Commit
·
6a33562
1
Parent(s):
2a3559e
ThyroidTumorClassification
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- pytorch_model.bin +1 -1
- train_results.json +6 -6
- trainer_state.json +137 -1373
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_f1": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 0.
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
-
"total_flos": 1.
|
| 9 |
-
"train_loss": 0.
|
| 10 |
-
"train_runtime":
|
| 11 |
-
"train_samples_per_second":
|
| 12 |
-
"train_steps_per_second": 0.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 19.73,
|
| 3 |
+
"eval_f1": 0.9312977099236641,
|
| 4 |
+
"eval_loss": 0.5987359285354614,
|
| 5 |
+
"eval_runtime": 0.4648,
|
| 6 |
+
"eval_samples_per_second": 176.425,
|
| 7 |
+
"eval_steps_per_second": 6.455,
|
| 8 |
+
"total_flos": 1.620732679243776e+17,
|
| 9 |
+
"train_loss": 0.8578881025314331,
|
| 10 |
+
"train_runtime": 102.9255,
|
| 11 |
+
"train_samples_per_second": 63.347,
|
| 12 |
+
"train_steps_per_second": 0.389
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_f1": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 0.
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 19.73,
|
| 3 |
+
"eval_f1": 0.9312977099236641,
|
| 4 |
+
"eval_loss": 0.5987359285354614,
|
| 5 |
+
"eval_runtime": 0.4648,
|
| 6 |
+
"eval_samples_per_second": 176.425,
|
| 7 |
+
"eval_steps_per_second": 6.455
|
| 8 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 111347349
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ce7c3b86954dad0996ccf49be4284947b73504f1521f15cd9a889bf12083e90
|
| 3 |
size 111347349
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos": 1.
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 19.73,
|
| 3 |
+
"total_flos": 1.620732679243776e+17,
|
| 4 |
+
"train_loss": 0.8578881025314331,
|
| 5 |
+
"train_runtime": 102.9255,
|
| 6 |
+
"train_samples_per_second": 63.347,
|
| 7 |
+
"train_steps_per_second": 0.389
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1447 +1,211 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "Cvt-finetuned-thyroid/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
-
"epoch": 0.
|
| 12 |
-
"eval_f1": 0.
|
| 13 |
-
"eval_loss": 0.
|
| 14 |
-
"eval_runtime": 0.
|
| 15 |
-
"eval_samples_per_second":
|
| 16 |
-
"eval_steps_per_second": 5.
|
| 17 |
"step": 2
|
| 18 |
},
|
| 19 |
{
|
| 20 |
-
"epoch": 1.
|
| 21 |
-
"eval_f1": 0.
|
| 22 |
-
"eval_loss": 0.
|
| 23 |
-
"eval_runtime": 0.
|
| 24 |
-
"eval_samples_per_second":
|
| 25 |
-
"eval_steps_per_second": 6.
|
| 26 |
"step": 4
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 2.
|
| 30 |
-
"eval_f1": 0.
|
| 31 |
-
"eval_loss": 0.
|
| 32 |
-
"eval_runtime": 0.
|
| 33 |
-
"eval_samples_per_second":
|
| 34 |
-
"eval_steps_per_second":
|
| 35 |
"step": 6
|
| 36 |
},
|
| 37 |
{
|
| 38 |
-
"epoch": 3.
|
| 39 |
-
"eval_f1": 0.
|
| 40 |
-
"eval_loss": 0.
|
| 41 |
-
"eval_runtime": 0.
|
| 42 |
-
"eval_samples_per_second":
|
| 43 |
-
"eval_steps_per_second": 6.
|
| 44 |
"step": 8
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 4.
|
| 48 |
-
"eval_f1": 0.
|
| 49 |
-
"eval_loss": 0.
|
| 50 |
-
"eval_runtime": 0.
|
| 51 |
-
"eval_samples_per_second":
|
| 52 |
-
"eval_steps_per_second":
|
| 53 |
"step": 10
|
| 54 |
},
|
| 55 |
{
|
| 56 |
-
"epoch": 5.
|
| 57 |
-
"eval_f1": 0.
|
| 58 |
-
"eval_loss": 0.
|
| 59 |
-
"eval_runtime": 0.
|
| 60 |
-
"eval_samples_per_second":
|
| 61 |
-
"eval_steps_per_second":
|
| 62 |
"step": 12
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch": 6.
|
| 66 |
-
"eval_f1": 0.
|
| 67 |
-
"eval_loss": 0.
|
| 68 |
-
"eval_runtime": 0.
|
| 69 |
-
"eval_samples_per_second":
|
| 70 |
-
"eval_steps_per_second":
|
| 71 |
"step": 14
|
| 72 |
},
|
| 73 |
{
|
| 74 |
-
"epoch": 7.
|
| 75 |
-
"eval_f1": 0.
|
| 76 |
-
"eval_loss": 0.
|
| 77 |
-
"eval_runtime": 0.
|
| 78 |
-
"eval_samples_per_second":
|
| 79 |
-
"eval_steps_per_second":
|
| 80 |
"step": 16
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 8.
|
| 84 |
-
"eval_f1": 0.
|
| 85 |
-
"eval_loss": 0.
|
| 86 |
-
"eval_runtime": 0.
|
| 87 |
-
"eval_samples_per_second":
|
| 88 |
-
"eval_steps_per_second":
|
| 89 |
"step": 18
|
| 90 |
},
|
| 91 |
{
|
| 92 |
-
"epoch": 9.
|
| 93 |
-
"eval_f1": 0.
|
| 94 |
-
"eval_loss": 0.
|
| 95 |
-
"eval_runtime": 0.
|
| 96 |
-
"eval_samples_per_second":
|
| 97 |
-
"eval_steps_per_second":
|
| 98 |
"step": 20
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch": 10.
|
| 102 |
-
"eval_f1": 0.
|
| 103 |
-
"eval_loss": 0.
|
| 104 |
-
"eval_runtime": 0.
|
| 105 |
-
"eval_samples_per_second":
|
| 106 |
-
"eval_steps_per_second":
|
| 107 |
"step": 22
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 11.
|
| 111 |
-
"eval_f1": 0.
|
| 112 |
-
"eval_loss": 0.
|
| 113 |
-
"eval_runtime": 0.
|
| 114 |
-
"eval_samples_per_second":
|
| 115 |
-
"eval_steps_per_second":
|
| 116 |
"step": 24
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch": 12.
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 25
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 12.
|
| 126 |
-
"eval_f1": 0.
|
| 127 |
-
"eval_loss": 0.
|
| 128 |
-
"eval_runtime": 0.
|
| 129 |
-
"eval_samples_per_second":
|
| 130 |
-
"eval_steps_per_second":
|
| 131 |
"step": 26
|
| 132 |
},
|
| 133 |
{
|
| 134 |
-
"epoch": 13.
|
| 135 |
-
"eval_f1": 0.
|
| 136 |
-
"eval_loss": 0.
|
| 137 |
-
"eval_runtime": 0.
|
| 138 |
-
"eval_samples_per_second":
|
| 139 |
-
"eval_steps_per_second":
|
| 140 |
"step": 28
|
| 141 |
},
|
| 142 |
{
|
| 143 |
-
"epoch": 14.
|
| 144 |
-
"eval_f1": 0.
|
| 145 |
-
"eval_loss": 0.
|
| 146 |
-
"eval_runtime": 0.
|
| 147 |
-
"eval_samples_per_second":
|
| 148 |
-
"eval_steps_per_second":
|
| 149 |
"step": 30
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 15.
|
| 153 |
-
"eval_f1": 0.
|
| 154 |
-
"eval_loss": 0.
|
| 155 |
-
"eval_runtime": 0.
|
| 156 |
-
"eval_samples_per_second":
|
| 157 |
-
"eval_steps_per_second":
|
| 158 |
"step": 32
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch": 16.
|
| 162 |
-
"eval_f1": 0.
|
| 163 |
-
"eval_loss": 0.
|
| 164 |
-
"eval_runtime": 0.
|
| 165 |
-
"eval_samples_per_second":
|
| 166 |
-
"eval_steps_per_second":
|
| 167 |
"step": 34
|
| 168 |
},
|
| 169 |
{
|
| 170 |
-
"epoch": 17.
|
| 171 |
-
"eval_f1": 0.
|
| 172 |
-
"eval_loss": 0.
|
| 173 |
-
"eval_runtime": 0.
|
| 174 |
-
"eval_samples_per_second":
|
| 175 |
-
"eval_steps_per_second":
|
| 176 |
"step": 36
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch": 18.
|
| 180 |
-
"eval_f1": 0.
|
| 181 |
-
"eval_loss": 0.
|
| 182 |
-
"eval_runtime": 0.
|
| 183 |
-
"eval_samples_per_second":
|
| 184 |
-
"eval_steps_per_second":
|
| 185 |
"step": 38
|
| 186 |
},
|
| 187 |
{
|
| 188 |
-
"epoch": 19.
|
| 189 |
-
"eval_f1": 0.
|
| 190 |
-
"eval_loss": 0.
|
| 191 |
-
"eval_runtime": 0.
