},
{
"epoch": 1.7,
"learning_rate": 4.001450097427965e-06,
"loss": 0.6467,
"step": 575
},
{
"epoch": 1.71,
"learning_rate": 3.986063523217439e-06,
"loss": 0.759,
"step": 576
},
{
"epoch": 1.71,
"learning_rate": 3.970686956150595e-06,
"loss": 0.2621,
"step": 577
},
{
"epoch": 1.71,
"learning_rate": 3.955320547987943e-06,
"loss": 0.9187,
"step": 578
},
{
"epoch": 1.72,
"learning_rate": 3.939964450389728e-06,
"loss": 0.6237,
"step": 579
},
{
"epoch": 1.72,
"learning_rate": 3.924618814914435e-06,
"loss": 0.2489,
"step": 580
},
{
"epoch": 1.72,
"learning_rate": 3.909283793017289e-06,
"loss": 0.8546,
"step": 581
},
{
"epoch": 1.72,
"learning_rate": 3.8939595360487655e-06,
"loss": 0.9047,
"step": 582
},
{
"epoch": 1.73,
"learning_rate": 3.8786461952530955e-06,
"loss": 0.7214,
"step": 583
},
{
"epoch": 1.73,
"learning_rate": 3.863343921766769e-06,
"loss": 0.2606,
"step": 584
},
{
"epoch": 1.73,
"learning_rate": 3.8480528666170495e-06,
"loss": 0.7732,
"step": 585
},
{
"epoch": 1.74,
"learning_rate": 3.832773180720475e-06,
"loss": 0.7386,
"step": 586
},
{
"epoch": 1.74,
"learning_rate": 3.817505014881378e-06,
"loss": 0.7765,
"step": 587
},
{
"epoch": 1.74,
"learning_rate": 3.8022485197903924e-06,
"loss": 0.7934,
"step": 588
},
{
"epoch": 1.75,
"learning_rate": 3.787003846022964e-06,
"loss": 0.5822,
"step": 589
},
{
"epoch": 1.75,
"learning_rate": 3.7717711440378695e-06,
"loss": 0.5377,
"step": 590
},
{
"epoch": 1.75,
"learning_rate": 3.756550564175727e-06,
"loss": 0.7971,
"step": 591
},
{
"epoch": 1.75,
"learning_rate": 3.7413422566575153e-06,
"loss": 0.5596,
"step": 592
},
{
"epoch": 1.76,
"learning_rate": 3.7261463715830902e-06,
"loss": 0.556,
"step": 593
},
{
"epoch": 1.76,
"learning_rate": 3.7109630589297014e-06,
"loss": 0.5852,
"step": 594
},
{
"epoch": 1.76,
"learning_rate": 3.695792468550517e-06,
"loss": 0.6998,
"step": 595
},
{
"epoch": 1.77,
"learning_rate": 3.680634750173137e-06,
"loss": 0.3787,
"step": 596
},
{
"epoch": 1.77,
"learning_rate": 3.6654900533981234e-06,
"loss": 0.4347,
"step": 597
},
{
"epoch": 1.77,
"learning_rate": 3.6503585276975196e-06,
"loss": 0.5151,
"step": 598
},
{
"epoch": 1.77,
"learning_rate": 3.635240322413375e-06,
"loss": 0.8255,
"step": 599
},
{
"epoch": 1.78,
"learning_rate": 3.6201355867562725e-06,
"loss": 0.762,
"step": 600
},
{
"epoch": 1.78,
"learning_rate": 3.6050444698038547e-06,
"loss": 0.606,
"step": 601
},
{
"epoch": 1.78,
"learning_rate": 3.5899671204993535e-06,
"loss": 0.4158,
"step": 602
},
{
"epoch": 1.79,
"learning_rate": 3.5749036876501196e-06,
"loss": 0.6422,
"step": 603
},
{
"epoch": 1.79,
"learning_rate": 3.559854319926156e-06,
"loss": 0.4695,
"step": 604
},
{
"epoch": 1.79,
"learning_rate": 3.5448191658586423e-06,
"loss": 0.4557,
"step": 605
},
{
"epoch": 1.8,
"learning_rate": 3.5297983738384813e-06,
"loss": 0.3492,
"step": 606
},
{
"epoch": 1.8,
"learning_rate": 3.5147920921148267e-06,
"loss": 0.4855,
"step": 607
},
{
"epoch": 1.8,
"learning_rate": 3.49980046879362e-06,
"loss": 0.4001,
"step": 608
},
{
"epoch": 1.8,
"learning_rate": 3.484823651836131e-06,
"loss": 0.5279,
"step": 609
},
{
"epoch": 1.81,
"learning_rate": 3.4698617890574972e-06,
"loss": 0.446,
"step": 610
},
{
"epoch": 1.81,
"learning_rate": 3.4549150281252635e-06,
"loss": 0.6631,
"step": 611
},
{
"epoch": 1.81,
"learning_rate": 3.4399835165579266e-06,
"loss": 0.623,
"step": 612
},
{
"epoch": 1.82,
"learning_rate": 3.4250674017234774e-06,
"loss": 0.6952,
"step": 613
},
{
"epoch": 1.82,
"learning_rate": 3.4101668308379465e-06,
"loss": 0.5226,
"step": 614
},
{
"epoch": 1.82,
"learning_rate": 3.3952819509639534e-06,
"loss": 0.2537,
"step": 615
},
{
"epoch": 1.83,
"learning_rate": 3.3804129090092542e-06,
"loss": 0.6659,
"step": 616
},
{
"epoch": 1.83,
"learning_rate": 3.3655598517252886e-06,
"loss": 0.4981,
"step": 617
},
{
"epoch": 1.83,
"learning_rate": 3.350722925705736e-06,
"loss": 0.6212,
"step": 618
},
{
"epoch": 1.83,
"learning_rate": 3.3359022773850673e-06,
"loss": 0.5847,
"step": 619
},
{
"epoch": 1.84,
"learning_rate": 3.3210980530370974e-06,
"loss": 0.239,
"step": 620
},
{
"epoch": 1.84,
"learning_rate": 3.3063103987735433e-06,
"loss": 0.4722,
"step": 621
},
{
"epoch": 1.84,
"learning_rate": 3.2915394605425836e-06,
"loss": 1.0309,
"step": 622
},
{
"epoch": 1.85,
"learning_rate": 3.2767853841274154e-06,
"loss": 0.8042,
"step": 623
},
{
"epoch": 1.85,
"learning_rate": 3.262048315144816e-06,
"loss": 0.4395,
"step": 624
},
{
"epoch": 1.85,
"learning_rate": 3.247328399043706e-06,
"loss": 0.7189,
"step": 625
},
{
"epoch": 1.85,
"learning_rate": 3.2326257811037154e-06,
"loss": 0.4421,
"step": 626
},
{
"epoch": 1.86,
"learning_rate": 3.217940606433747e-06,
"loss": 0.6944,
"step": 627
},
{
"epoch": 1.86,
"learning_rate": 3.2032730199705477e-06,
"loss": 0.9294,
"step": 628
},
{
"epoch": 1.86,
"learning_rate": 3.188623166477272e-06,
"loss": 0.4487,
"step": 629
},
{
