Delete trainer_state.json with huggingface_hub
Browse files- trainer_state.json +0 -313
trainer_state.json
DELETED
|
@@ -1,313 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.12358882947118241,
|
| 5 |
-
"eval_steps": 62,
|
| 6 |
-
"global_step": 40,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.0030897207367795603,
|
| 13 |
-
"grad_norm": 5275.14208984375,
|
| 14 |
-
"learning_rate": 4.0000000000000004e-11,
|
| 15 |
-
"loss": 11.2955,
|
| 16 |
-
"step": 1
|
| 17 |
-
},
|
| 18 |
-
{
|
| 19 |
-
"epoch": 0.006179441473559121,
|
| 20 |
-
"grad_norm": 5153.88330078125,
|
| 21 |
-
"learning_rate": 8.000000000000001e-11,
|
| 22 |
-
"loss": 11.2979,
|
| 23 |
-
"step": 2
|
| 24 |
-
},
|
| 25 |
-
{
|
| 26 |
-
"epoch": 0.009269162210338681,
|
| 27 |
-
"grad_norm": 5132.99609375,
|
| 28 |
-
"learning_rate": 1.2e-10,
|
| 29 |
-
"loss": 11.1883,
|
| 30 |
-
"step": 3
|
| 31 |
-
},
|
| 32 |
-
{
|
| 33 |
-
"epoch": 0.012358882947118241,
|
| 34 |
-
"grad_norm": 5524.13623046875,
|
| 35 |
-
"learning_rate": 1.6000000000000002e-10,
|
| 36 |
-
"loss": 11.5185,
|
| 37 |
-
"step": 4
|
| 38 |
-
},
|
| 39 |
-
{
|
| 40 |
-
"epoch": 0.015448603683897801,
|
| 41 |
-
"grad_norm": 5275.44970703125,
|
| 42 |
-
"learning_rate": 2e-10,
|
| 43 |
-
"loss": 11.202,
|
| 44 |
-
"step": 5
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.018538324420677363,
|
| 48 |
-
"grad_norm": 4898.6171875,
|
| 49 |
-
"learning_rate": 1.9984815164333162e-10,
|
| 50 |
-
"loss": 11.0454,
|
| 51 |
-
"step": 6
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.02162804515745692,
|
| 55 |
-
"grad_norm": 5124.095703125,
|
| 56 |
-
"learning_rate": 1.9939306773179497e-10,
|
| 57 |
-
"loss": 11.1537,
|
| 58 |
-
"step": 7
|
| 59 |
-
},
|
| 60 |
-
{
|
| 61 |
-
"epoch": 0.024717765894236483,
|
| 62 |
-
"grad_norm": 5025.26025390625,
|
| 63 |
-
"learning_rate": 1.9863613034027225e-10,
|
| 64 |
-
"loss": 11.1345,
|
| 65 |
-
"step": 8
|
| 66 |
-
},
|
| 67 |
-
{
|
| 68 |
-
"epoch": 0.027807486631016044,
|
| 69 |
-
"grad_norm": 4950.40576171875,
|
| 70 |
-
"learning_rate": 1.9757963826274357e-10,
|
| 71 |
-
"loss": 11.1104,
|
| 72 |
-
"step": 9
|
| 73 |
-
},
|
| 74 |
-
{
|
| 75 |
-
"epoch": 0.030897207367795602,
|
| 76 |
-
"grad_norm": 4390.81787109375,
|
| 77 |
-
"learning_rate": 1.9622680003092504e-10,
|
| 78 |
-
"loss": 10.546,
|
| 79 |
-
"step": 10
|
| 80 |
-
},
|
| 81 |
-
{
|
| 82 |
-
"epoch": 0.03398692810457516,
|
| 83 |
-
"grad_norm": 4808.67626953125,
|
| 84 |
-
"learning_rate": 1.9458172417006345e-10,
|
| 85 |
-
"loss": 10.8947,
|
| 86 |
-
"step": 11
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.037076648841354726,
|
| 90 |
-
"grad_norm": 5642.1826171875,
|
| 91 |
-
"learning_rate": 1.9264940672148017e-10,
|
| 92 |
-
"loss": 11.7941,
|
| 93 |
-
"step": 12
|
| 94 |
-
},
|
| 95 |
-
{
|
| 96 |
-
"epoch": 0.040166369578134284,
|
| 97 |
-
"grad_norm": 4708.16650390625,
|
| 98 |
-
"learning_rate": 1.9043571606975777e-10,
|
| 99 |
-
"loss": 10.8668,
|
| 100 |
-
"step": 13
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 0.04325609031491384,
|
| 104 |
-
"grad_norm": 5043.15966796875,
|
| 105 |
-
"learning_rate": 1.879473751206489e-10,
|
| 106 |
-
"loss": 11.0351,
|
| 107 |
-
"step": 14