|
| 192 |
-
"eval_samples_per_second":
|
| 193 |
-
"eval_steps_per_second":
|
| 194 |
"step": 40
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch":
|
| 198 |
-
"
|
| 199 |
-
"
|
| 200 |
-
"
|
| 201 |
-
"
|
| 202 |
-
"
|
| 203 |
-
"
|
| 204 |
-
},
|
| 205 |
-
{
|
| 206 |
-
"epoch": 21.89,
|
| 207 |
-
"eval_f1": 0.888888888888889,
|
| 208 |
-
"eval_loss": 0.4898264706134796,
|
| 209 |
-
"eval_runtime": 0.5144,
|
| 210 |
-
"eval_samples_per_second": 136.069,
|
| 211 |
-
"eval_steps_per_second": 5.832,
|
| 212 |
-
"step": 44
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 22.89,
|
| 216 |
-
"eval_f1": 0.888888888888889,
|
| 217 |
-
"eval_loss": 0.4883803129196167,
|
| 218 |
-
"eval_runtime": 0.5035,
|
| 219 |
-
"eval_samples_per_second": 139.037,
|
| 220 |
-
"eval_steps_per_second": 5.959,
|
| 221 |
-
"step": 46
|
| 222 |
-
},
|
| 223 |
-
{
|
| 224 |
-
"epoch": 23.89,
|
| 225 |
-
"eval_f1": 0.888888888888889,
|
| 226 |
-
"eval_loss": 0.4872118830680847,
|
| 227 |
-
"eval_runtime": 0.5072,
|
| 228 |
-
"eval_samples_per_second": 138.01,
|
| 229 |
-
"eval_steps_per_second": 5.915,
|
| 230 |
-
"step": 48
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 24.89,
|
| 234 |
-
"learning_rate": 9.25925925925926e-06,
|
| 235 |
-
"loss": 0.5044,
|
| 236 |
-
"step": 50
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 24.89,
|
| 240 |
-
"eval_f1": 0.888888888888889,
|
| 241 |
-
"eval_loss": 0.4861142933368683,
|
| 242 |
-
"eval_runtime": 0.5085,
|
| 243 |
-
"eval_samples_per_second": 137.671,
|
| 244 |
-
"eval_steps_per_second": 5.9,
|
| 245 |
-
"step": 50
|
| 246 |
-
},
|
| 247 |
-
{
|
| 248 |
-
"epoch": 25.89,
|
| 249 |
-
"eval_f1": 0.888888888888889,
|
| 250 |
-
"eval_loss": 0.4849891662597656,
|
| 251 |
-
"eval_runtime": 0.5374,
|
| 252 |
-
"eval_samples_per_second": 130.249,
|
| 253 |
-
"eval_steps_per_second": 5.582,
|
| 254 |
-
"step": 52
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 26.89,
|
| 258 |
-
"eval_f1": 0.888888888888889,
|
| 259 |
-
"eval_loss": 0.483755499124527,
|
| 260 |
-
"eval_runtime": 0.51,
|
| 261 |
-
"eval_samples_per_second": 137.255,
|
| 262 |
-
"eval_steps_per_second": 5.882,
|
| 263 |
-
"step": 54
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 27.89,
|
| 267 |
-
"eval_f1": 0.888888888888889,
|
| 268 |
-
"eval_loss": 0.4827323257923126,
|
| 269 |
-
"eval_runtime": 0.5139,
|
| 270 |
-
"eval_samples_per_second": 136.21,
|
| 271 |
-
"eval_steps_per_second": 5.838,
|
| 272 |
-
"step": 56
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 28.89,
|
| 276 |
-
"eval_f1": 0.888888888888889,
|
| 277 |
-
"eval_loss": 0.4820977449417114,
|
| 278 |
-
"eval_runtime": 0.5117,
|
| 279 |
-
"eval_samples_per_second": 136.794,
|
| 280 |
-
"eval_steps_per_second": 5.863,
|
| 281 |
-
"step": 58
|
| 282 |
-
},
|
| 283 |
-
{
|
| 284 |
-
"epoch": 29.89,
|
| 285 |
-
"eval_f1": 0.888888888888889,
|
| 286 |
-
"eval_loss": 0.4817977249622345,
|
| 287 |
-
"eval_runtime": 0.5176,
|
| 288 |
-
"eval_samples_per_second": 135.231,
|
| 289 |
-
"eval_steps_per_second": 5.796,
|
| 290 |
-
"step": 60
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 30.89,
|
| 294 |
-
"eval_f1": 0.888888888888889,
|
| 295 |
-
"eval_loss": 0.4813471734523773,
|
| 296 |
-
"eval_runtime": 0.5076,
|
| 297 |
-
"eval_samples_per_second": 137.901,
|
| 298 |
-
"eval_steps_per_second": 5.91,
|
| 299 |
-
"step": 62
|
| 300 |
-
},
|
| 301 |
-
{
|
| 302 |
-
"epoch": 31.89,
|
| 303 |
-
"eval_f1": 0.888888888888889,
|
| 304 |
-
"eval_loss": 0.4812641739845276,
|
| 305 |
-
"eval_runtime": 0.5176,
|
| 306 |
-
"eval_samples_per_second": 135.242,
|
| 307 |
-
"eval_steps_per_second": 5.796,
|
| 308 |
-
"step": 64
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 32.89,
|
| 312 |
-
"eval_f1": 0.888888888888889,
|
| 313 |
-
"eval_loss": 0.48070254921913147,
|
| 314 |
-
"eval_runtime": 0.5078,
|
| 315 |
-
"eval_samples_per_second": 137.854,
|
| 316 |
-
"eval_steps_per_second": 5.908,
|
| 317 |
-
"step": 66
|
| 318 |
-
},
|
| 319 |
-
{
|
| 320 |
-
"epoch": 33.89,
|
| 321 |
-
"eval_f1": 0.888888888888889,
|
| 322 |
-
"eval_loss": 0.48028674721717834,
|
| 323 |
-
"eval_runtime": 0.5091,
|
| 324 |
-
"eval_samples_per_second": 137.488,
|
| 325 |
-
"eval_steps_per_second": 5.892,
|
| 326 |
-
"step": 68
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 34.89,
|
| 330 |
-
"eval_f1": 0.888888888888889,
|
| 331 |
-
"eval_loss": 0.4801904261112213,
|
| 332 |
-
"eval_runtime": 0.5047,
|
| 333 |
-
"eval_samples_per_second": 138.694,
|
| 334 |
-
"eval_steps_per_second": 5.944,
|
| 335 |
-
"step": 70
|
| 336 |
-
},
|
| 337 |
-
{
|
| 338 |
-
"epoch": 35.89,
|
| 339 |
-
"eval_f1": 0.888888888888889,
|
| 340 |
-
"eval_loss": 0.4799834191799164,
|
| 341 |
-
"eval_runtime": 0.5083,
|
| 342 |
-
"eval_samples_per_second": 137.704,
|
| 343 |
-
"eval_steps_per_second": 5.902,
|
| 344 |
-
"step": 72
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 36.89,
|
| 348 |
-
"eval_f1": 0.888888888888889,
|
| 349 |
-
"eval_loss": 0.47955021262168884,
|
| 350 |
-
"eval_runtime": 0.5047,
|
| 351 |
-
"eval_samples_per_second": 138.688,
|
| 352 |
-
"eval_steps_per_second": 5.944,
|
| 353 |
-
"step": 74
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 37.44,
|
| 357 |
-
"learning_rate": 8.333333333333334e-06,
|
| 358 |
-
"loss": 0.4434,
|
| 359 |
-
"step": 75
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 37.89,
|
| 363 |
-
"eval_f1": 0.888888888888889,
|
| 364 |
-
"eval_loss": 0.47954249382019043,
|
| 365 |
-
"eval_runtime": 0.5098,
|
| 366 |
-
"eval_samples_per_second": 137.317,
|
| 367 |
-
"eval_steps_per_second": 5.885,
|
| 368 |
-
"step": 76
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 38.89,
|
| 372 |
-
"eval_f1": 0.888888888888889,
|
| 373 |
-
"eval_loss": 0.47896233201026917,
|
| 374 |
-
"eval_runtime": 0.5167,
|
| 375 |
-
"eval_samples_per_second": 135.481,
|
| 376 |
-
"eval_steps_per_second": 5.806,
|
| 377 |
-
"step": 78
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 39.89,
|
| 381 |
-
"eval_f1": 0.888888888888889,
|
| 382 |
-
"eval_loss": 0.4780120849609375,
|
| 383 |
-
"eval_runtime": 0.5174,
|
| 384 |
-
"eval_samples_per_second": 135.289,
|
| 385 |
-
"eval_steps_per_second": 5.798,
|
| 386 |
-
"step": 80
|
| 387 |
-
},
|
| 388 |
-
{
|
| 389 |
-
"epoch": 40.89,
|
| 390 |
-
"eval_f1": 0.888888888888889,
|
| 391 |
-
"eval_loss": 0.4772844612598419,
|
| 392 |
-
"eval_runtime": 0.5162,
|
| 393 |
-
"eval_samples_per_second": 135.603,
|
| 394 |
-
"eval_steps_per_second": 5.812,
|
| 395 |
-
"step": 82
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 41.89,
|
| 399 |
-
"eval_f1": 0.888888888888889,
|
| 400 |
-
"eval_loss": 0.47639530897140503,
|
| 401 |
-
"eval_runtime": 0.5186,
|
| 402 |
-
"eval_samples_per_second": 134.98,
|
| 403 |
-
"eval_steps_per_second": 5.785,
|
| 404 |
-
"step": 84
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 42.89,
