grammar-synthesis-large / trainer_state.json
pszemraj's picture
load model from drive and convert
4a407cf
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4643978008769782,
"global_step": 652,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.517241379310345e-06,
"loss": 0.4754,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.103448275862069e-05,
"loss": 0.497,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 1.6551724137931037e-05,
"loss": 0.5361,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 2.206896551724138e-05,
"loss": 0.4945,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 2.7586206896551727e-05,
"loss": 0.4847,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 3.310344827586207e-05,
"loss": 0.4741,
"step": 12
},
{
"epoch": 0.01,
"learning_rate": 3.862068965517242e-05,
"loss": 0.4742,
"step": 14
},
{
"epoch": 0.01,
"learning_rate": 4.413793103448276e-05,
"loss": 0.4544,
"step": 16
},
{
"epoch": 0.01,
"learning_rate": 4.9655172413793107e-05,
"loss": 0.4779,
"step": 18
},
{
"epoch": 0.01,
"learning_rate": 5.517241379310345e-05,
"loss": 0.4374,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 6.068965517241379e-05,
"loss": 0.5045,
"step": 22
},
{
"epoch": 0.02,
"learning_rate": 6.620689655172415e-05,
"loss": 0.46,
"step": 24
},
{
"epoch": 0.02,
"learning_rate": 7.17241379310345e-05,
"loss": 0.4489,
"step": 26
},
{
"epoch": 0.02,
"learning_rate": 7.724137931034484e-05,
"loss": 0.4535,
"step": 28
},
{
"epoch": 0.02,
"learning_rate": 7.999989544228434e-05,
"loss": 0.4802,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 7.999905898383868e-05,
"loss": 0.5038,
"step": 32
},
{
"epoch": 0.02,
"learning_rate": 7.999738608443897e-05,
"loss": 0.4804,
"step": 34
},
{
"epoch": 0.03,
"learning_rate": 7.999487677906807e-05,
"loss": 0.4752,
"step": 36
},
{
"epoch": 0.03,
"learning_rate": 7.999153112019939e-05,
"loss": 0.4681,
"step": 38
},
{
"epoch": 0.03,
"learning_rate": 7.998734917779568e-05,
"loss": 0.456,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 7.998233103930771e-05,
"loss": 0.4724,
"step": 42
},
{
"epoch": 0.03,
"learning_rate": 7.99764768096724e-05,
"loss": 0.4514,
"step": 44
},
{
"epoch": 0.03,
"learning_rate": 7.99697866113105e-05,
"loss": 0.4601,
"step": 46
},
{
"epoch": 0.03,
"learning_rate": 7.996226058412426e-05,
"loss": 0.4754,
"step": 48
},
{
"epoch": 0.04,
"learning_rate": 7.995389888549428e-05,
"loss": 0.4395,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 7.994470169027636e-05,
"loss": 0.4795,
"step": 52
},
{
"epoch": 0.04,
"learning_rate": 7.993466919079781e-05,
"loss": 0.4506,
"step": 54
},
{
"epoch": 0.04,
"learning_rate": 7.992380159685338e-05,
"loss": 0.5032,
"step": 56
},
{
"epoch": 0.04,
"learning_rate": 7.991209913570094e-05,
"loss": 0.4629,
"step": 58
},
{
"epoch": 0.04,
"learning_rate": 7.989956205205669e-05,
"loss": 0.4653,
"step": 60
},
{
"epoch": 0.04,
"learning_rate": 7.988619060809005e-05,
"loss": 0.4462,
"step": 62
},
{
"epoch": 0.05,
"learning_rate": 7.98719850834182e-05,
"loss": 0.434,
"step": 64
},
{
"epoch": 0.05,
"learning_rate": 7.985694577510017e-05,
"loss": 0.4845,
"step": 66
},
{
"epoch": 0.05,
"learning_rate": 7.98410729976307e-05,
"loss": 0.4586,
"step": 68
},
{
"epoch": 0.05,
"learning_rate": 7.982436708293364e-05,
"loss": 0.4515,
"step": 70
},
{
"epoch": 0.05,
"learning_rate": 7.980682838035498e-05,
"loss": 0.4621,
"step": 72
},
{
"epoch": 0.05,
"learning_rate": 7.978845725665557e-05,
"loss": 0.4524,
"step": 74
},
{
"epoch": 0.05,
"learning_rate": 7.976925409600346e-05,
"loss": 0.4458,
"step": 76
},
{
"epoch": 0.06,
"learning_rate": 7.974921929996585e-05,