"epoch": 1.87,
"learning_rate": 3.1739911905420617e-06,
"loss": 1.0226,
"step": 630
},
{
"epoch": 1.87,
"learning_rate": 3.1593772365766107e-06,
"loss": 0.5889,
"step": 631
},
{
"epoch": 1.87,
"learning_rate": 3.144781448814746e-06,
"loss": 0.3073,
"step": 632
},
{
"epoch": 1.88,
"learning_rate": 3.130203971310999e-06,
"loss": 0.6123,
"step": 633
},
{
"epoch": 1.88,
"learning_rate": 3.1156449479391876e-06,
"loss": 0.4194,
"step": 634
},
{
"epoch": 1.88,
"learning_rate": 3.1011045223909954e-06,
"loss": 0.9148,
"step": 635
},
{
"epoch": 1.88,
"learning_rate": 3.0865828381745515e-06,
"loss": 0.3942,
"step": 636
},
{
"epoch": 1.89,
"learning_rate": 3.0720800386130176e-06,
"loss": 1.0434,
"step": 637
},
{
"epoch": 1.89,
"learning_rate": 3.0575962668431704e-06,
"loss": 0.5763,
"step": 638
},
{
"epoch": 1.89,
"learning_rate": 3.043131665813988e-06,
"loss": 0.8471,
"step": 639
},
{
"epoch": 1.9,
"learning_rate": 3.028686378285245e-06,
"loss": 0.5722,
"step": 640
},
{
"epoch": 1.9,
"learning_rate": 3.0142605468260976e-06,
"loss": 1.0777,
"step": 641
},
{
"epoch": 1.9,
"learning_rate": 2.9998543138136773e-06,
"loss": 0.4605,
"step": 642
},
{
"epoch": 1.91,
"learning_rate": 2.9854678214316875e-06,
"loss": 0.5976,
"step": 643
},
{
"epoch": 1.91,
"learning_rate": 2.9711012116690007e-06,
"loss": 0.7678,
"step": 644
},
{
"epoch": 1.91,
"learning_rate": 2.9567546263182554e-06,
"loss": 0.613,
"step": 645
},
{
"epoch": 1.91,
"learning_rate": 2.9424282069744564e-06,
"loss": 0.7015,
"step": 646
},
{
"epoch": 1.92,
"learning_rate": 2.92812209503358e-06,
"loss": 0.887,
"step": 647
},
{
"epoch": 1.92,
"learning_rate": 2.9138364316911747e-06,
"loss": 0.8143,
"step": 648
},
{
"epoch": 1.92,
"learning_rate": 2.899571357940969e-06,
"loss": 0.4772,
"step": 649
},
{
"epoch": 1.93,
"learning_rate": 2.8853270145734846e-06,
"loss": 0.4833,
"step": 650
},
{
"epoch": 1.93,
"learning_rate": 2.871103542174637e-06,
"loss": 0.4945,
"step": 651
},
{
"epoch": 1.93,
"learning_rate": 2.856901081124359e-06,
"loss": 0.7334,
"step": 652
},
{
"epoch": 1.93,
"learning_rate": 2.8427197715952047e-06,
"loss": 0.5597,
"step": 653
},
{
"epoch": 1.94,
"learning_rate": 2.8285597535509775e-06,
"loss": 0.8277,
"step": 654
},
{
"epoch": 1.94,
"learning_rate": 2.814421166745337e-06,
"loss": 0.6258,
"step": 655
},
{
"epoch": 1.94,
"learning_rate": 2.800304150720424e-06,
"loss": 0.5445,
"step": 656
},
{
"epoch": 1.95,
"learning_rate": 2.7862088448054936e-06,
"loss": 0.8419,
"step": 657
},
{
"epoch": 1.95,
"learning_rate": 2.772135388115519e-06,
"loss": 0.4667,
"step": 658
},
{
"epoch": 1.95,
"learning_rate": 2.7580839195498397e-06,
"loss": 0.506,
"step": 659
},
{
"epoch": 1.96,
"learning_rate": 2.7440545777907747e-06,
"loss": 0.5917,
"step": 660
},
{
"epoch": 1.96,
"learning_rate": 2.7300475013022666e-06,
"loss": 0.4623,
"step": 661
},
{
"epoch": 1.96,
"learning_rate": 2.716062828328502e-06,
"loss": 0.4437,
"step": 662
},
{
"epoch": 1.96,
"learning_rate": 2.7021006968925613e-06,
"loss": 0.506,
"step": 663
},
{
"epoch": 1.97,
"learning_rate": 2.6881612447950425e-06,
"loss": 0.4927,
"step": 664
},
{
"epoch": 1.97,
"learning_rate": 2.6742446096127086e-06,
"loss": 0.6321,
"step": 665
},
{
"epoch": 1.97,
"learning_rate": 2.6603509286971342e-06,
"loss": 1.0067,
"step": 666
},
{
"epoch": 1.98,
"learning_rate": 2.646480339173337e-06,
"loss": 0.7787,
"step": 667
},
{
"epoch": 1.98,
"learning_rate": 2.6326329779384397e-06,
"loss": 1.0528,
"step": 668
},
{
"epoch": 1.98,
"learning_rate": 2.618808981660304e-06,
"loss": 0.599,
"step": 669
},
{
"epoch": 1.99,
"learning_rate": 2.6050084867761953e-06,
"loss": 0.5039,
"step": 670
},
{
"epoch": 1.99,
"learning_rate": 2.5912316294914232e-06,
"loss": 0.7395,
"step": 671
},
{
"epoch": 1.99,
"learning_rate": 2.5774785457780107e-06,
"loss": 1.0472,
"step": 672
},
{
"epoch": 1.99,
"learning_rate": 2.5637493713733376e-06,
"loss": 0.8206,
"step": 673
},
{
"epoch": 2.0,
"learning_rate": 2.5500442417788176e-06,
"loss": 0.4127,
"step": 674
},
{
"epoch": 2.0,
"learning_rate": 2.536363292258543e-06,
"loss": 0.6619,
"step": 675
},
{
"epoch": 2.0,
"learning_rate": 2.5227066578379624e-06,
"loss": 0.4064,
"step": 676
},
{
"epoch": 2.01,
"learning_rate": 2.509074473302546e-06,
"loss": 0.6855,
"step": 677
},
{
"epoch": 2.01,
"learning_rate": 2.49546687319645e-06,
"loss": 0.6626,
"step": 678
},
{
"epoch": 2.01,
"learning_rate": 2.4818839918211963e-06,
"loss": 0.365,
"step": 679
},
{
"epoch": 2.01,
"learning_rate": 2.4683259632343363e-06,
"loss": 0.7547,
"step": 680
},
{
"epoch": 2.02,
"learning_rate": 2.4547929212481436e-06,
"loss": 0.8106,
"step": 681
},
{
"epoch": 2.02,
"learning_rate": 2.4412849994282744e-06,
"loss": 0.4343,
"step": 682
},
{
"epoch": 2.02,
"learning_rate": 2.4278023310924676e-06,
"loss": 0.5923,
"step": 683
},
{
"epoch": 2.03,
"learning_rate": 2.4143450493092146e-06,
"loss": 0.4262,
"step": 684
},
{
"epoch": 2.03,