|
| 408 |
-
"eval_f1": 0.888888888888889,
|
| 409 |
-
"eval_loss": 0.47554805874824524,
|
| 410 |
-
"eval_runtime": 0.5172,
|
| 411 |
-
"eval_samples_per_second": 135.355,
|
| 412 |
-
"eval_steps_per_second": 5.801,
|
| 413 |
-
"step": 86
|
| 414 |
-
},
|
| 415 |
-
{
|
| 416 |
-
"epoch": 43.89,
|
| 417 |
-
"eval_f1": 0.888888888888889,
|
| 418 |
-
"eval_loss": 0.4745935797691345,
|
| 419 |
-
"eval_runtime": 0.514,
|
| 420 |
-
"eval_samples_per_second": 136.19,
|
| 421 |
-
"eval_steps_per_second": 5.837,
|
| 422 |
-
"step": 88
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 44.89,
|
| 426 |
-
"eval_f1": 0.888888888888889,
|
| 427 |
-
"eval_loss": 0.4736994504928589,
|
| 428 |
-
"eval_runtime": 0.5108,
|
| 429 |
-
"eval_samples_per_second": 137.03,
|
| 430 |
-
"eval_steps_per_second": 5.873,
|
| 431 |
-
"step": 90
|
| 432 |
-
},
|
| 433 |
-
{
|
| 434 |
-
"epoch": 45.89,
|
| 435 |
-
"eval_f1": 0.888888888888889,
|
| 436 |
-
"eval_loss": 0.4725368618965149,
|
| 437 |
-
"eval_runtime": 0.5067,
|
| 438 |
-
"eval_samples_per_second": 138.137,
|
| 439 |
-
"eval_steps_per_second": 5.92,
|
| 440 |
-
"step": 92
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 46.89,
|
| 444 |
-
"eval_f1": 0.888888888888889,
|
| 445 |
-
"eval_loss": 0.47122785449028015,
|
| 446 |
-
"eval_runtime": 0.5187,
|
| 447 |
-
"eval_samples_per_second": 134.95,
|
| 448 |
-
"eval_steps_per_second": 5.784,
|
| 449 |
-
"step": 94
|
| 450 |
-
},
|
| 451 |
-
{
|
| 452 |
-
"epoch": 47.89,
|
| 453 |
-
"eval_f1": 0.888888888888889,
|
| 454 |
-
"eval_loss": 0.4697120189666748,
|
| 455 |
-
"eval_runtime": 0.5106,
|
| 456 |
-
"eval_samples_per_second": 137.089,
|
| 457 |
-
"eval_steps_per_second": 5.875,
|
| 458 |
-
"step": 96
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 48.89,
|
| 462 |
-
"eval_f1": 0.888888888888889,
|
| 463 |
-
"eval_loss": 0.4686456322669983,
|
| 464 |
-
"eval_runtime": 0.5071,
|
| 465 |
-
"eval_samples_per_second": 138.045,
|
| 466 |
-
"eval_steps_per_second": 5.916,
|
| 467 |
-
"step": 98
|
| 468 |
-
},
|
| 469 |
-
{
|
| 470 |
-
"epoch": 49.89,
|
| 471 |
-
"learning_rate": 7.4074074074074075e-06,
|
| 472 |
-
"loss": 0.4,
|
| 473 |
-
"step": 100
|
| 474 |
-
},
|
| 475 |
-
{
|
| 476 |
-
"epoch": 49.89,
|
| 477 |
-
"eval_f1": 0.888888888888889,
|
| 478 |
-
"eval_loss": 0.46755868196487427,
|
| 479 |
-
"eval_runtime": 0.5094,
|
| 480 |
-
"eval_samples_per_second": 137.403,
|
| 481 |
-
"eval_steps_per_second": 5.889,
|
| 482 |
-
"step": 100
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 50.89,
|
| 486 |
-
"eval_f1": 0.888888888888889,
|
| 487 |
-
"eval_loss": 0.4666549265384674,
|
| 488 |
-
"eval_runtime": 0.5092,
|
| 489 |
-
"eval_samples_per_second": 137.458,
|
| 490 |
-
"eval_steps_per_second": 5.891,
|
| 491 |
-
"step": 102
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 51.89,
|
| 495 |
-
"eval_f1": 0.888888888888889,
|
| 496 |
-
"eval_loss": 0.46638351678848267,
|
| 497 |
-
"eval_runtime": 0.5124,
|
| 498 |
-
"eval_samples_per_second": 136.625,
|
| 499 |
-
"eval_steps_per_second": 5.855,
|
| 500 |
-
"step": 104
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 52.89,
|
| 504 |
-
"eval_f1": 0.888888888888889,
|
| 505 |
-
"eval_loss": 0.4666298031806946,
|
| 506 |
-
"eval_runtime": 0.5135,
|
| 507 |
-
"eval_samples_per_second": 136.326,
|
| 508 |
-
"eval_steps_per_second": 5.843,
|
| 509 |
-
"step": 106
|
| 510 |
-
},
|
| 511 |
-
{
|
| 512 |
-
"epoch": 53.89,
|
| 513 |
-
"eval_f1": 0.888888888888889,
|
| 514 |
-
"eval_loss": 0.46660488843917847,
|
| 515 |
-
"eval_runtime": 0.5175,
|
| 516 |
-
"eval_samples_per_second": 135.261,
|
| 517 |
-
"eval_steps_per_second": 5.797,
|
| 518 |
-
"step": 108
|
| 519 |
-
},
|
| 520 |
-
{
|
| 521 |
-
"epoch": 54.89,
|
| 522 |
-
"eval_f1": 0.888888888888889,
|
| 523 |
-
"eval_loss": 0.4668794870376587,
|
| 524 |
-
"eval_runtime": 0.5009,
|
| 525 |
-
"eval_samples_per_second": 139.747,
|
| 526 |
-
"eval_steps_per_second": 5.989,
|
| 527 |
-
"step": 110
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 55.89,
|
| 531 |
-
"eval_f1": 0.888888888888889,
|
| 532 |
-
"eval_loss": 0.46678298711776733,
|
| 533 |
-
"eval_runtime": 0.5078,
|
| 534 |
-
"eval_samples_per_second": 137.844,
|
| 535 |
-
"eval_steps_per_second": 5.908,
|
| 536 |
-
"step": 112
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 56.89,
|
| 540 |
-
"eval_f1": 0.888888888888889,
|
| 541 |
-
"eval_loss": 0.46638283133506775,
|
| 542 |
-
"eval_runtime": 0.5157,
|
| 543 |
-
"eval_samples_per_second": 135.74,
|
| 544 |
-
"eval_steps_per_second": 5.817,
|
| 545 |
-
"step": 114
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 57.89,
|
| 549 |
-
"eval_f1": 0.888888888888889,
|
| 550 |
-
"eval_loss": 0.46611693501472473,
|
| 551 |
-
"eval_runtime": 0.5257,
|
| 552 |
-
"eval_samples_per_second": 133.154,
|
| 553 |
-
"eval_steps_per_second": 5.707,
|
| 554 |
-
"step": 116
|
| 555 |
-
},
|
| 556 |
-
{
|
| 557 |
-
"epoch": 58.89,
|
| 558 |
-
"eval_f1": 0.888888888888889,
|
| 559 |
-
"eval_loss": 0.4656626582145691,
|
| 560 |
-
"eval_runtime": 0.5108,
|
| 561 |
-
"eval_samples_per_second": 137.041,
|
| 562 |
-
"eval_steps_per_second": 5.873,
|
| 563 |
-
"step": 118
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 59.89,
|
| 567 |
-
"eval_f1": 0.888888888888889,
|
| 568 |
-
"eval_loss": 0.4653801918029785,
|
| 569 |
-
"eval_runtime": 0.5184,
|
| 570 |
-
"eval_samples_per_second": 135.019,
|
| 571 |
-
"eval_steps_per_second": 5.787,
|
| 572 |
-
"step": 120
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 60.89,
|
| 576 |
-
"eval_f1": 0.888888888888889,
|
| 577 |
-
"eval_loss": 0.46494531631469727,
|
| 578 |
-
"eval_runtime": 0.5265,
|
| 579 |
-
"eval_samples_per_second": 132.948,
|
| 580 |
-
"eval_steps_per_second": 5.698,
|
| 581 |
-
"step": 122
|
| 582 |
-
},
|
| 583 |
-
{
|
| 584 |
-
"epoch": 61.89,
|
| 585 |
-
"eval_f1": 0.888888888888889,
|
| 586 |
-
"eval_loss": 0.4643649458885193,
|
| 587 |
-
"eval_runtime": 0.5176,
|
| 588 |
-
"eval_samples_per_second": 135.236,
|
| 589 |
-
"eval_steps_per_second": 5.796,
|
| 590 |
-
"step": 124
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 62.44,
|
| 594 |
-
"learning_rate": 6.481481481481482e-06,
|
| 595 |
-
"loss": 0.3712,
|
| 596 |
-
"step": 125
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 62.89,
|
| 600 |
-
"eval_f1": 0.888888888888889,
|
| 601 |
-
"eval_loss": 0.46361738443374634,
|
| 602 |
-
"eval_runtime": 0.5177,
|
| 603 |
-
"eval_samples_per_second": 135.21,
|
| 604 |
-
"eval_steps_per_second": 5.795,
|
| 605 |
-
"step": 126
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 63.89,
|
| 609 |
-
"eval_f1": 0.888888888888889,
|
| 610 |
-
"eval_loss": 0.46249672770500183,
|
| 611 |
-
"eval_runtime": 0.5106,
|
| 612 |
-
"eval_samples_per_second": 137.097,
|
| 613 |
-
"eval_steps_per_second": 5.876,
|
| 614 |