"loss": 0.4992,
"step": 78
},
{
"epoch": 0.06,
"learning_rate": 7.97283532875007e-05,
"loss": 0.4931,
"step": 80
},
{
"epoch": 0.06,
"learning_rate": 7.970665649494793e-05,
"loss": 0.469,
"step": 82
},
{
"epoch": 0.06,
"learning_rate": 7.968412937602039e-05,
"loss": 0.5017,
"step": 84
},
{
"epoch": 0.06,
"learning_rate": 7.966077240179428e-05,
"loss": 0.4753,
"step": 86
},
{
"epoch": 0.06,
"learning_rate": 7.963658606069931e-05,
"loss": 0.4618,
"step": 88
},
{
"epoch": 0.06,
"learning_rate": 7.961157085850858e-05,
"loss": 0.4356,
"step": 90
},
{
"epoch": 0.07,
"learning_rate": 7.958572731832784e-05,
"loss": 0.5153,
"step": 92
},
{
"epoch": 0.07,
"learning_rate": 7.955905598058471e-05,
"loss": 0.4933,
"step": 94
},
{
"epoch": 0.07,
"learning_rate": 7.95315574030173e-05,
"loss": 0.4749,
"step": 96
},
{
"epoch": 0.07,
"learning_rate": 7.950323216066252e-05,
"loss": 0.4162,
"step": 98
},
{
"epoch": 0.07,
"learning_rate": 7.947408084584417e-05,
"loss": 0.4894,
"step": 100
},
{
"epoch": 0.07,
"learning_rate": 7.944410406816039e-05,
"loss": 0.4281,
"step": 102
},
{
"epoch": 0.07,
"learning_rate": 7.941330245447103e-05,
"loss": 0.4245,
"step": 104
},
{
"epoch": 0.08,
"learning_rate": 7.938167664888458e-05,
"loss": 0.4468,
"step": 106
},
{
"epoch": 0.08,
"learning_rate": 7.934922731274452e-05,
"loss": 0.4765,
"step": 108
},
{
"epoch": 0.08,
"learning_rate": 7.931595512461566e-05,
"loss": 0.4411,
"step": 110
},
{
"epoch": 0.08,
"learning_rate": 7.928186078026991e-05,
"loss": 0.4727,
"step": 112
},
{
"epoch": 0.08,
"learning_rate": 7.924694499267169e-05,
"loss": 0.4123,
"step": 114
},
{
"epoch": 0.08,
"learning_rate": 7.921120849196303e-05,
"loss": 0.4703,
"step": 116
},
{
"epoch": 0.08,
"learning_rate": 7.917465202544833e-05,
"loss": 0.4674,
"step": 118
},
{
"epoch": 0.09,
"learning_rate": 7.913727635757873e-05,
"loss": 0.4812,
"step": 120
},
{
"epoch": 0.09,
"learning_rate": 7.909908226993609e-05,
"loss": 0.4304,
"step": 122
},
{
"epoch": 0.09,
"learning_rate": 7.906007056121667e-05,
"loss": 0.4449,
"step": 124
},
{
"epoch": 0.09,
"learning_rate": 7.902024204721445e-05,
"loss": 0.4433,
"step": 126
},
{
"epoch": 0.09,
"learning_rate": 7.897959756080402e-05,
"loss": 0.4602,
"step": 128
},
{
"epoch": 0.09,
"learning_rate": 7.89381379519232e-05,
"loss": 0.4698,
"step": 130
},
{
"epoch": 0.09,
"learning_rate": 7.889586408755526e-05,
"loss": 0.3831,
"step": 132
},
{
"epoch": 0.1,
"learning_rate": 7.885277685171078e-05,
"loss": 0.4394,
"step": 134
},
{
"epoch": 0.1,
"learning_rate": 7.880887714540917e-05,
"loss": 0.4302,
"step": 136
},
{
"epoch": 0.1,
"learning_rate": 7.876416588665984e-05,
"loss": 0.4079,
"step": 138
},
{
"epoch": 0.1,
"learning_rate": 7.871864401044297e-05,
"loss": 0.4639,
"step": 140
},
{
"epoch": 0.1,
"learning_rate": 7.867231246869002e-05,
"loss": 0.4498,
"step": 142
},
{
"epoch": 0.1,
"learning_rate": 7.862517223026373e-05,
"loss": 0.4289,
"step": 144
},
{
"epoch": 0.1,
"learning_rate": 7.857722428093797e-05,
"loss": 0.4515,
"step": 146
},
{
"epoch": 0.11,
"learning_rate": 7.8528469623377e-05,
"loss": 0.4322,
"step": 148
},
{
"epoch": 0.11,
"learning_rate": 7.847890927711464e-05,
"loss": 0.434,
"step": 150
},
{
"epoch": 0.11,
"learning_rate": 7.842854427853285e-05,
"loss": 0.4431,
"step": 152
},
{
"epoch": 0.11,
"learning_rate": 7.83773756808401e-05,
"loss": 0.4089,
"step": 154
},
{
"epoch": 0.11,
"learning_rate": 7.83254045540493e-05,
"loss": 0.488,
"step": 156
},
{
"epoch": 0.11,
"learning_rate": 7.827263198495554e-05,
"loss": 0.4526,
"step": 158
},
{
"epoch": 0.11,
"learning_rate": 7.821905907711319e-05,
"loss": 0.4184,
"step": 160
},
{
"epoch": 0.12,