"learning_rate": 2.4009132868964525e-06,
"loss": 0.8337,
"step": 685
},
{
"epoch": 2.03,
"learning_rate": 2.387507176420256e-06,
"loss": 0.4889,
"step": 686
},
{
"epoch": 2.04,
"learning_rate": 2.3741268501935212e-06,
"loss": 0.4006,
"step": 687
},
{
"epoch": 2.04,
"learning_rate": 2.3607724402746685e-06,
"loss": 0.3731,
"step": 688
},
{
"epoch": 2.04,
"learning_rate": 2.3474440784663287e-06,
"loss": 0.7166,
"step": 689
},
{
"epoch": 2.04,
"learning_rate": 2.334141896314057e-06,
"loss": 0.5694,
"step": 690
},
{
"epoch": 2.05,
"learning_rate": 2.320866025105016e-06,
"loss": 0.4453,
"step": 691
},
{
"epoch": 2.05,
"learning_rate": 2.3076165958666992e-06,
"loss": 0.5588,
"step": 692
},
{
"epoch": 2.05,
"learning_rate": 2.294393739365621e-06,
"loss": 0.489,
"step": 693
},
{
"epoch": 2.06,
"learning_rate": 2.281197586106037e-06,
"loss": 0.4052,
"step": 694
},
{
"epoch": 2.06,
"learning_rate": 2.268028266328655e-06,
"loss": 0.3563,
"step": 695
},
{
"epoch": 2.06,
"learning_rate": 2.254885910009341e-06,
"loss": 0.4939,
"step": 696
},
{
"epoch": 2.07,
"learning_rate": 2.2417706468578495e-06,
"loss": 0.559,
"step": 697
},
{
"epoch": 2.07,
"learning_rate": 2.228682606316529e-06,
"loss": 0.5561,
"step": 698
},
{
"epoch": 2.07,
"learning_rate": 2.2156219175590623e-06,
"loss": 0.431,
"step": 699
},
{
"epoch": 2.07,
"learning_rate": 2.2025887094891657e-06,
"loss": 0.464,
"step": 700
},
{
"epoch": 2.08,
"learning_rate": 2.1895831107393485e-06,
"loss": 0.7037,
"step": 701
},
{
"epoch": 2.08,
"learning_rate": 2.1766052496696155e-06,
"loss": 0.4289,
"step": 702
},
{
"epoch": 2.08,
"learning_rate": 2.1636552543662187e-06,
"loss": 0.4766,
"step": 703
},
{
"epoch": 2.09,
"learning_rate": 2.1507332526403814e-06,
"loss": 0.3871,
"step": 704
},
{
"epoch": 2.09,
"learning_rate": 2.137839372027047e-06,
"loss": 0.6175,
"step": 705
},
{
"epoch": 2.09,
"learning_rate": 2.124973739783609e-06,
"loss": 0.5485,
"step": 706
},
{
"epoch": 2.09,
"learning_rate": 2.112136482888663e-06,
"loss": 0.6003,
"step": 707
},
{
"epoch": 2.1,
"learning_rate": 2.0993277280407547e-06,
"loss": 0.5157,
"step": 708
},
{
"epoch": 2.1,
"learning_rate": 2.0865476016571206e-06,
"loss": 0.4904,
"step": 709
},
{
"epoch": 2.1,
"learning_rate": 2.0737962298724513e-06,
"loss": 0.4212,
"step": 710
},
{
"epoch": 2.11,
"learning_rate": 2.061073738537635e-06,
"loss": 0.4394,
"step": 711
},
{
"epoch": 2.11,
"learning_rate": 2.0483802532185286e-06,
"loss": 0.3351,
"step": 712
},
{
"epoch": 2.11,
"learning_rate": 2.035715899194704e-06,
"loss": 0.5716,
"step": 713
},
{
"epoch": 2.12,
"learning_rate": 2.0230808014582265e-06,
"loss": 0.4579,
"step": 714
},
{
"epoch": 2.12,
"learning_rate": 2.0104750847124075e-06,
"loss": 0.5376,
"step": 715
},
{
"epoch": 2.12,
"learning_rate": 1.9978988733705807e-06,
"loss": 0.5306,
"step": 716
},
{
"epoch": 2.12,
"learning_rate": 1.9853522915548777e-06,
"loss": 0.9483,
"step": 717
},
{
"epoch": 2.13,
"learning_rate": 1.9728354630949935e-06,
"loss": 1.2231,
"step": 718
},
{
"epoch": 2.13,
"learning_rate": 1.9603485115269743e-06,
"loss": 0.7498,
"step": 719
},
{
"epoch": 2.13,
"learning_rate": 1.9478915600919877e-06,
"loss": 0.3547,
"step": 720
},
{
"epoch": 2.14,
"learning_rate": 1.9354647317351187e-06,
"loss": 0.5417,
"step": 721
},
{
"epoch": 2.14,
"learning_rate": 1.9230681491041425e-06,
"loss": 0.2974,
"step": 722
},
{
"epoch": 2.14,
"learning_rate": 1.910701934548329e-06,
"loss": 0.7658,
"step": 723
},
{
"epoch": 2.15,
"learning_rate": 1.8983662101172217e-06,
"loss": 0.4535,
"step": 724
},
{
"epoch": 2.15,
"learning_rate": 1.8860610975594384e-06,
"loss": 0.7777,
"step": 725
},
{
"epoch": 2.15,
"learning_rate": 1.873786718321476e-06,
"loss": 0.6766,
"step": 726
},
{
"epoch": 2.15,
"learning_rate": 1.8615431935464984e-06,
"loss": 0.4513,
"step": 727
},
{
"epoch": 2.16,
"learning_rate": 1.8493306440731557e-06,
"loss": 0.7229,
"step": 728
},
{
"epoch": 2.16,
"learning_rate": 1.837149190434378e-06,
"loss": 0.324,
"step": 729
},
{
"epoch": 2.16,
"learning_rate": 1.824998952856198e-06,
"loss": 0.3033,
"step": 730
},
{
"epoch": 2.17,
"learning_rate": 1.8128800512565514e-06,
"loss": 0.6436,
"step": 731
},
{
"epoch": 2.17,
"learning_rate": 1.800792605244109e-06,
"loss": 1.0939,
"step": 732
},
{
"epoch": 2.17,
"learning_rate": 1.7887367341170781e-06,
"loss": 0.6184,
"step": 733
},
{
"epoch": 2.17,
"learning_rate": 1.7767125568620442e-06,
"loss": 0.8289,
"step": 734
},
{
"epoch": 2.18,
"learning_rate": 1.7647201921527802e-06,
"loss": 0.3752,
"step": 735
},
{
"epoch": 2.18,
"learning_rate": 1.7527597583490825e-06,
"loss": 0.5272,
"step": 736
},
{
"epoch": 2.18,
"learning_rate": 1.7408313734956074e-06,
"loss": 0.4365,
"step": 737
},
{
"epoch": 2.19,
"learning_rate": 1.7289351553206952e-06,
"loss": 0.5517,
"step": 738
},
{
"epoch": 2.19,
"learning_rate": 1.7170712212352187e-06,
"loss": 0.5039,
"step": 739
},
{
"epoch": 2.19,