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 0.04634581105169341,
|
| 111 |
-
"grad_norm": 5204.0048828125,
|
| 112 |
-
"learning_rate": 1.851919408838327e-10,
|
| 113 |
-
"loss": 11.2322,
|
| 114 |
-
"step": 15
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.049435531788472965,
|
| 118 |
-
"grad_norm": 4756.47705078125,
|
| 119 |
-
"learning_rate": 1.821777815225245e-10,
|
| 120 |
-
"loss": 10.8011,
|
| 121 |
-
"step": 16
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.052525252525252523,
|
| 125 |
-
"grad_norm": 4809.21533203125,
|
| 126 |
-
"learning_rate": 1.7891405093963936e-10,
|
| 127 |
-
"loss": 10.8063,
|
| 128 |
-
"step": 17
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.05561497326203209,
|
| 132 |
-
"grad_norm": 5400.25927734375,
|
| 133 |
-
"learning_rate": 1.7541066097768963e-10,
|
| 134 |
-
"loss": 11.4944,
|
| 135 |
-
"step": 18
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.05870469399881165,
|
| 139 |
-
"grad_norm": 5103.818359375,
|
| 140 |
-
"learning_rate": 1.7167825131684515e-10,
|
| 141 |
-
"loss": 11.2059,
|
| 142 |
-
"step": 19
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.061794414735591205,
|
| 146 |
-
"grad_norm": 4988.18798828125,
|
| 147 |
-
"learning_rate": 1.6772815716257412e-10,
|
| 148 |
-
"loss": 11.0932,
|
| 149 |
-
"step": 20
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.06488413547237076,
|
| 153 |
-
"grad_norm": 5250.4130859375,
|
| 154 |
-
"learning_rate": 1.6357237482099684e-10,
|
| 155 |
-
"loss": 11.3196,
|
| 156 |
-
"step": 21
|
| 157 |
-
},
|
| 158 |
-
{
|
| 159 |
-
"epoch": 0.06797385620915032,
|
| 160 |
-
"grad_norm": 5046.6142578125,
|
| 161 |
-
"learning_rate": 1.59223525266498e-10,
|
| 162 |
-
"loss": 11.1777,
|
| 163 |
-
"step": 22
|
| 164 |
-
},
|
| 165 |
-
{
|
| 166 |
-
"epoch": 0.0710635769459299,
|
| 167 |
-
"grad_norm": 4959.35400390625,
|
| 168 |
-
"learning_rate": 1.5469481581224272e-10,
|
| 169 |
-
"loss": 11.2175,
|
| 170 |
-
"step": 23
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.07415329768270945,
|
| 174 |
-
"grad_norm": 5114.306640625,
|
| 175 |
-
"learning_rate": 1.5e-10,
|
| 176 |
-
"loss": 11.0837,
|
| 177 |
-
"step": 24
|
| 178 |
-
},
|
| 179 |
-
{
|
| 180 |
-
"epoch": 0.07724301841948901,
|
| 181 |
-
"grad_norm": 4886.6083984375,
|
| 182 |
-
"learning_rate": 1.4515333583108894e-10,
|
| 183 |
-
"loss": 10.9479,
|
| 184 |
-
"step": 25
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.08033273915626857,
|
| 188 |
-
"grad_norm": 5151.87548828125,
|
| 189 |
-
"learning_rate": 1.4016954246529695e-10,
|
| 190 |
-
"loss": 11.1946,
|
| 191 |
-
"step": 26
|
| 192 |
-
},
|
| 193 |
-
{
|
| 194 |
-
"epoch": 0.08342245989304813,
|
| 195 |
-
"grad_norm": 4926.22314453125,
|
| 196 |
-
"learning_rate": 1.3506375551927545e-10,
|
| 197 |
-
"loss": 11.1436,
|
| 198 |
-
"step": 27
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.08651218062982768,
|
| 202 |
-
"grad_norm": 5186.583984375,
|
| 203 |
-
"learning_rate": 1.2985148110016947e-10,
|
| 204 |
-
"loss": 11.3389,
|
| 205 |
-
"step": 28
|
| 206 |
-
},
|
| 207 |
-
{
|
| 208 |
-
"epoch": 0.08960190136660724,
|
| 209 |
-
"grad_norm": 4836.59130859375,
|
| 210 |
-
"learning_rate": 1.2454854871407994e-10,