"learning_rate": 7.816468695081297e-05,
"loss": 0.4176,
"step": 162
},
{
"epoch": 0.12,
"learning_rate": 7.810951674305845e-05,
"loss": 0.4562,
"step": 164
},
{
"epoch": 0.12,
"learning_rate": 7.805354960754233e-05,
"loss": 0.4403,
"step": 166
},
{
"epoch": 0.12,
"learning_rate": 7.799678671462222e-05,
"loss": 0.4238,
"step": 168
},
{
"epoch": 0.12,
"learning_rate": 7.793922925129626e-05,
"loss": 0.4911,
"step": 170
},
{
"epoch": 0.12,
"learning_rate": 7.788087842117826e-05,
"loss": 0.4704,
"step": 172
},
{
"epoch": 0.12,
"learning_rate": 7.782173544447253e-05,
"loss": 0.4355,
"step": 174
},
{
"epoch": 0.13,
"learning_rate": 7.776180155794836e-05,
"loss": 0.4341,
"step": 176
},
{
"epoch": 0.13,
"learning_rate": 7.770107801491414e-05,
"loss": 0.4198,
"step": 178
},
{
"epoch": 0.13,
"learning_rate": 7.763956608519123e-05,
"loss": 0.4756,
"step": 180
},
{
"epoch": 0.13,
"learning_rate": 7.757726705508732e-05,
"loss": 0.4315,
"step": 182
},
{
"epoch": 0.13,
"learning_rate": 7.751418222736954e-05,
"loss": 0.4167,
"step": 184
},
{
"epoch": 0.13,
"learning_rate": 7.745031292123727e-05,
"loss": 0.4316,
"step": 186
},
{
"epoch": 0.13,
"learning_rate": 7.738566047229454e-05,
"loss": 0.4103,
"step": 188
},
{
"epoch": 0.14,
"learning_rate": 7.732022623252203e-05,
"loss": 0.4527,
"step": 190
},
{
"epoch": 0.14,
"learning_rate": 7.725401157024887e-05,
"loss": 0.4587,
"step": 192
},
{
"epoch": 0.14,
"learning_rate": 7.718701787012406e-05,
"loss": 0.457,
"step": 194
},
{
"epoch": 0.14,
"learning_rate": 7.711924653308737e-05,
"loss": 0.4463,
"step": 196
},
{
"epoch": 0.14,
"learning_rate": 7.705069897634022e-05,
"loss": 0.3972,
"step": 198
},
{
"epoch": 0.14,
"learning_rate": 7.69813766333159e-05,
"loss": 0.4375,
"step": 200
},
{
"epoch": 0.14,
"learning_rate": 7.691128095364969e-05,
"loss": 0.4472,
"step": 202
},
{
"epoch": 0.15,
"learning_rate": 7.684041340314852e-05,
"loss": 0.4198,
"step": 204
},
{
"epoch": 0.15,
"learning_rate": 7.676877546376028e-05,
"loss": 0.4365,
"step": 206
},
{
"epoch": 0.15,
"learning_rate": 7.669636863354287e-05,
"loss": 0.4393,
"step": 208
},
{
"epoch": 0.15,
"learning_rate": 7.662319442663286e-05,
"loss": 0.373,
"step": 210
},
{
"epoch": 0.15,
"learning_rate": 7.654925437321386e-05,
"loss": 0.4099,
"step": 212
},
{
"epoch": 0.15,
"learning_rate": 7.647455001948443e-05,
"loss": 0.4473,
"step": 214
},
{
"epoch": 0.15,
"learning_rate": 7.639908292762585e-05,
"loss": 0.4578,
"step": 216
},
{
"epoch": 0.16,
"learning_rate": 7.632285467576943e-05,
"loss": 0.4263,
"step": 218
},
{
"epoch": 0.16,
"learning_rate": 7.624586685796343e-05,
"loss": 0.402,
"step": 220
},
{
"epoch": 0.16,
"learning_rate": 7.616812108413986e-05,
"loss": 0.4347,
"step": 222
},
{
"epoch": 0.16,
"learning_rate": 7.608961898008066e-05,
"loss": 0.45,
"step": 224
},
{
"epoch": 0.16,
"learning_rate": 7.601036218738382e-05,
"loss": 0.4445,
"step": 226
},
{
"epoch": 0.16,
"learning_rate": 7.593035236342903e-05,
"loss": 0.421,
"step": 228
},
{
"epoch": 0.16,
"learning_rate": 7.5849591181343e-05,
"loss": 0.4048,
"step": 230
},
{
"epoch": 0.17,
"learning_rate": 7.576808032996444e-05,
"loss": 0.4372,
"step": 232
},
{
"epoch": 0.17,
"learning_rate": 7.568582151380881e-05,
"loss": 0.3985,
"step": 234
},
{
"epoch": 0.17,
"learning_rate": 7.560281645303266e-05,
"loss": 0.4387,
"step": 236
},
{
"epoch": 0.17,
"learning_rate": 7.551906688339762e-05,
"loss": 0.3997,
"step": 238
},
{
"epoch": 0.17,
"learning_rate": 7.543457455623413e-05,
"loss": 0.3969,
"step": 240
},
{
"epoch": 0.17,
"learning_rate": 7.534934123840486e-05,
"loss": 0.4212,
"step": 242
},
{
"epoch": 0.17,