"learning_rate": 1.7052396883314154e-06,
"loss": 0.4664,
"step": 740
},
{
"epoch": 2.2,
"learning_rate": 1.6934406733817417e-06,
"loss": 0.681,
"step": 741
},
{
"epoch": 2.2,
"learning_rate": 1.6816742928377072e-06,
"loss": 0.5911,
"step": 742
},
{
"epoch": 2.2,
"learning_rate": 1.6699406628287423e-06,
"loss": 0.6502,
"step": 743
},
{
"epoch": 2.2,
"learning_rate": 1.658239899161036e-06,
"loss": 0.4529,
"step": 744
},
{
"epoch": 2.21,
"learning_rate": 1.6465721173164e-06,
"loss": 0.5244,
"step": 745
},
{
"epoch": 2.21,
"learning_rate": 1.6349374324511347e-06,
"loss": 0.9191,
"step": 746
},
{
"epoch": 2.21,
"learning_rate": 1.6233359593948777e-06,
"loss": 0.3934,
"step": 747
},
{
"epoch": 2.22,
"learning_rate": 1.6117678126494895e-06,
"loss": 0.3773,
"step": 748
},
{
"epoch": 2.22,
"learning_rate": 1.600233106387904e-06,
"loss": 0.8239,
"step": 749
},
{
"epoch": 2.22,
"learning_rate": 1.5887319544530182e-06,
"loss": 0.7353,
"step": 750
},
{
"epoch": 2.23,
"learning_rate": 1.5772644703565564e-06,
"loss": 0.4856,
"step": 751
},
{
"epoch": 2.23,
"learning_rate": 1.5658307672779594e-06,
"loss": 0.1405,
"step": 752
},
{
"epoch": 2.23,
"learning_rate": 1.554430958063259e-06,
"loss": 1.1349,
"step": 753
},
{
"epoch": 2.23,
"learning_rate": 1.5430651552239684e-06,
"loss": 0.5385,
"step": 754
},
{
"epoch": 2.24,
"learning_rate": 1.531733470935976e-06,
"loss": 0.7236,
"step": 755
},
{
"epoch": 2.24,
"learning_rate": 1.5204360170384286e-06,
"loss": 0.5191,
"step": 756
},
{
"epoch": 2.24,
"learning_rate": 1.5091729050326376e-06,
"loss": 0.5902,
"step": 757
},
{
"epoch": 2.25,
"learning_rate": 1.4979442460809684e-06,
"loss": 0.4046,
"step": 758
},
{
"epoch": 2.25,
"learning_rate": 1.4867501510057548e-06,
"loss": 0.8249,
"step": 759
},
{
"epoch": 2.25,
"learning_rate": 1.4755907302881927e-06,
"loss": 0.4916,
"step": 760
},
{
"epoch": 2.25,
"learning_rate": 1.4644660940672628e-06,
"loss": 0.5995,
"step": 761
},
{
"epoch": 2.26,
"learning_rate": 1.4533763521386319e-06,
"loss": 0.5577,
"step": 762
},
{
"epoch": 2.26,
"learning_rate": 1.4423216139535735e-06,
"loss": 0.2862,
"step": 763
},
{
"epoch": 2.26,
"learning_rate": 1.4313019886178942e-06,
"loss": 0.4645,
"step": 764
},
{
"epoch": 2.27,
"learning_rate": 1.420317584890844e-06,
"loss": 0.885,
"step": 765
},
{
"epoch": 2.27,
"learning_rate": 1.4093685111840567e-06,
"loss": 0.6885,
"step": 766
},
{
"epoch": 2.27,
"learning_rate": 1.3984548755604655e-06,
"loss": 0.8162,
"step": 767
},
{
"epoch": 2.28,
"learning_rate": 1.3875767857332512e-06,
"loss": 0.8972,
"step": 768
},
{
"epoch": 2.28,
"learning_rate": 1.3767343490647668e-06,
"loss": 0.5285,
"step": 769
},
{
"epoch": 2.28,
"learning_rate": 1.3659276725654863e-06,
"loss": 0.8215,
"step": 770
},
{
"epoch": 2.28,
"learning_rate": 1.3551568628929434e-06,
"loss": 0.8666,
"step": 771
},
{
"epoch": 2.29,
"learning_rate": 1.3444220263506797e-06,
"loss": 0.3384,
"step": 772
},
{
"epoch": 2.29,
"learning_rate": 1.333723268887201e-06,
"loss": 0.5464,
"step": 773
},
{
"epoch": 2.29,
"learning_rate": 1.3230606960949204e-06,
"loss": 1.0275,
"step": 774
},
{
"epoch": 2.3,
"learning_rate": 1.312434413209131e-06,
"loss": 0.5065,
"step": 775
},
{
"epoch": 2.3,
"learning_rate": 1.301844525106951e-06,
"loss": 0.4504,
"step": 776
},
{
"epoch": 2.3,
"learning_rate": 1.2912911363063048e-06,
"loss": 0.5338,
"step": 777
},
{
"epoch": 2.31,
"learning_rate": 1.2807743509648745e-06,
"loss": 0.9552,
"step": 778
},
{
"epoch": 2.31,
"learning_rate": 1.2702942728790897e-06,
"loss": 0.3828,
"step": 779
},
{
"epoch": 2.31,
"learning_rate": 1.2598510054830888e-06,
"loss": 0.4786,
"step": 780
},
{
"epoch": 2.31,
"learning_rate": 1.2494446518477022e-06,
"loss": 0.4632,
"step": 781
},
{
"epoch": 2.32,
"learning_rate": 1.2390753146794438e-06,
"loss": 0.6768,
"step": 782
},
{
"epoch": 2.32,
"learning_rate": 1.2287430963194807e-06,
"loss": 1.0365,
"step": 783
},
{
"epoch": 2.32,
"learning_rate": 1.218448098742641e-06,
"loss": 0.752,
"step": 784
},
{
"epoch": 2.33,
"learning_rate": 1.2081904235563908e-06,
"loss": 1.0462,
"step": 785
},
{
"epoch": 2.33,
"learning_rate": 1.1979701719998454e-06,
"loss": 0.6615,
"step": 786
},
{
"epoch": 2.33,
"learning_rate": 1.18778744494276e-06,
"loss": 0.6062,
"step": 787
},
{
"epoch": 2.33,
"learning_rate": 1.1776423428845423e-06,
"loss": 0.4378,
"step": 788
},
{
"epoch": 2.34,
"learning_rate": 1.1675349659532514e-06,
"loss": 0.3697,
"step": 789
},
{
"epoch": 2.34,
"learning_rate": 1.1574654139046171e-06,
"loss": 0.8439,
"step": 790
},
{
"epoch": 2.34,
"learning_rate": 1.1474337861210543e-06,
"loss": 0.519,
"step": 791
},
{
"epoch": 2.35,
"learning_rate": 1.1374401816106778e-06,
"loss": 0.7717,
"step": 792
},
{
"epoch": 2.35,
"learning_rate": 1.1274846990063314e-06,
"loss": 0.9027,
"step": 793
},
{
"epoch": 2.35,
"learning_rate": 1.1175674365646067e-06,
"loss": 0.8879,
"step": 794
},
{
"epoch": 2.36,