-
"step": 128
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 64.89,
|
| 618 |
-
"eval_f1": 0.888888888888889,
|
| 619 |
-
"eval_loss": 0.4611242711544037,
|
| 620 |
-
"eval_runtime": 0.5134,
|
| 621 |
-
"eval_samples_per_second": 136.352,
|
| 622 |
-
"eval_steps_per_second": 5.844,
|
| 623 |
-
"step": 130
|
| 624 |
-
},
|
| 625 |
-
{
|
| 626 |
-
"epoch": 65.89,
|
| 627 |
-
"eval_f1": 0.888888888888889,
|
| 628 |
-
"eval_loss": 0.4598376154899597,
|
| 629 |
-
"eval_runtime": 0.5045,
|
| 630 |
-
"eval_samples_per_second": 138.759,
|
| 631 |
-
"eval_steps_per_second": 5.947,
|
| 632 |
-
"step": 132
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 66.89,
|
| 636 |
-
"eval_f1": 0.888888888888889,
|
| 637 |
-
"eval_loss": 0.45868563652038574,
|
| 638 |
-
"eval_runtime": 0.5061,
|
| 639 |
-
"eval_samples_per_second": 138.303,
|
| 640 |
-
"eval_steps_per_second": 5.927,
|
| 641 |
-
"step": 134
|
| 642 |
-
},
|
| 643 |
-
{
|
| 644 |
-
"epoch": 67.89,
|
| 645 |
-
"eval_f1": 0.888888888888889,
|
| 646 |
-
"eval_loss": 0.4579010605812073,
|
| 647 |
-
"eval_runtime": 0.5154,
|
| 648 |
-
"eval_samples_per_second": 135.819,
|
| 649 |
-
"eval_steps_per_second": 5.821,
|
| 650 |
-
"step": 136
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 68.89,
|
| 654 |
-
"eval_f1": 0.888888888888889,
|
| 655 |
-
"eval_loss": 0.4574340879917145,
|
| 656 |
-
"eval_runtime": 0.5107,
|
| 657 |
-
"eval_samples_per_second": 137.062,
|
| 658 |
-
"eval_steps_per_second": 5.874,
|
| 659 |
-
"step": 138
|
| 660 |
-
},
|
| 661 |
-
{
|
| 662 |
-
"epoch": 69.89,
|
| 663 |
-
"eval_f1": 0.888888888888889,
|
| 664 |
-
"eval_loss": 0.45691609382629395,
|
| 665 |
-
"eval_runtime": 0.5186,
|
| 666 |
-
"eval_samples_per_second": 134.99,
|
| 667 |
-
"eval_steps_per_second": 5.785,
|
| 668 |
-
"step": 140
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 70.89,
|
| 672 |
-
"eval_f1": 0.888888888888889,
|
| 673 |
-
"eval_loss": 0.45680859684944153,
|
| 674 |
-
"eval_runtime": 0.5147,
|
| 675 |
-
"eval_samples_per_second": 136.006,
|
| 676 |
-
"eval_steps_per_second": 5.829,
|
| 677 |
-
"step": 142
|
| 678 |
-
},
|
| 679 |
-
{
|
| 680 |
-
"epoch": 71.89,
|
| 681 |
-
"eval_f1": 0.888888888888889,
|
| 682 |
-
"eval_loss": 0.4564048945903778,
|
| 683 |
-
"eval_runtime": 0.5157,
|
| 684 |
-
"eval_samples_per_second": 135.741,
|
| 685 |
-
"eval_steps_per_second": 5.817,
|
| 686 |
-
"step": 144
|
| 687 |
-
},
|
| 688 |
-
{
|
| 689 |
-
"epoch": 72.89,
|
| 690 |
-
"eval_f1": 0.888888888888889,
|
| 691 |
-
"eval_loss": 0.4561113119125366,
|
| 692 |
-
"eval_runtime": 0.5039,
|
| 693 |
-
"eval_samples_per_second": 138.917,
|
| 694 |
-
"eval_steps_per_second": 5.954,
|
| 695 |
-
"step": 146
|
| 696 |
-
},
|
| 697 |
-
{
|
| 698 |
-
"epoch": 73.89,
|
| 699 |
-
"eval_f1": 0.888888888888889,
|
| 700 |
-
"eval_loss": 0.45622721314430237,
|
| 701 |
-
"eval_runtime": 0.514,
|
| 702 |
-
"eval_samples_per_second": 136.175,
|
| 703 |
-
"eval_steps_per_second": 5.836,
|
| 704 |
-
"step": 148
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 74.89,
|
| 708 |
-
"learning_rate": 5.555555555555557e-06,
|
| 709 |
-
"loss": 0.3419,
|
| 710 |
-
"step": 150
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 74.89,
|
| 714 |
-
"eval_f1": 0.888888888888889,
|
| 715 |
-
"eval_loss": 0.4559585452079773,
|
| 716 |
-
"eval_runtime": 0.5107,
|
| 717 |
-
"eval_samples_per_second": 137.074,
|
| 718 |
-
"eval_steps_per_second": 5.875,
|
| 719 |
-
"step": 150
|
| 720 |
-
},
|
| 721 |
-
{
|
| 722 |
-
"epoch": 75.89,
|
| 723 |
-
"eval_f1": 0.888888888888889,
|
| 724 |
-
"eval_loss": 0.4556769132614136,
|
| 725 |
-
"eval_runtime": 0.5134,
|
| 726 |
-
"eval_samples_per_second": 136.339,
|
| 727 |
-
"eval_steps_per_second": 5.843,
|
| 728 |
-
"step": 152
|
| 729 |
-
},
|
| 730 |
-
{
|
| 731 |
-
"epoch": 76.89,
|
| 732 |
-
"eval_f1": 0.888888888888889,
|
| 733 |
-
"eval_loss": 0.45586857199668884,
|
| 734 |
-
"eval_runtime": 0.5119,
|
| 735 |
-
"eval_samples_per_second": 136.757,
|
| 736 |
-
"eval_steps_per_second": 5.861,
|
| 737 |
-
"step": 154
|
| 738 |
-
},
|
| 739 |
-
{
|
| 740 |
-
"epoch": 77.89,
|
| 741 |
-
"eval_f1": 0.888888888888889,
|
| 742 |
-
"eval_loss": 0.45571208000183105,
|
| 743 |
-
"eval_runtime": 0.5116,
|
| 744 |
-
"eval_samples_per_second": 136.818,
|
| 745 |
-
"eval_steps_per_second": 5.864,
|
| 746 |
-
"step": 156
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 78.89,
|
| 750 |
-
"eval_f1": 0.888888888888889,
|
| 751 |
-
"eval_loss": 0.4558698832988739,
|
| 752 |
-
"eval_runtime": 0.5082,
|
| 753 |
-
"eval_samples_per_second": 137.744,
|
| 754 |
-
"eval_steps_per_second": 5.903,
|
| 755 |
-
"step": 158
|
| 756 |
-
},
|
| 757 |
-
{
|
| 758 |
-
"epoch": 79.89,
|
| 759 |
-
"eval_f1": 0.888888888888889,
|
| 760 |
-
"eval_loss": 0.4559585154056549,
|
| 761 |
-
"eval_runtime": 0.5127,
|
| 762 |
-
"eval_samples_per_second": 136.542,
|
| 763 |
-
"eval_steps_per_second": 5.852,
|
| 764 |
-
"step": 160
|
| 765 |
-
},
|
| 766 |
-
{
|
| 767 |
-
"epoch": 80.89,
|
| 768 |
-
"eval_f1": 0.888888888888889,
|
| 769 |
-
"eval_loss": 0.4561022222042084,
|
| 770 |
-
"eval_runtime": 0.5136,
|
| 771 |
-
"eval_samples_per_second": 136.283,
|
| 772 |
-
"eval_steps_per_second": 5.841,
|
| 773 |
-
"step": 162
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 81.89,
|
| 777 |
-
"eval_f1": 0.888888888888889,
|
| 778 |
-
"eval_loss": 0.4561418294906616,
|
| 779 |
-
"eval_runtime": 0.5185,
|
| 780 |
-
"eval_samples_per_second": 135.002,
|
| 781 |
-
"eval_steps_per_second": 5.786,
|
| 782 |
-
"step": 164
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 82.89,
|
| 786 |
-
"eval_f1": 0.888888888888889,
|
| 787 |
-
"eval_loss": 0.45633062720298767,
|
| 788 |
-
"eval_runtime": 0.5583,
|
| 789 |
-
"eval_samples_per_second": 125.383,
|
| 790 |
-
"eval_steps_per_second": 5.374,
|
| 791 |
-
"step": 166
|
| 792 |
-
},
|
| 793 |
-
{
|
| 794 |
-
"epoch": 83.89,
|
| 795 |
-
"eval_f1": 0.896,
|
| 796 |
-
"eval_loss": 0.45609140396118164,
|
| 797 |
-
"eval_runtime": 0.5096,
|
| 798 |
-
"eval_samples_per_second": 137.376,
|
| 799 |
-
"eval_steps_per_second": 5.888,
|
| 800 |
-
"step": 168
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 84.89,
|
| 804 |
-
"eval_f1": 0.896,
|
| 805 |
-
"eval_loss": 0.4557640850543976,
|
| 806 |
-
"eval_runtime": 0.5089,
|
| 807 |
-
"eval_samples_per_second": 137.544,
|
| 808 |
-
"eval_steps_per_second": 5.895,
|
| 809 |
-
"step": 170
|
| 810 |
-
},
|
| 811 |
-
{
|
| 812 |
-
"epoch": 85.89,
|
| 813 |
-
"eval_f1": 0.896,
|
| 814 |
-
"eval_loss": 0.4552680552005768,
|
| 815 |
-
"eval_runtime": 0.5124,
|
| 816 |
-
"eval_samples_per_second": 136.601,
|
| 817 |
-
"eval_steps_per_second": 5.854,
|
| 818 |
-
"step": 172
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 86.89,
|
| 822 |
-
"eval_f1": 0.896,
|
| 823 |
-
"eval_loss": 0.45485004782676697,
|