"learning_rate": 7.526336871226763e-05,
"loss": 0.4026,
"step": 244
},
{
"epoch": 0.18,
"learning_rate": 7.51766587756383e-05,
"loss": 0.424,
"step": 246
},
{
"epoch": 0.18,
"learning_rate": 7.50892132417531e-05,
"loss": 0.4037,
"step": 248
},
{
"epoch": 0.18,
"learning_rate": 7.500103393923068e-05,
"loss": 0.395,
"step": 250
},
{
"epoch": 0.18,
"learning_rate": 7.491212271203391e-05,
"loss": 0.385,
"step": 252
},
{
"epoch": 0.18,
"learning_rate": 7.482248141943133e-05,
"loss": 0.4292,
"step": 254
},
{
"epoch": 0.18,
"learning_rate": 7.473211193595825e-05,
"loss": 0.4002,
"step": 256
},
{
"epoch": 0.18,
"learning_rate": 7.464101615137756e-05,
"loss": 0.4208,
"step": 258
},
{
"epoch": 0.19,
"learning_rate": 7.454919597064017e-05,
"loss": 0.3817,
"step": 260
},
{
"epoch": 0.19,
"learning_rate": 7.445665331384526e-05,
"loss": 0.3668,
"step": 262
},
{
"epoch": 0.19,
"learning_rate": 7.436339011620007e-05,
"loss": 0.4409,
"step": 264
},
{
"epoch": 0.19,
"learning_rate": 7.426940832797941e-05,
"loss": 0.3886,
"step": 266
},
{
"epoch": 0.19,
"learning_rate": 7.417470991448494e-05,
"loss": 0.4372,
"step": 268
},
{
"epoch": 0.19,
"learning_rate": 7.407929685600403e-05,
"loss": 0.4362,
"step": 270
},
{
"epoch": 0.19,
"learning_rate": 7.398317114776836e-05,
"loss": 0.4164,
"step": 272
},
{
"epoch": 0.2,
"learning_rate": 7.388633479991219e-05,
"loss": 0.4498,
"step": 274
},
{
"epoch": 0.2,
"learning_rate": 7.378878983743037e-05,
"loss": 0.3623,
"step": 276
},
{
"epoch": 0.2,
"learning_rate": 7.369053830013588e-05,
"loss": 0.4217,
"step": 278
},
{
"epoch": 0.2,
"learning_rate": 7.35915822426173e-05,
"loss": 0.4554,
"step": 280
},
{
"epoch": 0.2,
"learning_rate": 7.349192373419583e-05,
"loss": 0.4505,
"step": 282
},
{
"epoch": 0.2,
"learning_rate": 7.339156485888191e-05,
"loss": 0.3735,
"step": 284
},
{
"epoch": 0.2,
"learning_rate": 7.329050771533176e-05,
"loss": 0.4257,
"step": 286
},
{
"epoch": 0.21,
"learning_rate": 7.31887544168034e-05,
"loss": 0.469,
"step": 288
},
{
"epoch": 0.21,
"learning_rate": 7.308630709111261e-05,
"loss": 0.3918,
"step": 290
},
{
"epoch": 0.21,
"learning_rate": 7.29831678805882e-05,
"loss": 0.4332,
"step": 292
},
{
"epoch": 0.21,
"learning_rate": 7.287933894202743e-05,
"loss": 0.4651,
"step": 294
},
{
"epoch": 0.21,
"learning_rate": 7.277482244665079e-05,
"loss": 0.3642,
"step": 296
},
{
"epoch": 0.21,
"learning_rate": 7.266962058005659e-05,
"loss": 0.4475,
"step": 298
},
{
"epoch": 0.21,
"learning_rate": 7.256373554217536e-05,
"loss": 0.4169,
"step": 300
},
{
"epoch": 0.22,
"learning_rate": 7.245716954722374e-05,
"loss": 0.4228,
"step": 302
},
{
"epoch": 0.22,
"learning_rate": 7.234992482365821e-05,
"loss": 0.4214,
"step": 304
},
{
"epoch": 0.22,
"learning_rate": 7.22420036141285e-05,
"loss": 0.4339,
"step": 306
},
{
"epoch": 0.22,
"learning_rate": 7.21334081754307e-05,
"loss": 0.4068,
"step": 308
},
{
"epoch": 0.22,
"learning_rate": 7.202414077846003e-05,
"loss": 0.4048,
"step": 310
},
{
"epoch": 0.22,
"learning_rate": 7.191420370816339e-05,
"loss": 0.4318,
"step": 312
},
{
"epoch": 0.22,
"learning_rate": 7.180359926349158e-05,
"loss": 0.3897,
"step": 314
},
{
"epoch": 0.23,
"learning_rate": 7.169232975735115e-05,
"loss": 0.4575,
"step": 316
},
{
"epoch": 0.23,
"learning_rate": 7.158039751655619e-05,
"loss": 0.442,
"step": 318
},
{
"epoch": 0.23,
"learning_rate": 7.146780488177948e-05,
"loss": 0.3831,
"step": 320
},
{
"epoch": 0.23,
"learning_rate": 7.135455420750371e-05,
"loss": 0.3654,
"step": 322
},
{
"epoch": 0.23,
"learning_rate": 7.124064786197213e-05,
"loss": 0.4198,
"step": 324
},
{
"epoch": 0.23,
"learning_rate": 7.112608822713908e-05,