"learning_rate": 1.1076884921648834e-06,
"loss": 0.2537,
"step": 795
},
{
"epoch": 2.36,
"learning_rate": 1.097847963308351e-06,
"loss": 0.3626,
"step": 796
},
{
"epoch": 2.36,
"learning_rate": 1.0880459471170597e-06,
"loss": 0.6451,
"step": 797
},
{
"epoch": 2.36,
"learning_rate": 1.0782825403329488e-06,
"loss": 0.5883,
"step": 798
},
{
"epoch": 2.37,
"learning_rate": 1.0685578393169054e-06,
"loss": 0.3654,
"step": 799
},
{
"epoch": 2.37,
"learning_rate": 1.0588719400478004e-06,
"loss": 0.7633,
"step": 800
},
{
"epoch": 2.37,
"learning_rate": 1.049224938121548e-06,
"loss": 0.3511,
"step": 801
},
{
"epoch": 2.38,
"learning_rate": 1.0396169287501652e-06,
"loss": 0.7186,
"step": 802
},
{
"epoch": 2.38,
"learning_rate": 1.0300480067608232e-06,
"loss": 0.4396,
"step": 803
},
{
"epoch": 2.38,
"learning_rate": 1.020518266594921e-06,
"loss": 0.8757,
"step": 804
},
{
"epoch": 2.39,
"learning_rate": 1.0110278023071445e-06,
"loss": 0.4438,
"step": 805
},
{
"epoch": 2.39,
"learning_rate": 1.0015767075645472e-06,
"loss": 0.5048,
"step": 806
},
{
"epoch": 2.39,
"learning_rate": 9.921650756456164e-07,
"loss": 0.5697,
"step": 807
},
{
"epoch": 2.39,
"learning_rate": 9.82792999439362e-07,
"loss": 0.6669,
"step": 808
},
{
"epoch": 2.4,
"learning_rate": 9.734605714443906e-07,
"loss": 0.2139,
"step": 809
},
{
"epoch": 2.4,
"learning_rate": 9.641678837679985e-07,
"loss": 0.5889,
"step": 810
},
{
"epoch": 2.4,
"learning_rate": 9.549150281252633e-07,
"loss": 0.628,
"step": 811
},
{
"epoch": 2.41,
"learning_rate": 9.457020958381324e-07,
"loss": 0.3455,
"step": 812
},
{
"epoch": 2.41,
"learning_rate": 9.365291778345303e-07,
"loss": 0.5204,
"step": 813
},
{
"epoch": 2.41,
"learning_rate": 9.273963646474527e-07,
"loss": 0.3506,
"step": 814
},
{
"epoch": 2.41,
"learning_rate": 9.183037464140804e-07,
"loss": 0.4729,
"step": 815
},
{
"epoch": 2.42,
"learning_rate": 9.09251412874882e-07,
"loss": 1.218,
"step": 816
},
{
"epoch": 2.42,
"learning_rate": 9.002394533727382e-07,
"loss": 0.9412,
"step": 817
},
{
"epoch": 2.42,
"learning_rate": 8.912679568520494e-07,
"loss": 0.3594,
"step": 818
},
{
"epoch": 2.43,
"learning_rate": 8.823370118578628e-07,
"loss": 1.2984,
"step": 819
},
{
"epoch": 2.43,
"learning_rate": 8.734467065350022e-07,
"loss": 0.8734,
"step": 820
},
{
"epoch": 2.43,
"learning_rate": 8.645971286271903e-07,
"loss": 0.3182,
"step": 821
},
{
"epoch": 2.44,
"learning_rate": 8.557883654761906e-07,
"loss": 0.7474,
"step": 822
},
{
"epoch": 2.44,
"learning_rate": 8.470205040209362e-07,
"loss": 0.7064,
"step": 823
},
{
"epoch": 2.44,
"learning_rate": 8.382936307966838e-07,
"loss": 0.3532,
"step": 824
},
{
"epoch": 2.44,
"learning_rate": 8.296078319341444e-07,
"loss": 0.4798,
"step": 825
},
{
"epoch": 2.45,
"learning_rate": 8.209631931586499e-07,
"loss": 0.6227,
"step": 826
},
{
"epoch": 2.45,
"learning_rate": 8.123597997892918e-07,
"loss": 0.5941,
"step": 827
},
{
"epoch": 2.45,
"learning_rate": 8.037977367380922e-07,
"loss": 0.481,
"step": 828
},
{
"epoch": 2.46,
"learning_rate": 7.952770885091548e-07,
"loss": 0.6646,
"step": 829
},
{
"epoch": 2.46,
"learning_rate": 7.867979391978398e-07,
"loss": 0.6778,
"step": 830
},
{
"epoch": 2.46,
"learning_rate": 7.783603724899258e-07,
"loss": 0.3959,
"step": 831
},
{
"epoch": 2.47,
"learning_rate": 7.699644716607896e-07,
"loss": 0.4402,
"step": 832
},
{
"epoch": 2.47,
"learning_rate": 7.61610319574585e-07,
"loss": 0.8016,
"step": 833
},
{
"epoch": 2.47,
"learning_rate": 7.532979986834177e-07,
"loss": 0.3912,
"step": 834
},
{
"epoch": 2.47,
"learning_rate": 7.450275910265415e-07,
"loss": 0.5042,
"step": 835
},
{
"epoch": 2.48,
"learning_rate": 7.367991782295392e-07,
"loss": 0.7444,
"step": 836
},
{
"epoch": 2.48,
"learning_rate": 7.286128415035249e-07,
"loss": 0.6005,
"step": 837
},
{
"epoch": 2.48,
"learning_rate": 7.204686616443352e-07,
"loss": 0.9143,
"step": 838
},
{
"epoch": 2.49,
"learning_rate": 7.123667190317396e-07,
"loss": 0.4768,
"step": 839
},
{
"epoch": 2.49,
"learning_rate": 7.043070936286395e-07,
"loss": 0.5258,
"step": 840
},
{
"epoch": 2.49,
"learning_rate": 6.962898649802824e-07,
"loss": 0.81,
"step": 841
},
{
"epoch": 2.49,
"learning_rate": 6.883151122134812e-07,
"loss": 1.5241,
"step": 842
},
{
"epoch": 2.5,
"learning_rate": 6.803829140358237e-07,
"loss": 0.4831,
"step": 843
},
{
"epoch": 2.5,
"learning_rate": 6.724933487349061e-07,
"loss": 0.7074,
"step": 844
},
{
"epoch": 2.5,
"learning_rate": 6.646464941775499e-07,
"loss": 0.2782,
"step": 845
},
{
"epoch": 2.51,
"learning_rate": 6.568424278090446e-07,
"loss": 0.9711,
"step": 846
},
{
"epoch": 2.51,
"learning_rate": 6.490812266523716e-07,
"loss": 0.5261,
"step": 847
},
{
"epoch": 2.51,
"learning_rate": 6.413629673074562e-07,
"loss": 0.5923,
"step": 848
},
{
"epoch": 2.52,
"learning_rate": 6.336877259504004e-07,
"loss": 0.2051,
"step": 849
},
{
"epoch": 2.52,
"learning_rate": 6.260555783327366e-07,