| 824 |
-
"eval_runtime": 0.5155,
|
| 825 |
-
"eval_samples_per_second": 135.799,
|
| 826 |
-
"eval_steps_per_second": 5.82,
|
| 827 |
-
"step": 174
|
| 828 |
-
},
|
| 829 |
-
{
|
| 830 |
-
"epoch": 87.44,
|
| 831 |
-
"learning_rate": 4.62962962962963e-06,
|
| 832 |
-
"loss": 0.3221,
|
| 833 |
-
"step": 175
|
| 834 |
-
},
|
| 835 |
-
{
|
| 836 |
-
"epoch": 87.89,
|
| 837 |
-
"eval_f1": 0.896,
|
| 838 |
-
"eval_loss": 0.45451802015304565,
|
| 839 |
-
"eval_runtime": 0.5105,
|
| 840 |
-
"eval_samples_per_second": 137.109,
|
| 841 |
-
"eval_steps_per_second": 5.876,
|
| 842 |
-
"step": 176
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 88.89,
|
| 846 |
-
"eval_f1": 0.896,
|
| 847 |
-
"eval_loss": 0.4541896879673004,
|
| 848 |
-
"eval_runtime": 0.5149,
|
| 849 |
-
"eval_samples_per_second": 135.937,
|
| 850 |
-
"eval_steps_per_second": 5.826,
|
| 851 |
-
"step": 178
|
| 852 |
-
},
|
| 853 |
-
{
|
| 854 |
-
"epoch": 89.89,
|
| 855 |
-
"eval_f1": 0.896,
|
| 856 |
-
"eval_loss": 0.45373955368995667,
|
| 857 |
-
"eval_runtime": 0.5183,
|
| 858 |
-
"eval_samples_per_second": 135.051,
|
| 859 |
-
"eval_steps_per_second": 5.788,
|
| 860 |
-
"step": 180
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 90.89,
|
| 864 |
-
"eval_f1": 0.896,
|
| 865 |
-
"eval_loss": 0.4535920023918152,
|
| 866 |
-
"eval_runtime": 0.5109,
|
| 867 |
-
"eval_samples_per_second": 137.024,
|
| 868 |
-
"eval_steps_per_second": 5.872,
|
| 869 |
-
"step": 182
|
| 870 |
-
},
|
| 871 |
-
{
|
| 872 |
-
"epoch": 91.89,
|
| 873 |
-
"eval_f1": 0.896,
|
| 874 |
-
"eval_loss": 0.45348674058914185,
|
| 875 |
-
"eval_runtime": 0.5161,
|
| 876 |
-
"eval_samples_per_second": 135.628,
|
| 877 |
-
"eval_steps_per_second": 5.813,
|
| 878 |
-
"step": 184
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 92.89,
|
| 882 |
-
"eval_f1": 0.896,
|
| 883 |
-
"eval_loss": 0.45325374603271484,
|
| 884 |
-
"eval_runtime": 0.5128,
|
| 885 |
-
"eval_samples_per_second": 136.516,
|
| 886 |
-
"eval_steps_per_second": 5.851,
|
| 887 |
-
"step": 186
|
| 888 |
-
},
|
| 889 |
-
{
|
| 890 |
-
"epoch": 93.89,
|
| 891 |
-
"eval_f1": 0.896,
|
| 892 |
-
"eval_loss": 0.4530419111251831,
|
| 893 |
-
"eval_runtime": 0.5198,
|
| 894 |
-
"eval_samples_per_second": 134.677,
|
| 895 |
-
"eval_steps_per_second": 5.772,
|
| 896 |
-
"step": 188
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 94.89,
|
| 900 |
-
"eval_f1": 0.896,
|
| 901 |
-
"eval_loss": 0.452932745218277,
|
| 902 |
-
"eval_runtime": 0.5123,
|
| 903 |
-
"eval_samples_per_second": 136.651,
|
| 904 |
-
"eval_steps_per_second": 5.856,
|
| 905 |
-
"step": 190
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 95.89,
|
| 909 |
-
"eval_f1": 0.896,
|
| 910 |
-
"eval_loss": 0.45291921496391296,
|
| 911 |
-
"eval_runtime": 0.5108,
|
| 912 |
-
"eval_samples_per_second": 137.03,
|
| 913 |
-
"eval_steps_per_second": 5.873,
|
| 914 |
-
"step": 192
|
| 915 |
-
},
|
| 916 |
-
{
|
| 917 |
-
"epoch": 96.89,
|
| 918 |
-
"eval_f1": 0.896,
|
| 919 |
-
"eval_loss": 0.45280784368515015,
|
| 920 |
-
"eval_runtime": 0.5125,
|
| 921 |
-
"eval_samples_per_second": 136.577,
|
| 922 |
-
"eval_steps_per_second": 5.853,
|
| 923 |
-
"step": 194
|
| 924 |
-
},
|
| 925 |
-
{
|
| 926 |
-
"epoch": 97.89,
|
| 927 |
-
"eval_f1": 0.896,
|
| 928 |
-
"eval_loss": 0.4528333246707916,
|
| 929 |
-
"eval_runtime": 0.5077,
|
| 930 |
-
"eval_samples_per_second": 137.871,
|
| 931 |
-
"eval_steps_per_second": 5.909,
|
| 932 |
-
"step": 196
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 98.89,
|
| 936 |
-
"eval_f1": 0.896,
|
| 937 |
-
"eval_loss": 0.45277661085128784,
|
| 938 |
-
"eval_runtime": 0.508,
|
| 939 |
-
"eval_samples_per_second": 137.803,
|
| 940 |
-
"eval_steps_per_second": 5.906,
|
| 941 |
-
"step": 198
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 99.89,
|
| 945 |
-
"learning_rate": 3.7037037037037037e-06,
|
| 946 |
-
"loss": 0.297,
|
| 947 |
-
"step": 200
|
| 948 |
-
},
|
| 949 |
-
{
|
| 950 |
-
"epoch": 99.89,
|
| 951 |
-
"eval_f1": 0.896,
|
| 952 |
-
"eval_loss": 0.45276370644569397,
|
| 953 |
-
"eval_runtime": 0.5103,
|
| 954 |
-
"eval_samples_per_second": 137.186,
|
| 955 |
-
"eval_steps_per_second": 5.879,
|
| 956 |
-
"step": 200
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 100.89,
|
| 960 |
-
"eval_f1": 0.896,
|
| 961 |
-
"eval_loss": 0.4528166651725769,
|
| 962 |
-
"eval_runtime": 0.5087,
|
| 963 |
-
"eval_samples_per_second": 137.613,
|
| 964 |
-
"eval_steps_per_second": 5.898,
|
| 965 |
-
"step": 202
|
| 966 |
-
},
|
| 967 |
-
{
|
| 968 |
-
"epoch": 101.89,
|
| 969 |
-
"eval_f1": 0.896,
|
| 970 |
-
"eval_loss": 0.45277735590934753,
|
| 971 |
-
"eval_runtime": 0.5107,
|
| 972 |
-
"eval_samples_per_second": 137.056,
|
| 973 |
-
"eval_steps_per_second": 5.874,
|
| 974 |
-
"step": 204
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 102.89,
|
| 978 |
-
"eval_f1": 0.896,
|
| 979 |
-
"eval_loss": 0.4523409307003021,
|
| 980 |
-
"eval_runtime": 0.519,
|
| 981 |
-
"eval_samples_per_second": 134.862,
|
| 982 |
-
"eval_steps_per_second": 5.78,
|
| 983 |
-
"step": 206
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 103.89,
|
| 987 |
-
"eval_f1": 0.896,
|
| 988 |
-
"eval_loss": 0.4522373080253601,
|
| 989 |
-
"eval_runtime": 0.5234,
|
| 990 |
-
"eval_samples_per_second": 133.744,
|
| 991 |
-
"eval_steps_per_second": 5.732,
|
| 992 |
-
"step": 208
|
| 993 |
-
},
|
| 994 |
-
{
|
| 995 |
-
"epoch": 104.89,
|
| 996 |
-
"eval_f1": 0.896,
|
| 997 |
-
"eval_loss": 0.4519382417201996,
|
| 998 |
-
"eval_runtime": 0.5099,
|
| 999 |
-
"eval_samples_per_second": 137.277,
|
| 1000 |
-
"eval_steps_per_second": 5.883,
|
| 1001 |
-
"step": 210
|
| 1002 |
-
},
|
| 1003 |
-
{
|
| 1004 |
-
"epoch": 105.89,
|
| 1005 |
-
"eval_f1": 0.896,
|
| 1006 |
-
"eval_loss": 0.4515880048274994,
|
| 1007 |
-
"eval_runtime": 0.5111,
|
| 1008 |
-
"eval_samples_per_second": 136.947,
|
| 1009 |
-
"eval_steps_per_second": 5.869,
|
| 1010 |
-
"step": 212
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 106.89,
|
| 1014 |
-
"eval_f1": 0.896,
|
| 1015 |
-
"eval_loss": 0.4515409767627716,
|
| 1016 |
-
"eval_runtime": 0.5212,
|
| 1017 |
-
"eval_samples_per_second": 134.295,
|
| 1018 |
-
"eval_steps_per_second": 5.755,
|
| 1019 |
-
"step": 214
|
| 1020 |
-
},
|
| 1021 |
-
{
|
| 1022 |
-
"epoch": 107.89,
|
| 1023 |
-
"eval_f1": 0.896,
|
| 1024 |
-
"eval_loss": 0.45120465755462646,
|
| 1025 |
-
"eval_runtime": 0.5234,
|
| 1026 |
-
"eval_samples_per_second": 133.737,
|
| 1027 |
-
"eval_steps_per_second": 5.732,
|
| 1028 |
-
"step": 216
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 108.89,
|
| 1032 |
-
"eval_f1": 0.9032258064516129,
|
| 1033 |
-
"eval_loss": 0.45062127709388733,
|
| 1034 |
-
"eval_runtime": 0.5116,
|
| 1035 |
-
"eval_samples_per_second": 136.817,
|
| 1036 |
-
"eval_steps_per_second": 5.864,
|
| 1037 |
-
"step": 218
|
| 1038 |
-
},
|