"loss": 0.4102,
"step": 326
},
{
"epoch": 0.23,
"learning_rate": 7.10108776986202e-05,
"loss": 0.375,
"step": 328
},
{
"epoch": 0.24,
"learning_rate": 7.089501868564224e-05,
"loss": 0.4253,
"step": 330
},
{
"epoch": 0.24,
"learning_rate": 7.077851361099281e-05,
"loss": 0.4087,
"step": 332
},
{
"epoch": 0.24,
"learning_rate": 7.066136491096961e-05,
"loss": 0.4172,
"step": 334
},
{
"epoch": 0.24,
"learning_rate": 7.054357503532953e-05,
"loss": 0.4306,
"step": 336
},
{
"epoch": 0.24,
"learning_rate": 7.042514644723741e-05,
"loss": 0.3946,
"step": 338
},
{
"epoch": 0.24,
"learning_rate": 7.030608162321455e-05,
"loss": 0.4295,
"step": 340
},
{
"epoch": 0.24,
"learning_rate": 7.018638305308693e-05,
"loss": 0.4249,
"step": 342
},
{
"epoch": 0.25,
"learning_rate": 7.006605323993301e-05,
"loss": 0.3931,
"step": 344
},
{
"epoch": 0.25,
"learning_rate": 6.994509470003167e-05,
"loss": 0.3852,
"step": 346
},
{
"epoch": 0.25,
"learning_rate": 6.982350996280926e-05,
"loss": 0.3973,
"step": 348
},
{
"epoch": 0.25,
"learning_rate": 6.970130157078697e-05,
"loss": 0.422,
"step": 350
},
{
"epoch": 0.25,
"learning_rate": 6.957847207952752e-05,
"loss": 0.3904,
"step": 352
},
{
"epoch": 0.25,
"learning_rate": 6.945502405758174e-05,
"loss": 0.3854,
"step": 354
},
{
"epoch": 0.25,
"learning_rate": 6.933096008643489e-05,
"loss": 0.3958,
"step": 356
},
{
"epoch": 0.25,
"learning_rate": 6.92062827604527e-05,
"loss": 0.4068,
"step": 358
},
{
"epoch": 0.26,
"learning_rate": 6.9080994686827e-05,
"loss": 0.3632,
"step": 360
},
{
"epoch": 0.26,
"learning_rate": 6.895509848552134e-05,
"loss": 0.4087,
"step": 362
},
{
"epoch": 0.26,
"learning_rate": 6.882859678921612e-05,
"loss": 0.3684,
"step": 364
},
{
"epoch": 0.26,
"learning_rate": 6.870149224325358e-05,
"loss": 0.4083,
"step": 366
},
{
"epoch": 0.26,
"learning_rate": 6.85737875055824e-05,
"loss": 0.4025,
"step": 368
},
{
"epoch": 0.26,
"learning_rate": 6.844548524670226e-05,
"loss": 0.3936,
"step": 370
},
{
"epoch": 0.26,
"learning_rate": 6.831658814960786e-05,
"loss": 0.4363,
"step": 372
},
{
"epoch": 0.27,
"learning_rate": 6.818709890973289e-05,
"loss": 0.4119,
"step": 374
},
{
"epoch": 0.27,
"learning_rate": 6.805702023489361e-05,
"loss": 0.3917,
"step": 376
},
{
"epoch": 0.27,
"learning_rate": 6.792635484523231e-05,
"loss": 0.4127,
"step": 378
},
{
"epoch": 0.27,
"learning_rate": 6.779510547316034e-05,
"loss": 0.3575,
"step": 380
},
{
"epoch": 0.27,
"learning_rate": 6.766327486330103e-05,
"loss": 0.3646,
"step": 382
},
{
"epoch": 0.27,
"learning_rate": 6.753086577243225e-05,
"loss": 0.4003,
"step": 384
},
{
"epoch": 0.27,
"learning_rate": 6.739788096942879e-05,
"loss": 0.4276,
"step": 386
},
{
"epoch": 0.28,
"learning_rate": 6.726432323520447e-05,
"loss": 0.3895,
"step": 388
},
{
"epoch": 0.28,
"learning_rate": 6.713019536265395e-05,
"loss": 0.392,
"step": 390
},
{
"epoch": 0.28,
"learning_rate": 6.699550015659437e-05,
"loss": 0.3909,
"step": 392
},
{
"epoch": 0.28,
"learning_rate": 6.686024043370667e-05,
"loss": 0.39,
"step": 394
},
{
"epoch": 0.28,
"learning_rate": 6.672441902247666e-05,
"loss": 0.3835,
"step": 396
},
{
"epoch": 0.28,
"learning_rate": 6.658803876313592e-05,
"loss": 0.4331,
"step": 398
},
{
"epoch": 0.28,
"learning_rate": 6.645110250760243e-05,
"loss": 0.3894,
"step": 400
},
{
"epoch": 0.29,
"learning_rate": 6.631361311942085e-05,
"loss": 0.3964,
"step": 402
},
{
"epoch": 0.29,
"learning_rate": 6.617557347370266e-05,
"loss": 0.3844,
"step": 404
},
{
"epoch": 0.29,
"learning_rate": 6.603698645706614e-05,
"loss": 0.3761,
"step": 406
},
{
"epoch": 0.29,
"learning_rate": 6.589785496757587e-05,
"loss": 0.399,
"step": 408