"loss": 0.3705,
"step": 850
},
{
"epoch": 2.52,
"learning_rate": 6.184665997806832e-07,
"loss": 0.6495,
"step": 851
},
{
"epoch": 2.52,
"learning_rate": 6.109208651943921e-07,
"loss": 0.443,
"step": 852
},
{
"epoch": 2.53,
"learning_rate": 6.034184490472195e-07,
"loss": 0.5884,
"step": 853
},
{
"epoch": 2.53,
"learning_rate": 5.959594253849821e-07,
"loss": 0.3667,
"step": 854
},
{
"epoch": 2.53,
"learning_rate": 5.885438678252342e-07,
"loss": 0.4444,
"step": 855
},
{
"epoch": 2.54,
"learning_rate": 5.811718495565327e-07,
"loss": 0.5174,
"step": 856
},
{
"epoch": 2.54,
"learning_rate": 5.738434433377244e-07,
"loss": 0.5387,
"step": 857
},
{
"epoch": 2.54,
"learning_rate": 5.665587214972173e-07,
"loss": 0.3972,
"step": 858
},
{
"epoch": 2.55,
"learning_rate": 5.593177559322776e-07,
"loss": 0.4253,
"step": 859
},
{
"epoch": 2.55,
"learning_rate": 5.521206181083111e-07,
"loss": 0.6145,
"step": 860
},
{
"epoch": 2.55,
"learning_rate": 5.449673790581611e-07,
"loss": 0.2021,
"step": 861
},
{
"epoch": 2.55,
"learning_rate": 5.378581093814112e-07,
"loss": 0.7266,
"step": 862
},
{
"epoch": 2.56,
"learning_rate": 5.307928792436812e-07,
"loss": 0.5027,
"step": 863
},
{
"epoch": 2.56,
"learning_rate": 5.237717583759421e-07,
"loss": 0.6208,
"step": 864
},
{
"epoch": 2.56,
"learning_rate": 5.167948160738206e-07,
"loss": 0.9094,
"step": 865
},
{
"epoch": 2.57,
"learning_rate": 5.098621211969224e-07,
"loss": 0.3827,
"step": 866
},
{
"epoch": 2.57,
"learning_rate": 5.029737421681446e-07,
"loss": 0.472,
"step": 867
},
{
"epoch": 2.57,
"learning_rate": 4.961297469730097e-07,
"loss": 0.8333,
"step": 868
},
{
"epoch": 2.57,
"learning_rate": 4.893302031589864e-07,
"loss": 0.2821,
"step": 869
},
{
"epoch": 2.58,
"learning_rate": 4.825751778348259e-07,
"loss": 0.533,
"step": 870
},
{
"epoch": 2.58,
"learning_rate": 4.758647376699033e-07,
"loss": 0.6419,
"step": 871
},
{
"epoch": 2.58,
"learning_rate": 4.691989488935511e-07,
"loss": 0.7215,
"step": 872
},
{
"epoch": 2.59,
"learning_rate": 4.625778772944156e-07,
"loss": 0.3634,
"step": 873
},
{
"epoch": 2.59,
"learning_rate": 4.5600158821979933e-07,
"loss": 0.6101,
"step": 874
},
{
"epoch": 2.59,
"learning_rate": 4.494701465750217e-07,
"loss": 0.65,
"step": 875
},
{
"epoch": 2.6,
"learning_rate": 4.4298361682277355e-07,
"loss": 0.3687,
"step": 876
},
{
"epoch": 2.6,
"learning_rate": 4.3654206298248625e-07,
"loss": 0.6719,
"step": 877
},
{
"epoch": 2.6,
"learning_rate": 4.301455486296946e-07,
"loss": 0.4012,
"step": 878
},
{
"epoch": 2.6,
"learning_rate": 4.237941368954124e-07,
"loss": 0.515,
"step": 879
},
{
"epoch": 2.61,
"learning_rate": 4.1748789046551055e-07,
"loss": 0.7022,
"step": 880
},
{
"epoch": 2.61,
"learning_rate": 4.112268715800943e-07,
"loss": 0.6836,
"step": 881
},
{
"epoch": 2.61,
"learning_rate": 4.0501114203289395e-07,
"loss": 0.5604,
"step": 882
},
{
"epoch": 2.62,
"learning_rate": 3.9884076317064813e-07,
"loss": 1.1093,
"step": 883
},
{
"epoch": 2.62,
"learning_rate": 3.9271579589250817e-07,
"loss": 0.2598,
"step": 884
},
{
"epoch": 2.62,
"learning_rate": 3.866363006494256e-07,
"loss": 0.9246,
"step": 885
},
{
"epoch": 2.63,
"learning_rate": 3.8060233744356634e-07,
"loss": 0.9875,
"step": 886
},
{
"epoch": 2.63,
"learning_rate": 3.7461396582771035e-07,
"loss": 0.7367,
"step": 887
},
{
"epoch": 2.63,
"learning_rate": 3.6867124490466697e-07,
"loss": 0.4154,
"step": 888
},
{
"epoch": 2.63,
"learning_rate": 3.627742333266937e-07,
"loss": 0.4498,
"step": 889
},
{
"epoch": 2.64,
"learning_rate": 3.569229892949133e-07,
"loss": 0.3582,
"step": 890
},
{
"epoch": 2.64,
"learning_rate": 3.511175705587433e-07,
"loss": 1.0801,
"step": 891
},
{
"epoch": 2.64,
"learning_rate": 3.4535803441532125e-07,
"loss": 0.2941,
"step": 892
},
{
"epoch": 2.65,
"learning_rate": 3.396444377089453e-07,
"loss": 0.7146,
"step": 893
},
{
"epoch": 2.65,
"learning_rate": 3.3397683683050685e-07,
"loss": 0.7243,
"step": 894
},
{
"epoch": 2.65,
"learning_rate": 3.283552877169399e-07,
"loss": 0.6498,
"step": 895
},
{
"epoch": 2.65,
"learning_rate": 3.227798458506637e-07,
"loss": 0.5064,
"step": 896
},
{
"epoch": 2.66,
"learning_rate": 3.172505662590386e-07,
"loss": 1.1218,
"step": 897
},
{
"epoch": 2.66,
"learning_rate": 3.1176750351382235e-07,
"loss": 0.5226,
"step": 898
},
{
"epoch": 2.66,
"learning_rate": 3.0633071173062966e-07,
"loss": 0.6652,
"step": 899
},
{
"epoch": 2.67,
"learning_rate": 3.0094024456840176e-07,
"loss": 0.9665,
"step": 900
},
{
"epoch": 2.67,
"learning_rate": 2.9559615522887275e-07,
"loss": 0.4859,
"step": 901
},
{
"epoch": 2.67,
"learning_rate": 2.9029849645604735e-07,
"loss": 0.5401,
"step": 902
},
{
"epoch": 2.68,
"learning_rate": 2.850473205356774e-07,
"loss": 1.1925,
"step": 903
},
{
"epoch": 2.68,
"learning_rate": 2.798426792947517e-07,
"loss": 0.5312,
"step": 904
},
{
"epoch": 2.68,
"learning_rate": 2.746846241009765e-07,
"loss": 0.3988,