| 1039 |
-
{
|
| 1040 |
-
"epoch": 109.89,
|
| 1041 |
-
"eval_f1": 0.9032258064516129,
|
| 1042 |
-
"eval_loss": 0.450234055519104,
|
| 1043 |
-
"eval_runtime": 0.5074,
|
| 1044 |
-
"eval_samples_per_second": 137.954,
|
| 1045 |
-
"eval_steps_per_second": 5.912,
|
| 1046 |
-
"step": 220
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 110.89,
|
| 1050 |
-
"eval_f1": 0.9032258064516129,
|
| 1051 |
-
"eval_loss": 0.45027095079421997,
|
| 1052 |
-
"eval_runtime": 0.5078,
|
| 1053 |
-
"eval_samples_per_second": 137.84,
|
| 1054 |
-
"eval_steps_per_second": 5.907,
|
| 1055 |
-
"step": 222
|
| 1056 |
-
},
|
| 1057 |
-
{
|
| 1058 |
-
"epoch": 111.89,
|
| 1059 |
-
"eval_f1": 0.9032258064516129,
|
| 1060 |
-
"eval_loss": 0.4502160847187042,
|
| 1061 |
-
"eval_runtime": 0.5143,
|
| 1062 |
-
"eval_samples_per_second": 136.105,
|
| 1063 |
-
"eval_steps_per_second": 5.833,
|
| 1064 |
-
"step": 224
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 112.44,
|
| 1068 |
-
"learning_rate": 2.7777777777777783e-06,
|
| 1069 |
-
"loss": 0.2809,
|
| 1070 |
-
"step": 225
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 112.89,
|
| 1074 |
-
"eval_f1": 0.9032258064516129,
|
| 1075 |
-
"eval_loss": 0.45027267932891846,
|
| 1076 |
-
"eval_runtime": 0.5062,
|
| 1077 |
-
"eval_samples_per_second": 138.288,
|
| 1078 |
-
"eval_steps_per_second": 5.927,
|
| 1079 |
-
"step": 226
|
| 1080 |
-
},
|
| 1081 |
-
{
|
| 1082 |
-
"epoch": 113.89,
|
| 1083 |
-
"eval_f1": 0.9032258064516129,
|
| 1084 |
-
"eval_loss": 0.45001256465911865,
|
| 1085 |
-
"eval_runtime": 0.5176,
|
| 1086 |
-
"eval_samples_per_second": 135.243,
|
| 1087 |
-
"eval_steps_per_second": 5.796,
|
| 1088 |
-
"step": 228
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 114.89,
|
| 1092 |
-
"eval_f1": 0.9032258064516129,
|
| 1093 |
-
"eval_loss": 0.44984087347984314,
|
| 1094 |
-
"eval_runtime": 0.5057,
|
| 1095 |
-
"eval_samples_per_second": 138.435,
|
| 1096 |
-
"eval_steps_per_second": 5.933,
|
| 1097 |
-
"step": 230
|
| 1098 |
-
},
|
| 1099 |
-
{
|
| 1100 |
-
"epoch": 115.89,
|
| 1101 |
-
"eval_f1": 0.9032258064516129,
|
| 1102 |
-
"eval_loss": 0.4496540129184723,
|
| 1103 |
-
"eval_runtime": 0.5087,
|
| 1104 |
-
"eval_samples_per_second": 137.605,
|
| 1105 |
-
"eval_steps_per_second": 5.897,
|
| 1106 |
-
"step": 232
|
| 1107 |
-
},
|
| 1108 |
-
{
|
| 1109 |
-
"epoch": 116.89,
|
| 1110 |
-
"eval_f1": 0.9032258064516129,
|
| 1111 |
-
"eval_loss": 0.4495893120765686,
|
| 1112 |
-
"eval_runtime": 0.5162,
|
| 1113 |
-
"eval_samples_per_second": 135.612,
|
| 1114 |
-
"eval_steps_per_second": 5.812,
|
| 1115 |
-
"step": 234
|
| 1116 |
-
},
|
| 1117 |
-
{
|
| 1118 |
-
"epoch": 117.89,
|
| 1119 |
-
"eval_f1": 0.9032258064516129,
|
| 1120 |
-
"eval_loss": 0.4495932459831238,
|
| 1121 |
-
"eval_runtime": 0.5162,
|
| 1122 |
-
"eval_samples_per_second": 135.596,
|
| 1123 |
-
"eval_steps_per_second": 5.811,
|
| 1124 |
-
"step": 236
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 118.89,
|
| 1128 |
-
"eval_f1": 0.9032258064516129,
|
| 1129 |
-
"eval_loss": 0.4495743215084076,
|
| 1130 |
-
"eval_runtime": 0.5143,
|
| 1131 |
-
"eval_samples_per_second": 136.098,
|
| 1132 |
-
"eval_steps_per_second": 5.833,
|
| 1133 |
-
"step": 238
|
| 1134 |
-
},
|
| 1135 |
-
{
|
| 1136 |
-
"epoch": 119.89,
|
| 1137 |
-
"eval_f1": 0.9032258064516129,
|
| 1138 |
-
"eval_loss": 0.4495793282985687,
|
| 1139 |
-
"eval_runtime": 0.5201,
|
| 1140 |
-
"eval_samples_per_second": 134.594,
|
| 1141 |
-
"eval_steps_per_second": 5.768,
|
| 1142 |
-
"step": 240
|
| 1143 |
-
},
|
| 1144 |
-
{
|
| 1145 |
-
"epoch": 120.89,
|
| 1146 |
-
"eval_f1": 0.9032258064516129,
|
| 1147 |
-
"eval_loss": 0.44972047209739685,
|
| 1148 |
-
"eval_runtime": 0.517,
|
| 1149 |
-
"eval_samples_per_second": 135.393,
|
| 1150 |
-
"eval_steps_per_second": 5.803,
|
| 1151 |
-
"step": 242
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 121.89,
|
| 1155 |
-
"eval_f1": 0.9032258064516129,
|
| 1156 |
-
"eval_loss": 0.4496006965637207,
|
| 1157 |
-
"eval_runtime": 0.5201,
|
| 1158 |
-
"eval_samples_per_second": 134.597,
|
| 1159 |
-
"eval_steps_per_second": 5.768,
|
| 1160 |
-
"step": 244
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 122.89,
|
| 1164 |
-
"eval_f1": 0.9032258064516129,
|
| 1165 |
-
"eval_loss": 0.4495578706264496,
|
| 1166 |
-
"eval_runtime": 0.5087,
|
| 1167 |
-
"eval_samples_per_second": 137.594,
|
| 1168 |
-
"eval_steps_per_second": 5.897,
|
| 1169 |
-
"step": 246
|
| 1170 |
-
},
|
| 1171 |
-
{
|
| 1172 |
-
"epoch": 123.89,
|
| 1173 |
-
"eval_f1": 0.9032258064516129,
|
| 1174 |
-
"eval_loss": 0.44961830973625183,
|
| 1175 |
-
"eval_runtime": 0.5091,
|
| 1176 |
-
"eval_samples_per_second": 137.492,
|
| 1177 |
-
"eval_steps_per_second": 5.893,
|
| 1178 |
-
"step": 248
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 124.89,
|
| 1182 |
-
"learning_rate": 1.8518518518518519e-06,
|
| 1183 |
-
"loss": 0.2717,
|
| 1184 |
-
"step": 250
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 124.89,
|
| 1188 |
-
"eval_f1": 0.9032258064516129,
|
| 1189 |
-
"eval_loss": 0.4495493769645691,
|
| 1190 |
-
"eval_runtime": 0.5126,
|
| 1191 |
-
"eval_samples_per_second": 136.564,
|
| 1192 |
-
"eval_steps_per_second": 5.853,
|
| 1193 |
-
"step": 250
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 125.89,
|
| 1197 |
-
"eval_f1": 0.9032258064516129,
|
| 1198 |
-
"eval_loss": 0.44953247904777527,
|
| 1199 |
-
"eval_runtime": 0.5166,
|
| 1200 |
-
"eval_samples_per_second": 135.508,
|
| 1201 |
-
"eval_steps_per_second": 5.807,
|
| 1202 |
-
"step": 252
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 126.89,
|
| 1206 |
-
"eval_f1": 0.9032258064516129,
|
| 1207 |
-
"eval_loss": 0.44945859909057617,
|
| 1208 |
-
"eval_runtime": 0.5087,
|
| 1209 |
-
"eval_samples_per_second": 137.598,
|
| 1210 |
-
"eval_steps_per_second": 5.897,
|
| 1211 |
-
"step": 254
|
| 1212 |
-
},
|
| 1213 |
-
{
|
| 1214 |
-
"epoch": 127.89,
|
| 1215 |
-
"eval_f1": 0.9032258064516129,
|
| 1216 |
-
"eval_loss": 0.4493277072906494,
|
| 1217 |
-
"eval_runtime": 0.5129,
|
| 1218 |
-
"eval_samples_per_second": 136.49,
|
| 1219 |
-
"eval_steps_per_second": 5.85,
|
| 1220 |
-
"step": 256
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 128.89,
|
| 1224 |
-
"eval_f1": 0.9032258064516129,
|
| 1225 |
-
"eval_loss": 0.4492017924785614,
|
| 1226 |
-
"eval_runtime": 0.5132,
|
| 1227 |
-
"eval_samples_per_second": 136.411,
|
| 1228 |
-
"eval_steps_per_second": 5.846,
|
| 1229 |
-
"step": 258
|
| 1230 |
-
},
|
| 1231 |
-
{
|
| 1232 |
-
"epoch": 129.89,
|
| 1233 |
-
"eval_f1": 0.9032258064516129,
|
| 1234 |
-
"eval_loss": 0.44910600781440735,
|
| 1235 |
-
"eval_runtime": 0.5158,
|
| 1236 |
-
"eval_samples_per_second": 135.724,
|
| 1237 |
-
"eval_steps_per_second": 5.817,
|
| 1238 |
-
"step": 260
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 130.89,