},
{
"epoch": 0.29,
"learning_rate": 6.575818191468217e-05,
"loss": 0.4046,
"step": 410
},
{
"epoch": 0.29,
"learning_rate": 6.561797021916031e-05,
"loss": 0.402,
"step": 412
},
{
"epoch": 0.29,
"learning_rate": 6.547722281304935e-05,
"loss": 0.3824,
"step": 414
},
{
"epoch": 0.3,
"learning_rate": 6.533594263959095e-05,
"loss": 0.4048,
"step": 416
},
{
"epoch": 0.3,
"learning_rate": 6.519413265316762e-05,
"loss": 0.382,
"step": 418
},
{
"epoch": 0.3,
"learning_rate": 6.50517958192412e-05,
"loss": 0.3952,
"step": 420
},
{
"epoch": 0.3,
"learning_rate": 6.49089351142906e-05,
"loss": 0.3711,
"step": 422
},
{
"epoch": 0.3,
"learning_rate": 6.476555352574973e-05,
"loss": 0.3902,
"step": 424
},
{
"epoch": 0.3,
"learning_rate": 6.462165405194492e-05,
"loss": 0.4164,
"step": 426
},
{
"epoch": 0.3,
"learning_rate": 6.447723970203232e-05,
"loss": 0.4021,
"step": 428
},
{
"epoch": 0.31,
"learning_rate": 6.433231349593487e-05,
"loss": 0.3853,
"step": 430
},
{
"epoch": 0.31,
"learning_rate": 6.418687846427922e-05,
"loss": 0.385,
"step": 432
},
{
"epoch": 0.31,
"learning_rate": 6.404093764833235e-05,
"loss": 0.4017,
"step": 434
},
{
"epoch": 0.31,
"learning_rate": 6.389449409993791e-05,
"loss": 0.3751,
"step": 436
},
{
"epoch": 0.31,
"learning_rate": 6.374755088145253e-05,
"loss": 0.3937,
"step": 438
},
{
"epoch": 0.31,
"learning_rate": 6.360011106568158e-05,
"loss": 0.3854,
"step": 440
},
{
"epoch": 0.31,
"learning_rate": 6.345217773581512e-05,
"loss": 0.4066,
"step": 442
},
{
"epoch": 0.32,
"learning_rate": 6.330375398536333e-05,
"loss": 0.4047,
"step": 444
},
{
"epoch": 0.32,
"learning_rate": 6.315484291809181e-05,
"loss": 0.3834,
"step": 446
},
{
"epoch": 0.32,
"learning_rate": 6.300544764795669e-05,
"loss": 0.4043,
"step": 448
},
{
"epoch": 0.32,
"learning_rate": 6.285557129903954e-05,
"loss": 0.3703,
"step": 450
},
{
"epoch": 0.32,
"learning_rate": 6.270521700548196e-05,
"loss": 0.3748,
"step": 452
},
{
"epoch": 0.32,
"learning_rate": 6.255438791142017e-05,
"loss": 0.4021,
"step": 454
},
{
"epoch": 0.32,
"learning_rate": 6.240308717091913e-05,
"loss": 0.3973,
"step": 456
},
{
"epoch": 0.33,
"learning_rate": 6.22513179479067e-05,
"loss": 0.3821,
"step": 458
},
{
"epoch": 0.33,
"learning_rate": 6.209908341610736e-05,
"loss": 0.4095,
"step": 460
},
{
"epoch": 0.33,
"learning_rate": 6.194638675897591e-05,
"loss": 0.4244,
"step": 462
},
{
"epoch": 0.33,
"learning_rate": 6.179323116963095e-05,
"loss": 0.3658,
"step": 464
},
{
"epoch": 0.33,
"learning_rate": 6.163961985078798e-05,
"loss": 0.3782,
"step": 466
},
{
"epoch": 0.33,
"learning_rate": 6.148555601469252e-05,
"loss": 0.4021,
"step": 468
},
{
"epoch": 0.33,
"learning_rate": 6.133104288305293e-05,
"loss": 0.3933,
"step": 470
},
{
"epoch": 0.34,
"learning_rate": 6.1176083686973e-05,
"loss": 0.3814,
"step": 472
},
{
"epoch": 0.34,
"learning_rate": 6.1020681666884406e-05,
"loss": 0.3991,
"step": 474
},
{
"epoch": 0.34,
"learning_rate": 6.086484007247894e-05,
"loss": 0.4124,
"step": 476
},
{
"epoch": 0.34,
"learning_rate": 6.070856216264057e-05,
"loss": 0.369,
"step": 478
},
{
"epoch": 0.34,
"learning_rate": 6.055185120537728e-05,
"loss": 0.379,
"step": 480
},
{
"epoch": 0.34,
"learning_rate": 6.039471047775273e-05,
"loss": 0.4135,
"step": 482
},
{
"epoch": 0.34,
"learning_rate": 6.023714326581772e-05,
"loss": 0.3477,
"step": 484
},
{
"epoch": 0.35,
"learning_rate": 6.0079152864541506e-05,
"loss": 0.4139,
"step": 486
},
{
"epoch": 0.35,
"learning_rate": 5.992074257774285e-05,
"loss": 0.3711,
"step": 488
},
{
"epoch": 0.35,
"learning_rate": 5.976191571802097e-05,
"loss": 0.4052,
"step": 490
},
{
"epoch": 0.35,