"step": 905
},
{
"epoch": 2.68,
"learning_rate": 2.6957320586227354e-07,
"loss": 0.1643,
"step": 906
},
{
"epoch": 2.69,
"learning_rate": 2.6450847502627883e-07,
"loss": 0.3646,
"step": 907
},
{
"epoch": 2.69,
"learning_rate": 2.594904815798399e-07,
"loss": 0.4722,
"step": 908
},
{
"epoch": 2.69,
"learning_rate": 2.5451927504852757e-07,
"loss": 0.5324,
"step": 909
},
{
"epoch": 2.7,
"learning_rate": 2.49594904496141e-07,
"loss": 0.4732,
"step": 910
},
{
"epoch": 2.7,
"learning_rate": 2.447174185242324e-07,
"loss": 0.4848,
"step": 911
},
{
"epoch": 2.7,
"learning_rate": 2.3988686527161686e-07,
"loss": 0.3708,
"step": 912
},
{
"epoch": 2.71,
"learning_rate": 2.351032924139063e-07,
"loss": 0.2401,
"step": 913
},
{
"epoch": 2.71,
"learning_rate": 2.3036674716303277e-07,
"loss": 0.4895,
"step": 914
},
{
"epoch": 2.71,
"learning_rate": 2.2567727626678527e-07,
"loss": 0.2889,
"step": 915
},
{
"epoch": 2.71,
"learning_rate": 2.210349260083494e-07,
"loss": 0.1817,
"step": 916
},
{
"epoch": 2.72,
"learning_rate": 2.1643974220584729e-07,
"loss": 0.388,
"step": 917
},
{
"epoch": 2.72,
"learning_rate": 2.1189177021188888e-07,
"loss": 0.3138,
"step": 918
},
{
"epoch": 2.72,
"learning_rate": 2.0739105491312028e-07,
"loss": 0.2662,
"step": 919
},
{
"epoch": 2.73,
"learning_rate": 2.0293764072978618e-07,
"loss": 0.8694,
"step": 920
},
{
"epoch": 2.73,
"learning_rate": 1.9853157161528468e-07,
"loss": 0.4989,
"step": 921
},
{
"epoch": 2.73,
"learning_rate": 1.9417289105574054e-07,
"loss": 1.0212,
"step": 922
},
{
"epoch": 2.73,
"learning_rate": 1.8986164206957037e-07,
"loss": 1.2547,
"step": 923
},
{
"epoch": 2.74,
"learning_rate": 1.8559786720706185e-07,
"loss": 1.2131,
"step": 924
},
{
"epoch": 2.74,
"learning_rate": 1.8138160854995145e-07,
"loss": 1.0261,
"step": 925
},
{
"epoch": 2.74,
"learning_rate": 1.7721290771100964e-07,
"loss": 0.8173,
"step": 926
},
{
"epoch": 2.75,
"learning_rate": 1.7309180583363062e-07,
"loss": 0.7826,
"step": 927
},
{
"epoch": 2.75,
"learning_rate": 1.690183435914261e-07,
"loss": 0.4913,
"step": 928
},
{
"epoch": 2.75,
"learning_rate": 1.6499256118782503e-07,
"loss": 0.4313,
"step": 929
},
{
"epoch": 2.76,
"learning_rate": 1.6101449835567273e-07,
"loss": 0.4954,
"step": 930
},
{
"epoch": 2.76,
"learning_rate": 1.5708419435684463e-07,
"loss": 0.9359,
"step": 931
},
{
"epoch": 2.76,
"learning_rate": 1.532016879818532e-07,
"loss": 0.4006,
"step": 932
},
{
"epoch": 2.76,
"learning_rate": 1.4936701754947104e-07,
"loss": 0.5435,
"step": 933
},
{
"epoch": 2.77,
"learning_rate": 1.4558022090634504e-07,
"loss": 0.8151,
"step": 934
},
{
"epoch": 2.77,
"learning_rate": 1.4184133542663014e-07,
"loss": 0.6551,
"step": 935
},
{
"epoch": 2.77,
"learning_rate": 1.3815039801161723e-07,
"loss": 0.7367,
"step": 936
},
{
"epoch": 2.78,
"learning_rate": 1.3450744508936687e-07,
"loss": 0.4398,
"step": 937
},
{
"epoch": 2.78,
"learning_rate": 1.3091251261435568e-07,
"loss": 0.5791,
"step": 938
},
{
"epoch": 2.78,
"learning_rate": 1.2736563606711384e-07,
"loss": 0.4408,
"step": 939
},
{
"epoch": 2.79,
"learning_rate": 1.2386685045388313e-07,
"loss": 0.6381,
"step": 940
},
{
"epoch": 2.79,
"learning_rate": 1.2041619030626283e-07,
"loss": 0.2984,
"step": 941
},
{
"epoch": 2.79,
"learning_rate": 1.1701368968087711e-07,
"loss": 0.5103,
"step": 942
},
{
"epoch": 2.79,
"learning_rate": 1.136593821590326e-07,
"loss": 0.5749,
"step": 943
},
{
"epoch": 2.8,
"learning_rate": 1.1035330084639084e-07,
"loss": 0.6796,
"step": 944
},
{
"epoch": 2.8,
"learning_rate": 1.0709547837263967e-07,
"loss": 1.1137,
"step": 945
},
{
"epoch": 2.8,
"learning_rate": 1.038859468911707e-07,
"loss": 0.3819,
"step": 946
},
{
"epoch": 2.81,
"learning_rate": 1.007247380787657e-07,
"loss": 0.6517,
"step": 947
},
{
"epoch": 2.81,
"learning_rate": 9.761188313527792e-08,
"loss": 0.25,
"step": 948
},
{
"epoch": 2.81,
"learning_rate": 9.454741278333013e-08,
"loss": 0.4187,
"step": 949
},
{
"epoch": 2.81,
"learning_rate": 9.153135726800599e-08,
"loss": 0.6917,
"step": 950
},
{
"epoch": 2.82,
"learning_rate": 8.856374635655696e-08,
"loss": 0.677,
"step": 951
},
{
"epoch": 2.82,
"learning_rate": 8.564460933810414e-08,
"loss": 0.8416,
"step": 952
},
{
"epoch": 2.82,
"learning_rate": 8.277397502335194e-08,
"loss": 0.4108,
"step": 953
},
{
"epoch": 2.83,
"learning_rate": 7.995187174430152e-08,
"loss": 0.7571,
"step": 954
},
{
"epoch": 2.83,
"learning_rate": 7.717832735397335e-08,
"loss": 1.3161,
"step": 955
},
{
"epoch": 2.83,
"learning_rate": 7.445336922613067e-08,
"loss": 0.436,
"step": 956
},
{
"epoch": 2.84,
"learning_rate": 7.177702425500977e-08,
"loss": 0.5057,
"step": 957
},
{
"epoch": 2.84,
"learning_rate": 6.914931885505626e-08,
"loss": 0.4623,
"step": 958
},
{
"epoch": 2.84,
"learning_rate": 6.657027896065982e-08,
"loss": 0.6187,
"step": 959
},
{
"epoch": 2.84,
"learning_rate": 6.403993002590425e-08,
"loss": 0.3782,