|
| 211 |
-
"loss": 11.131,
|
| 212 |
-
"step": 29
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.09269162210338681,
|
| 216 |
-
"grad_norm": 5271.8818359375,
|
| 217 |
-
"learning_rate": 1.1917106319237384e-10,
|
| 218 |
-
"loss": 11.4785,
|
| 219 |
-
"step": 30
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.09578134284016637,
|
| 223 |
-
"grad_norm": 4995.57373046875,
|
| 224 |
-
"learning_rate": 1.1373535578184083e-10,
|
| 225 |
-
"loss": 11.1493,
|
| 226 |
-
"step": 31
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.09887106357694593,
|
| 230 |
-
"grad_norm": 4872.11376953125,
|
| 231 |
-
"learning_rate": 1.0825793454723326e-10,
|
| 232 |
-
"loss": 10.97,
|
| 233 |
-
"step": 32
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 0.10196078431372549,
|
| 237 |
-
"grad_norm": 5491.39013671875,
|
| 238 |
-
"learning_rate": 1.0275543423681622e-10,
|
| 239 |
-
"loss": 11.5456,
|
| 240 |
-
"step": 33
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 0.10505050505050505,
|
| 244 |
-
"grad_norm": 5149.216796875,
|
| 245 |
-
"learning_rate": 9.724456576318382e-11,
|
| 246 |
-
"loss": 11.257,
|
| 247 |
-
"step": 34
|
| 248 |
-
},
|
| 249 |
-
{
|
| 250 |
-
"epoch": 0.1081402257872846,
|
| 251 |
-
"grad_norm": 5288.52294921875,
|
| 252 |
-
"learning_rate": 9.174206545276678e-11,
|
| 253 |
-
"loss": 11.4861,
|
| 254 |
-
"step": 35
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.11122994652406418,
|
| 258 |
-
"grad_norm": 5499.283203125,
|
| 259 |
-
"learning_rate": 8.626464421815918e-11,
|
| 260 |
-
"loss": 11.4375,
|
| 261 |
-
"step": 36
|
| 262 |
-
},
|
| 263 |
-
{
|
| 264 |
-
"epoch": 0.11431966726084374,
|
| 265 |
-
"grad_norm": 4957.93310546875,
|
| 266 |
-
"learning_rate": 8.082893680762619e-11,
|
| 267 |
-
"loss": 11.0605,
|
| 268 |
-
"step": 37
|
| 269 |
-
},
|
| 270 |
-
{
|
| 271 |
-
"epoch": 0.1174093879976233,
|
| 272 |
-
"grad_norm": 4736.22705078125,
|
| 273 |
-
"learning_rate": 7.54514512859201e-11,
|
| 274 |
-
"loss": 10.8943,
|
| 275 |
-
"step": 38
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 0.12049910873440285,
|
| 279 |
-
"grad_norm": 5041.0615234375,
|
| 280 |
-
"learning_rate": 7.014851889983058e-11,
|
| 281 |
-
"loss": 11.2565,
|
| 282 |
-
"step": 39
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 0.12358882947118241,
|
| 286 |
-
"grad_norm": 5379.4970703125,
|
| 287 |
-
"learning_rate": 6.493624448072457e-11,
|
| 288 |
-
"loss": 11.4698,
|
| 289 |
-
"step": 40
|
| 290 |
-
}
|
| 291 |
-
],
|
| 292 |
-
"logging_steps": 1,
|
| 293 |
-
"max_steps": 62,
|
| 294 |
-
"num_input_tokens_seen": 0,
|
| 295 |
-
"num_train_epochs": 1,
|
| 296 |
-
"save_steps": 10,
|
| 297 |
-
"stateful_callbacks": {
|
| 298 |
-
"TrainerControl": {
|
| 299 |
-
"args": {
|
| 300 |
-
"should_epoch_stop": false,
|
| 301 |
-
"should_evaluate": false,
|
| 302 |
-
"should_log": false,
|
| 303 |
-
"should_save": true,
|
| 304 |
-
"should_training_stop": false
|
| 305 |
-
},
|
| 306 |
-
"attributes": {}
|
| 307 |
-
}
|
| 308 |
-
},
|
| 309 |
-
"total_flos": 6.637517228998656e+16,
|
| 310 |
-
"train_batch_size": 2,
|
| 311 |
-
"trial_name": null,
|
| 312 |
-
"trial_params": null
|
| 313 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|