"learning_rate": 5.960267560668625e-05,
"loss": 0.402,
"step": 492
},
{
"epoch": 0.35,
"learning_rate": 5.9443025573690795e-05,
"loss": 0.3992,
"step": 494
},
{
"epoch": 0.35,
"learning_rate": 5.928296895755882e-05,
"loss": 0.3895,
"step": 496
},
{
"epoch": 0.35,
"learning_rate": 5.9122509105316736e-05,
"loss": 0.4004,
"step": 498
},
{
"epoch": 0.36,
"learning_rate": 5.896164937242331e-05,
"loss": 0.3736,
"step": 500
},
{
"epoch": 0.36,
"learning_rate": 5.880039312269939e-05,
"loss": 0.3452,
"step": 502
},
{
"epoch": 0.36,
"learning_rate": 5.863874372825757e-05,
"loss": 0.3806,
"step": 504
},
{
"epoch": 0.36,
"learning_rate": 5.847670456943174e-05,
"loss": 0.3684,
"step": 506
},
{
"epoch": 0.36,
"learning_rate": 5.831427903470631e-05,
"loss": 0.3891,
"step": 508
},
{
"epoch": 0.36,
"learning_rate": 5.815147052064543e-05,
"loss": 0.3882,
"step": 510
},
{
"epoch": 0.36,
"learning_rate": 5.798828243182191e-05,
"loss": 0.3889,
"step": 512
},
{
"epoch": 0.37,
"learning_rate": 5.7824718180746034e-05,
"loss": 0.3882,
"step": 514
},
{
"epoch": 0.37,
"learning_rate": 5.766078118779424e-05,
"loss": 0.3863,
"step": 516
},
{
"epoch": 0.37,
"learning_rate": 5.749647488113755e-05,
"loss": 0.3629,
"step": 518
},
{
"epoch": 0.37,
"learning_rate": 5.7331802696669885e-05,
"loss": 0.3666,
"step": 520
},
{
"epoch": 0.37,
"learning_rate": 5.716676807793624e-05,
"loss": 0.3499,
"step": 522
},
{
"epoch": 0.37,
"learning_rate": 5.7001374476060646e-05,
"loss": 0.4004,
"step": 524
},
{
"epoch": 0.37,
"learning_rate": 5.683562534967401e-05,
"loss": 0.3468,
"step": 526
},
{
"epoch": 0.38,
"learning_rate": 5.666952416484183e-05,
"loss": 0.3729,
"step": 528
},
{
"epoch": 0.38,
"learning_rate": 5.650307439499164e-05,
"loss": 0.3763,
"step": 530
},
{
"epoch": 0.38,
"learning_rate": 5.6336279520840445e-05,
"loss": 0.3543,
"step": 532
},
{
"epoch": 0.38,
"learning_rate": 5.61691430303219e-05,
"loss": 0.4355,
"step": 534
},
{
"epoch": 0.38,
"learning_rate": 5.600166841851333e-05,
"loss": 0.3764,
"step": 536
},
{
"epoch": 0.38,
"learning_rate": 5.583385918756277e-05,
"loss": 0.3594,
"step": 538
},
{
"epoch": 0.38,
"learning_rate": 5.5665718846615584e-05,
"loss": 0.3589,
"step": 540
},
{
"epoch": 0.39,
"learning_rate": 5.549725091174117e-05,
"loss": 0.3726,
"step": 542
},
{
"epoch": 0.39,
"learning_rate": 5.532845890585941e-05,
"loss": 0.3803,
"step": 544
},
{
"epoch": 0.39,
"learning_rate": 5.5159346358667e-05,
"loss": 0.3661,
"step": 546
},
{
"epoch": 0.39,
"learning_rate": 5.4989916806563655e-05,
"loss": 0.3902,
"step": 548
},
{
"epoch": 0.39,
"learning_rate": 5.482017379257812e-05,
"loss": 0.3953,
"step": 550
},
{
"epoch": 0.39,
"learning_rate": 5.465012086629411e-05,
"loss": 0.3633,
"step": 552
},
{
"epoch": 0.39,
"learning_rate": 5.447976158377611e-05,
"loss": 0.3559,
"step": 554
},
{
"epoch": 0.4,
"learning_rate": 5.430909950749491e-05,
"loss": 0.3633,
"step": 556
},
{
"epoch": 0.4,
"learning_rate": 5.413813820625324e-05,
"loss": 0.3553,
"step": 558
},
{
"epoch": 0.4,
"learning_rate": 5.3966881255111024e-05,
"loss": 0.3908,
"step": 560
},
{
"epoch": 0.4,
"learning_rate": 5.3795332235310744e-05,
"loss": 0.3674,
"step": 562
},
{
"epoch": 0.4,
"learning_rate": 5.3623494734202415e-05,
"loss": 0.4049,
"step": 564
},
{
"epoch": 0.4,
"learning_rate": 5.345137234516864e-05,
"loss": 0.351,
"step": 566
},
{
"epoch": 0.4,
"learning_rate": 5.327896866754953e-05,
"loss": 0.3518,
"step": 568
},
{
"epoch": 0.41,
"learning_rate": 5.310628730656726e-05,
"loss": 0.3458,
"step": 570
},
{
"epoch": 0.41,
"learning_rate": 5.293333187325086e-05,
"loss": 0.3756,
"step": 572
},
{
"epoch": 0.41,