|
| 1242 |
-
"eval_f1": 0.9032258064516129,
|
| 1243 |
-
"eval_loss": 0.44898271560668945,
|
| 1244 |
-
"eval_runtime": 0.5171,
|
| 1245 |
-
"eval_samples_per_second": 135.359,
|
| 1246 |
-
"eval_steps_per_second": 5.801,
|
| 1247 |
-
"step": 262
|
| 1248 |
-
},
|
| 1249 |
-
{
|
| 1250 |
-
"epoch": 131.89,
|
| 1251 |
-
"eval_f1": 0.9032258064516129,
|
| 1252 |
-
"eval_loss": 0.44882574677467346,
|
| 1253 |
-
"eval_runtime": 0.5104,
|
| 1254 |
-
"eval_samples_per_second": 137.159,
|
| 1255 |
-
"eval_steps_per_second": 5.878,
|
| 1256 |
-
"step": 264
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 132.89,
|
| 1260 |
-
"eval_f1": 0.9032258064516129,
|
| 1261 |
-
"eval_loss": 0.44872909784317017,
|
| 1262 |
-
"eval_runtime": 0.5186,
|
| 1263 |
-
"eval_samples_per_second": 134.967,
|
| 1264 |
-
"eval_steps_per_second": 5.784,
|
| 1265 |
-
"step": 266
|
| 1266 |
-
},
|
| 1267 |
-
{
|
| 1268 |
-
"epoch": 133.89,
|
| 1269 |
-
"eval_f1": 0.9032258064516129,
|
| 1270 |
-
"eval_loss": 0.4485660791397095,
|
| 1271 |
-
"eval_runtime": 0.511,
|
| 1272 |
-
"eval_samples_per_second": 136.99,
|
| 1273 |
-
"eval_steps_per_second": 5.871,
|
| 1274 |
-
"step": 268
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 134.89,
|
| 1278 |
-
"eval_f1": 0.9032258064516129,
|
| 1279 |
-
"eval_loss": 0.44834819436073303,
|
| 1280 |
-
"eval_runtime": 0.5105,
|
| 1281 |
-
"eval_samples_per_second": 137.112,
|
| 1282 |
-
"eval_steps_per_second": 5.876,
|
| 1283 |
-
"step": 270
|
| 1284 |
-
},
|
| 1285 |
-
{
|
| 1286 |
-
"epoch": 135.89,
|
| 1287 |
-
"eval_f1": 0.9032258064516129,
|
| 1288 |
-
"eval_loss": 0.448197603225708,
|
| 1289 |
-
"eval_runtime": 0.5111,
|
| 1290 |
-
"eval_samples_per_second": 136.954,
|
| 1291 |
-
"eval_steps_per_second": 5.869,
|
| 1292 |
-
"step": 272
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 136.89,
|
| 1296 |
-
"eval_f1": 0.9032258064516129,
|
| 1297 |
-
"eval_loss": 0.4480949342250824,
|
| 1298 |
-
"eval_runtime": 0.5102,
|
| 1299 |
-
"eval_samples_per_second": 137.192,
|
| 1300 |
-
"eval_steps_per_second": 5.88,
|
| 1301 |
-
"step": 274
|
| 1302 |
-
},
|
| 1303 |
-
{
|
| 1304 |
-
"epoch": 137.44,
|
| 1305 |
-
"learning_rate": 9.259259259259259e-07,
|
| 1306 |
-
"loss": 0.2597,
|
| 1307 |
-
"step": 275
|
| 1308 |
-
},
|
| 1309 |
-
{
|
| 1310 |
-
"epoch": 137.89,
|
| 1311 |
-
"eval_f1": 0.9032258064516129,
|
| 1312 |
-
"eval_loss": 0.447935551404953,
|
| 1313 |
-
"eval_runtime": 0.5123,
|
| 1314 |
-
"eval_samples_per_second": 136.645,
|
| 1315 |
-
"eval_steps_per_second": 5.856,
|
| 1316 |
-
"step": 276
|
| 1317 |
-
},
|
| 1318 |
-
{
|
| 1319 |
-
"epoch": 138.89,
|
| 1320 |
-
"eval_f1": 0.9032258064516129,
|
| 1321 |
-
"eval_loss": 0.44773900508880615,
|
| 1322 |
-
"eval_runtime": 0.5066,
|
| 1323 |
-
"eval_samples_per_second": 138.167,
|
| 1324 |
-
"eval_steps_per_second": 5.921,
|
| 1325 |
-
"step": 278
|
| 1326 |
-
},
|
| 1327 |
-
{
|
| 1328 |
-
"epoch": 139.89,
|
| 1329 |
-
"eval_f1": 0.9032258064516129,
|
| 1330 |
-
"eval_loss": 0.44755300879478455,
|
| 1331 |
-
"eval_runtime": 0.5128,
|
| 1332 |
-
"eval_samples_per_second": 136.51,
|
| 1333 |
-
"eval_steps_per_second": 5.85,
|
| 1334 |
-
"step": 280
|
| 1335 |
-
},
|
| 1336 |
-
{
|
| 1337 |
-
"epoch": 140.89,
|
| 1338 |
-
"eval_f1": 0.9032258064516129,
|
| 1339 |
-
"eval_loss": 0.44738340377807617,
|
| 1340 |
-
"eval_runtime": 0.5088,
|
| 1341 |
-
"eval_samples_per_second": 137.582,
|
| 1342 |
-
"eval_steps_per_second": 5.896,
|
| 1343 |
-
"step": 282
|
| 1344 |
-
},
|
| 1345 |
-
{
|
| 1346 |
-
"epoch": 141.89,
|
| 1347 |
-
"eval_f1": 0.9032258064516129,
|
| 1348 |
-
"eval_loss": 0.4472770690917969,
|
| 1349 |
-
"eval_runtime": 0.5106,
|
| 1350 |
-
"eval_samples_per_second": 137.086,
|
| 1351 |
-
"eval_steps_per_second": 5.875,
|
| 1352 |
-
"step": 284
|
| 1353 |
-
},
|
| 1354 |
-
{
|
| 1355 |
-
"epoch": 142.89,
|
| 1356 |
-
"eval_f1": 0.9032258064516129,
|
| 1357 |
-
"eval_loss": 0.44719263911247253,
|
| 1358 |
-
"eval_runtime": 0.5111,
|
| 1359 |
-
"eval_samples_per_second": 136.971,
|
| 1360 |
-
"eval_steps_per_second": 5.87,
|
| 1361 |
-
"step": 286
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 143.89,
|
| 1365 |
-
"eval_f1": 0.9032258064516129,
|
| 1366 |
-
"eval_loss": 0.44712573289871216,
|
| 1367 |
-
"eval_runtime": 0.519,
|
| 1368 |
-
"eval_samples_per_second": 134.866,
|
| 1369 |
-
"eval_steps_per_second": 5.78,
|
| 1370 |
-
"step": 288
|
| 1371 |
-
},
|
| 1372 |
-
{
|
| 1373 |
-
"epoch": 144.89,
|
| 1374 |
-
"eval_f1": 0.9032258064516129,
|
| 1375 |
-
"eval_loss": 0.44704240560531616,
|
| 1376 |
-
"eval_runtime": 0.5125,
|
| 1377 |
-
"eval_samples_per_second": 136.589,
|
| 1378 |
-
"eval_steps_per_second": 5.854,
|
| 1379 |
-
"step": 290
|
| 1380 |
-
},
|
| 1381 |
-
{
|
| 1382 |
-
"epoch": 145.89,
|
| 1383 |
-
"eval_f1": 0.9032258064516129,
|
| 1384 |
-
"eval_loss": 0.4469843804836273,
|
| 1385 |
-
"eval_runtime": 0.5194,
|
| 1386 |
-
"eval_samples_per_second": 134.777,
|
| 1387 |
-
"eval_steps_per_second": 5.776,
|
| 1388 |
-
"step": 292
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 146.89,
|
| 1392 |
-
"eval_f1": 0.9032258064516129,
|
| 1393 |
-
"eval_loss": 0.44694092869758606,
|
| 1394 |
-
"eval_runtime": 0.5135,
|
| 1395 |
-
"eval_samples_per_second": 136.31,
|
| 1396 |
-
"eval_steps_per_second": 5.842,
|
| 1397 |
-
"step": 294
|
| 1398 |
-
},
|
| 1399 |
-
{
|
| 1400 |
-
"epoch": 147.89,
|
| 1401 |
-
"eval_f1": 0.9032258064516129,
|
| 1402 |
-
"eval_loss": 0.4469132423400879,
|
| 1403 |
-
"eval_runtime": 0.5162,
|
| 1404 |
-
"eval_samples_per_second": 135.603,
|
| 1405 |
-
"eval_steps_per_second": 5.812,
|
| 1406 |
-
"step": 296
|
| 1407 |
-
},
|
| 1408 |
-
{
|
| 1409 |
-
"epoch": 148.89,
|
| 1410 |
-
"eval_f1": 0.9032258064516129,
|
| 1411 |
-
"eval_loss": 0.44689762592315674,
|
| 1412 |
-
"eval_runtime": 0.514,
|
| 1413 |
-
"eval_samples_per_second": 136.176,
|
| 1414 |
-
"eval_steps_per_second": 5.836,
|
| 1415 |
-
"step": 298
|
| 1416 |
-
},
|
| 1417 |
-
{
|
| 1418 |
-
"epoch": 149.89,
|
| 1419 |
-
"learning_rate": 0.0,
|
| 1420 |
-
"loss": 0.2556,
|
| 1421 |
-
"step": 300
|
| 1422 |
-
},
|
| 1423 |
-
{
|
| 1424 |
-
"epoch": 149.89,
|
| 1425 |
-
"eval_f1": 0.9032258064516129,
|
| 1426 |
-
"eval_loss": 0.44689178466796875,
|
| 1427 |
-
"eval_runtime": 0.5161,
|
| 1428 |
-
"eval_samples_per_second": 135.631,
|
| 1429 |
-
"eval_steps_per_second": 5.813,
|
| 1430 |
-
"step": 300
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 149.89,
|
| 1434 |
-
"step": 300,
|
| 1435 |
-
"total_flos": 1.0435256966870508e+18,
|
| 1436 |
-
"train_loss": 0.3680222670237223,
|
| 1437 |
-
"train_runtime": 877.204,
|
| 1438 |
-
"train_samples_per_second": 47.366,
|
| 1439 |
-
"train_steps_per_second": 0.342
|
| 1440 |
}
|
| 1441 |
],
|
| 1442 |
-
"max_steps":
|
| 1443 |
-
"num_train_epochs":
|
| 1444 |
-
"total_flos": 1.