"step": 960
},
{
"epoch": 2.85,
"learning_rate": 6.15582970243117e-08,
"loss": 0.7128,
"step": 961
},
{
"epoch": 2.85,
"learning_rate": 5.9125404448597825e-08,
"loss": 0.5324,
"step": 962
},
{
"epoch": 2.85,
"learning_rate": 5.674127631043025e-08,
"loss": 0.5388,
"step": 963
},
{
"epoch": 2.86,
"learning_rate": 5.440593614019107e-08,
"loss": 0.408,
"step": 964
},
{
"epoch": 2.86,
"learning_rate": 5.2119406986745336e-08,
"loss": 0.2904,
"step": 965
},
{
"epoch": 2.86,
"learning_rate": 4.988171141721232e-08,
"loss": 0.5742,
"step": 966
},
{
"epoch": 2.87,
"learning_rate": 4.769287151674407e-08,
"loss": 0.504,
"step": 967
},
{
"epoch": 2.87,
"learning_rate": 4.5552908888306654e-08,
"loss": 0.2633,
"step": 968
},
{
"epoch": 2.87,
"learning_rate": 4.346184465246761e-08,
"loss": 0.652,
"step": 969
},
{
"epoch": 2.87,
"learning_rate": 4.1419699447186045e-08,
"loss": 0.4014,
"step": 970
},
{
"epoch": 2.88,
"learning_rate": 3.9426493427611177e-08,
"loss": 0.7626,
"step": 971
},
{
"epoch": 2.88,
"learning_rate": 3.748224626588137e-08,
"loss": 0.8499,
"step": 972
},
{
"epoch": 2.88,
"learning_rate": 3.558697715093207e-08,
"loss": 0.4229,
"step": 973
},
{
"epoch": 2.89,
"learning_rate": 3.374070478830316e-08,
"loss": 0.4941,
"step": 974
},
{
"epoch": 2.89,
"learning_rate": 3.194344739995803e-08,
"loss": 0.5811,
"step": 975
},
{
"epoch": 2.89,
"learning_rate": 3.019522272410202e-08,
"loss": 0.2732,
"step": 976
},
{
"epoch": 2.89,
"learning_rate": 2.8496048015005385e-08,
"loss": 0.4816,
"step": 977
},
{
"epoch": 2.9,
"learning_rate": 2.684594004283836e-08,
"loss": 0.5769,
"step": 978
},
{
"epoch": 2.9,
"learning_rate": 2.5244915093499134e-08,
"loss": 0.502,
"step": 979
},
{
"epoch": 2.9,
"learning_rate": 2.3692988968458398e-08,
"loss": 0.5027,
"step": 980
},
{
"epoch": 2.91,
"learning_rate": 2.219017698460002e-08,
"loss": 0.8383,
"step": 981
},
{
"epoch": 2.91,
"learning_rate": 2.0736493974071736e-08,
"loss": 0.608,
"step": 982
},
{
"epoch": 2.91,
"learning_rate": 1.9331954284137476e-08,
"loss": 0.6952,
"step": 983
},
{
"epoch": 2.92,
"learning_rate": 1.7976571777038044e-08,
"loss": 0.4672,
"step": 984
},
{
"epoch": 2.92,
"learning_rate": 1.6670359829850657e-08,
"loss": 0.3528,
"step": 985
},
{
"epoch": 2.92,
"learning_rate": 1.541333133436018e-08,
"loss": 0.2532,
"step": 986
},
{
"epoch": 2.92,
"learning_rate": 1.4205498696930332e-08,
"loss": 0.5655,
"step": 987
},
{
"epoch": 2.93,
"learning_rate": 1.3046873838381546e-08,
"loss": 0.7731,
"step": 988
},
{
"epoch": 2.93,
"learning_rate": 1.1937468193873869e-08,
"loss": 0.75,
"step": 989
},
{
"epoch": 2.93,
"learning_rate": 1.0877292712792586e-08,
"loss": 0.6898,
"step": 990
},
{
"epoch": 2.94,
"learning_rate": 9.866357858642206e-09,
"loss": 0.478,
"step": 991
},
{
"epoch": 2.94,
"learning_rate": 8.904673608940983e-09,
"loss": 0.467,
"step": 992
},
{
"epoch": 2.94,
"learning_rate": 7.992249455124889e-09,
"loss": 0.8591,
"step": 993
},
{
"epoch": 2.95,
"learning_rate": 7.129094402451575e-09,
"loss": 0.5595,
"step": 994
},
{
"epoch": 2.95,
"learning_rate": 6.315216969912663e-09,
"loss": 0.5404,
"step": 995
},
{
"epoch": 2.95,
"learning_rate": 5.5506251901504825e-09,
"loss": 0.6362,
"step": 996
},
{
"epoch": 2.95,
"learning_rate": 4.835326609376468e-09,
"loss": 0.8166,
"step": 997
},
{
"epoch": 2.96,
"learning_rate": 4.169328287299545e-09,
"loss": 0.5938,
"step": 998
},
{
"epoch": 2.96,
"learning_rate": 3.5526367970539765e-09,
"loss": 0.6241,
"step": 999
},
{
"epoch": 2.96,
"learning_rate": 2.9852582251355124e-09,
"loss": 0.4188,
"step": 1000
},
{
"epoch": 2.97,
"learning_rate": 2.4671981713420003e-09,
"loss": 0.5616,
"step": 1001
},
{
"epoch": 2.97,
"learning_rate": 1.9984617487173174e-09,
"loss": 0.6014,
"step": 1002
},
{
"epoch": 2.97,
"learning_rate": 1.5790535835003006e-09,
"loss": 0.8896,
"step": 1003
},
{
"epoch": 2.97,
"learning_rate": 1.2089778150797816e-09,
"loss": 0.8489,
"step": 1004
},
{
"epoch": 2.98,
"learning_rate": 8.88238095955174e-10,
"loss": 0.6794,
"step": 1005
},
{
"epoch": 2.98,
"learning_rate": 6.168375916970615e-10,
"loss": 0.506,
"step": 1006
},
{
"epoch": 2.98,
"learning_rate": 3.9477898091944135e-10,
"loss": 0.6728,
"step": 1007
},
{
"epoch": 2.99,
"learning_rate": 2.2206445525085886e-10,
"loss": 0.7106,
"step": 1008
},
{
"epoch": 2.99,
"learning_rate": 9.869571931442334e-11,
"loss": 0.8183,
"step": 1009
},
{
"epoch": 2.99,
"learning_rate": 2.467399070893439e-11,
"loss": 0.752,
"step": 1010
},
{
"epoch": 3.0,
"learning_rate": 0.0,
"loss": 0.3006,
"step": 1011
},
{
"epoch": 3.0,
"step": 1011,
"total_flos": 5.146073678262095e+20,
"train_loss": 0.6974123622319347,
"train_runtime": 2986.774,
"train_samples_per_second": 2.712,
"train_steps_per_second": 0.338
}
],
"logging_steps": 1.0,
"max_steps": 1011,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 5.146073678262095e+20,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}