"learning_rate": 5.276010598436061e-05,
"loss": 0.3508,
"step": 574
},
{
"epoch": 0.41,
"learning_rate": 5.2586613262312414e-05,
"loss": 0.3665,
"step": 576
},
{
"epoch": 0.41,
"learning_rate": 5.241285733510209e-05,
"loss": 0.3502,
"step": 578
},
{
"epoch": 0.41,
"learning_rate": 5.223884183622943e-05,
"loss": 0.3846,
"step": 580
},
{
"epoch": 0.41,
"learning_rate": 5.20645704046223e-05,
"loss": 0.3475,
"step": 582
},
{
"epoch": 0.42,
"learning_rate": 5.189004668456051e-05,
"loss": 0.4049,
"step": 584
},
{
"epoch": 0.42,
"learning_rate": 5.171527432559957e-05,
"loss": 0.3801,
"step": 586
},
{
"epoch": 0.42,
"learning_rate": 5.1540256982494425e-05,
"loss": 0.4386,
"step": 588
},
{
"epoch": 0.42,
"learning_rate": 5.1364998315123024e-05,
"loss": 0.3385,
"step": 590
},
{
"epoch": 0.42,
"learning_rate": 5.118950198840977e-05,
"loss": 0.3704,
"step": 592
},
{
"epoch": 0.42,
"learning_rate": 5.101377167224885e-05,
"loss": 0.3842,
"step": 594
},
{
"epoch": 0.42,
"learning_rate": 5.0837811041427534e-05,
"loss": 0.3807,
"step": 596
},
{
"epoch": 0.43,
"learning_rate": 5.0661623775549376e-05,
"loss": 0.377,
"step": 598
},
{
"epoch": 0.43,
"learning_rate": 5.048521355895714e-05,
"loss": 0.3896,
"step": 600
},
{
"epoch": 0.43,
"learning_rate": 5.030858408065587e-05,
"loss": 0.3721,
"step": 602
},
{
"epoch": 0.43,
"learning_rate": 5.01317390342357e-05,
"loss": 0.3744,
"step": 604
},
{
"epoch": 0.43,
"learning_rate": 4.995468211779459e-05,
"loss": 0.4038,
"step": 606
},
{
"epoch": 0.43,
"learning_rate": 4.977741703386107e-05,
"loss": 0.3491,
"step": 608
},
{
"epoch": 0.43,
"learning_rate": 4.9599947489316725e-05,
"loss": 0.3655,
"step": 610
},
{
"epoch": 0.44,
"learning_rate": 4.942227719531875e-05,
"loss": 0.3962,
"step": 612
},
{
"epoch": 0.44,
"learning_rate": 4.924440986722231e-05,
"loss": 0.3642,
"step": 614
},
{
"epoch": 0.44,
"learning_rate": 4.906634922450285e-05,
"loss": 0.3703,
"step": 616
},
{
"epoch": 0.44,
"learning_rate": 4.888809899067829e-05,
"loss": 0.3628,
"step": 618
},
{
"epoch": 0.44,
"learning_rate": 4.870966289323123e-05,
"loss": 0.3639,
"step": 620
},
{
"epoch": 0.44,
"learning_rate": 4.8531044663530944e-05,
"loss": 0.3929,
"step": 622
},
{
"epoch": 0.44,
"learning_rate": 4.835224803675537e-05,
"loss": 0.3438,
"step": 624
},
{
"epoch": 0.45,
"learning_rate": 4.817327675181297e-05,
"loss": 0.3349,
"step": 626
},
{
"epoch": 0.45,
"learning_rate": 4.799413455126461e-05,
"loss": 0.3395,
"step": 628
},
{
"epoch": 0.45,
"learning_rate": 4.7814825181245254e-05,
"loss": 0.3754,
"step": 630
},
{
"epoch": 0.45,
"learning_rate": 4.763535239138562e-05,
"loss": 0.3264,
"step": 632
},
{
"epoch": 0.45,
"learning_rate": 4.74557199347338e-05,
"loss": 0.3946,
"step": 634
},
{
"epoch": 0.45,
"learning_rate": 4.7275931567676735e-05,
"loss": 0.3699,
"step": 636
},
{
"epoch": 0.45,
"learning_rate": 4.709599104986172e-05,
"loss": 0.3708,
"step": 638
},
{
"epoch": 0.46,
"learning_rate": 4.691590214411772e-05,
"loss": 0.3557,
"step": 640
},
{
"epoch": 0.46,
"learning_rate": 4.6735668616376734e-05,
"loss": 0.3537,
"step": 642
},
{
"epoch": 0.46,
"learning_rate": 4.6555294235595047e-05,
"loss": 0.3601,
"step": 644
},
{
"epoch": 0.46,
"learning_rate": 4.637478277367436e-05,
"loss": 0.366,
"step": 646
},
{
"epoch": 0.46,
"learning_rate": 4.6194138005382945e-05,
"loss": 0.3928,
"step": 648
},
{
"epoch": 0.46,
"learning_rate": 4.6013363708276754e-05,
"loss": 0.3209,
"step": 650
},
{
"epoch": 0.46,
"learning_rate": 4.5832463662620353e-05,
"loss": 0.3796,
"step": 652
}
],
"max_steps": 1403,
"num_train_epochs": 1,
"total_flos": 3.846934282119414e+17,
"trial_name": null,
"trial_params": null
}