|
| 1445 |
"trial_name": null,
|
| 1446 |
"trial_params": null
|
| 1447 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.9312977099236641,
|
| 3 |
+
"best_model_checkpoint": "Cvt-finetuned-thyroid/checkpoint-36",
|
| 4 |
+
"epoch": 19.727272727272727,
|
| 5 |
+
"global_step": 40,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
+
"epoch": 0.73,
|
| 12 |
+
"eval_f1": 0.08695652173913043,
|
| 13 |
+
"eval_loss": 0.729124128818512,
|
| 14 |
+
"eval_runtime": 0.5602,
|
| 15 |
+
"eval_samples_per_second": 146.387,
|
| 16 |
+
"eval_steps_per_second": 5.356,
|
| 17 |
"step": 2
|
| 18 |
},
|
| 19 |
{
|
| 20 |
+
"epoch": 1.73,
|
| 21 |
+
"eval_f1": 0.24,
|
| 22 |
+
"eval_loss": 0.715528666973114,
|
| 23 |
+
"eval_runtime": 0.4576,
|
| 24 |
+
"eval_samples_per_second": 179.212,
|
| 25 |
+
"eval_steps_per_second": 6.557,
|
| 26 |
"step": 4
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 2.73,
|
| 30 |
+
"eval_f1": 0.4651162790697675,
|
| 31 |
+
"eval_loss": 0.6962855458259583,
|
| 32 |
+
"eval_runtime": 0.4553,
|
| 33 |
+
"eval_samples_per_second": 180.089,
|
| 34 |
+
"eval_steps_per_second": 6.589,
|
| 35 |
"step": 6
|
| 36 |
},
|
| 37 |
{
|
| 38 |
+
"epoch": 3.73,
|
| 39 |
+
"eval_f1": 0.6732673267326733,
|
| 40 |
+
"eval_loss": 0.6805053353309631,
|
| 41 |
+
"eval_runtime": 0.456,
|
| 42 |
+
"eval_samples_per_second": 179.84,
|
| 43 |
+
"eval_steps_per_second": 6.58,
|
| 44 |
"step": 8
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 4.73,
|
| 48 |
+
"eval_f1": 0.8035714285714285,
|
| 49 |
+
"eval_loss": 0.6671938896179199,
|
| 50 |
+
"eval_runtime": 0.4582,
|
| 51 |
+
"eval_samples_per_second": 178.951,
|
| 52 |
+
"eval_steps_per_second": 6.547,
|
| 53 |
"step": 10
|
| 54 |
},
|
| 55 |
{
|
| 56 |
+
"epoch": 5.73,
|
| 57 |
+
"eval_f1": 0.8666666666666666,
|
| 58 |
+
"eval_loss": 0.6562464237213135,
|
| 59 |
+
"eval_runtime": 0.4523,
|
| 60 |
+
"eval_samples_per_second": 181.301,
|
| 61 |
+
"eval_steps_per_second": 6.633,
|
| 62 |
"step": 12
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 6.73,
|
| 66 |
+
"eval_f1": 0.8688524590163934,
|
| 67 |
+
"eval_loss": 0.6468915343284607,
|
| 68 |
+
"eval_runtime": 0.4563,
|
| 69 |
+
"eval_samples_per_second": 179.688,
|
| 70 |
+
"eval_steps_per_second": 6.574,
|
| 71 |
"step": 14
|
| 72 |
},
|
| 73 |
{
|
| 74 |
+
"epoch": 7.73,
|
| 75 |
+
"eval_f1": 0.9047619047619049,
|
| 76 |
+
"eval_loss": 0.6390055418014526,
|
| 77 |
+
"eval_runtime": 0.4554,
|
| 78 |
+
"eval_samples_per_second": 180.045,
|
| 79 |
+
"eval_steps_per_second": 6.587,
|
| 80 |
"step": 16
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 8.73,
|
| 84 |
+
"eval_f1": 0.9218749999999999,
|
| 85 |
+
"eval_loss": 0.6321854591369629,
|
| 86 |
+
"eval_runtime": 0.4533,
|
| 87 |
+
"eval_samples_per_second": 180.88,
|
| 88 |
+
"eval_steps_per_second": 6.618,
|
| 89 |
"step": 18
|
| 90 |
},
|
| 91 |
{
|
| 92 |
+
"epoch": 9.73,
|
| 93 |
+
"eval_f1": 0.9218749999999999,
|
| 94 |
+
"eval_loss": 0.6260910034179688,
|
| 95 |
+
"eval_runtime": 0.4585,
|
| 96 |
+
"eval_samples_per_second": 178.853,
|
| 97 |
+
"eval_steps_per_second": 6.543,
|
| 98 |
"step": 20
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 10.73,
|
| 102 |
+
"eval_f1": 0.9147286821705426,
|
| 103 |
+
"eval_loss": 0.6207503080368042,
|
| 104 |
+
"eval_runtime": 0.4613,
|
| 105 |
+
"eval_samples_per_second": 177.741,
|
| 106 |
+
"eval_steps_per_second": 6.503,
|
| 107 |
"step": 22
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 11.73,
|
| 111 |
+
"eval_f1": 0.9147286821705426,
|
| 112 |
+
"eval_loss": 0.6160098910331726,
|
| 113 |
+
"eval_runtime": 0.4683,
|
| 114 |
+
"eval_samples_per_second": 175.09,
|
| 115 |
+
"eval_steps_per_second": 6.406,
|
| 116 |
"step": 24
|
| 117 |
},
|
| 118 |
{
|
| 119 |
+
"epoch": 12.36,
|
| 120 |
+
"learning_rate": 4.166666666666667e-06,
|
| 121 |
+
"loss": 0.8937,
|
| 122 |
"step": 25
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 12.73,
|
| 126 |
+
"eval_f1": 0.9147286821705426,
|
| 127 |
+
"eval_loss": 0.6118525862693787,
|
| 128 |
+
"eval_runtime": 0.4721,
|
| 129 |
+
"eval_samples_per_second": 173.678,
|
| 130 |
+
"eval_steps_per_second": 6.354,
|
| 131 |
"step": 26
|
| 132 |
},
|
| 133 |
{
|
| 134 |
+
"epoch": 13.73,
|
| 135 |
+
"eval_f1": 0.9147286821705426,
|
| 136 |
+
"eval_loss": 0.6081259846687317,
|
| 137 |
+
"eval_runtime": 0.4553,
|
| 138 |
+
"eval_samples_per_second": 180.118,
|
| 139 |
+
"eval_steps_per_second": 6.59,
|
| 140 |
"step": 28
|
| 141 |
},
|
| 142 |
{
|
| 143 |
+
"epoch": 14.73,
|
| 144 |
+
"eval_f1": 0.923076923076923,
|
| 145 |
+
"eval_loss": 0.6047901511192322,
|
| 146 |
+
"eval_runtime": 0.4555,
|
| 147 |
+
"eval_samples_per_second": 180.011,
|
| 148 |
+
"eval_steps_per_second": 6.586,
|
| 149 |
"step": 30
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 15.73,
|
| 153 |
+
"eval_f1": 0.923076923076923,
|
| 154 |
+
"eval_loss": 0.6022080183029175,
|
| 155 |
+
"eval_runtime": 0.4519,
|
| 156 |
+
"eval_samples_per_second": 181.441,
|
| 157 |
+
"eval_steps_per_second": 6.638,
|
| 158 |
"step": 32
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 16.73,
|
| 162 |
+
"eval_f1": 0.923076923076923,
|
| 163 |
+
"eval_loss": 0.6001817584037781,
|
| 164 |
+
"eval_runtime": 0.4544,
|
| 165 |
+
"eval_samples_per_second": 180.453,
|
| 166 |
+
"eval_steps_per_second": 6.602,
|
| 167 |
"step": 34
|
| 168 |
},
|
| 169 |
{
|
| 170 |
+
"epoch": 17.73,
|
| 171 |
+
"eval_f1": 0.9312977099236641,
|
| 172 |
+
"eval_loss": 0.5987359285354614,
|
| 173 |
+
"eval_runtime": 0.4559,
|
| 174 |
+
"eval_samples_per_second": 179.858,
|
| 175 |
+
"eval_steps_per_second": 6.58,
|
| 176 |
"step": 36
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 18.73,
|
| 180 |
+
"eval_f1": 0.9312977099236641,
|
| 181 |
+
"eval_loss": 0.5978376269340515,
|
| 182 |
+
"eval_runtime": 0.4536,
|
| 183 |
+
"eval_samples_per_second": 180.765,
|
| 184 |
+
"eval_steps_per_second": 6.613,
|
| 185 |
"step": 38
|
| 186 |
},
|
| 187 |
{
|
| 188 |
+
"epoch": 19.73,
|
| 189 |
+
"eval_f1": 0.9312977099236641,
|
| 190 |
+
"eval_loss": 0.5974743962287903,
|
| 191 |
+
"eval_runtime": 0.4555,
|
| 192 |
+
"eval_samples_per_second": 180.031,
|
| 193 |
+
"eval_steps_per_second": 6.586,
|
| 194 |
"step": 40
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 19.73,
|
| 198 |
+
"step": 40,
|
| 199 |
+
"total_flos": 1.620732679243776e+17,
|
| 200 |
+
"train_loss": 0.8578881025314331,
|
| 201 |
+
"train_runtime": 102.9255,
|
| 202 |
+
"train_samples_per_second": 63.347,
|
| 203 |
+
"train_steps_per_second": 0.389
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
}
|
| 205 |
],
|
| 206 |
+
"max_steps": 40,
|
| 207 |
+
"num_train_epochs": 20,
|
| 208 |
+
"total_flos": 1.620732679243776e+17,
|
| 209 |
"trial_name": null,
|
| 210 |
"trial_params": null
|
| 211 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3311
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98064b8c1615330c92e0bd8f4b34a86ff482df6adfeb93cfdfa4d377fa744a73
|
| 3 |
size 3311
|