root
commited on
Commit
·
aa4fb01
1
Parent(s):
ce3976c
model trained
Browse files- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +426 -168
- training_args.bin +2 -2
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1575738181
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20fdb5c2c4323d207318908a0c1e872739153bd4b1b3f357b7fd7f3270225435
|
| 3 |
size 1575738181
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 787895775
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1ae5619b0c0979a19f537880b5641b0f5fcdc9c9ff5dfa462f2489c466ddfeb
|
| 3 |
size 787895775
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14503
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad5e0a56dbbaa43f9d8745ae549f2f2bd42a7d1d9f6dc982adb3cb0b96d71acb
|
| 3 |
size 14503
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4694a64b29aceba9bc01f87f0f202f1ee1c1bba0130df96038ab6f4b43c4c590
|
| 3 |
size 623
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -10,347 +10,605 @@
|
|
| 10 |
{
|
| 11 |
"epoch": 0.11,
|
| 12 |
"learning_rate": 3e-06,
|
| 13 |
-
"loss":
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
"learning_rate": 6e-06,
|
| 19 |
-
"loss":
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
| 24 |
"learning_rate": 9e-06,
|
| 25 |
-
"loss": 3.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
"learning_rate": 1.2e-05,
|
| 31 |
-
"loss": 3.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
-
"epoch": 0.
|
| 36 |
"learning_rate": 1.5e-05,
|
| 37 |
-
"loss":
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
"learning_rate": 1.8e-05,
|
| 43 |
-
"loss":
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
"learning_rate": 2.1e-05,
|
| 49 |
-
"loss":
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
"learning_rate": 2.4e-05,
|
| 55 |
-
"loss":
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
-
"epoch": 0.
|
| 60 |
"learning_rate": 2.7000000000000002e-05,
|
| 61 |
-
"loss": 2.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch": 1.
|
| 66 |
"learning_rate": 3e-05,
|
| 67 |
-
"loss": 2.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch": 1.
|
| 72 |
-
"eval_loss": 2.
|
| 73 |
-
"eval_runtime":
|
| 74 |
-
"eval_samples_per_second":
|
| 75 |
-
"eval_steps_per_second": 0.
|
| 76 |
"step": 100
|
| 77 |
},
|
| 78 |
{
|
| 79 |
-
"epoch": 1.
|
| 80 |
-
"learning_rate": 2.
|
| 81 |
-
"loss": 2.
|
| 82 |
"step": 110
|
| 83 |
},
|
| 84 |
{
|
| 85 |
-
"epoch": 1.
|
| 86 |
-
"learning_rate": 2.
|
| 87 |
-
"loss": 2.
|
| 88 |
"step": 120
|
| 89 |
},
|
| 90 |
{
|
| 91 |
-
"epoch": 1.
|
| 92 |
-
"learning_rate": 2.
|
| 93 |
-
"loss": 2.
|
| 94 |
"step": 130
|
| 95 |
},
|
| 96 |
{
|
| 97 |
-
"epoch": 1.
|
| 98 |
-
"learning_rate": 2.
|
| 99 |
-
"loss":
|
| 100 |
"step": 140
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 1.
|
| 104 |
-
"learning_rate": 2.
|
| 105 |
-
"loss":
|
| 106 |
"step": 150
|
| 107 |
},
|
| 108 |
{
|
| 109 |
-
"epoch": 1.
|
| 110 |
-
"learning_rate": 2.
|
| 111 |
-
"loss":
|
| 112 |
"step": 160
|
| 113 |
},
|
| 114 |
{
|
| 115 |
-
"epoch": 1.
|
| 116 |
-
"learning_rate": 2.
|
| 117 |
-
"loss":
|
| 118 |
"step": 170
|
| 119 |
},
|
| 120 |
{
|
| 121 |
-
"epoch": 1.
|
| 122 |
-
"learning_rate": 2.
|
| 123 |
-
"loss":
|
| 124 |
"step": 180
|
| 125 |
},
|
| 126 |
{
|
| 127 |
-
"epoch": 2.
|
| 128 |
-
"learning_rate": 2.
|
| 129 |
-
"loss":
|
| 130 |
"step": 190
|
| 131 |
},
|
| 132 |
{
|
| 133 |
-
"epoch": 2.
|
| 134 |
-
"learning_rate": 2.
|
| 135 |
-
"loss":
|
| 136 |
"step": 200
|
| 137 |
},
|
| 138 |
{
|
| 139 |
-
"epoch": 2.
|
| 140 |
-
"eval_loss":
|
| 141 |
-
"eval_runtime":
|
| 142 |
-
"eval_samples_per_second": 4.
|
| 143 |
-
"eval_steps_per_second": 0.
|
| 144 |
"step": 200
|
| 145 |
},
|
| 146 |
{
|
| 147 |
-
"epoch": 2.
|
| 148 |
-
"learning_rate": 2.
|
| 149 |
-
"loss":
|
| 150 |
"step": 210
|
| 151 |
},
|
| 152 |
{
|
| 153 |
-
"epoch": 2.
|
| 154 |
-
"learning_rate": 2.
|
| 155 |
-
"loss":
|
| 156 |
"step": 220
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 2.
|
| 160 |
-
"learning_rate": 2.
|
| 161 |
-
"loss":
|
| 162 |
"step": 230
|
| 163 |
},
|
| 164 |
{
|
| 165 |
-
"epoch": 2.
|
| 166 |
-
"learning_rate": 2.
|
| 167 |
-
"loss": 1.
|
| 168 |
"step": 240
|
| 169 |
},
|
| 170 |
{
|
| 171 |
-
"epoch": 2.
|
| 172 |
-
"learning_rate": 2.
|
| 173 |
-
"loss": 1.
|
| 174 |
"step": 250
|
| 175 |
},
|
| 176 |
{
|
| 177 |
-
"epoch": 2.
|
| 178 |
-
"learning_rate": 2.
|
| 179 |
-
"loss": 1.
|
| 180 |
"step": 260
|
| 181 |
},
|
| 182 |
{
|
| 183 |
-
"epoch": 2.
|
| 184 |
-
"learning_rate": 2.
|
| 185 |
-
"loss": 1.
|
| 186 |
"step": 270
|
| 187 |
},
|
| 188 |
{
|
| 189 |
-
"epoch":
|
| 190 |
-
"learning_rate": 2.
|
| 191 |
-
"loss": 1.
|
| 192 |
"step": 280
|
| 193 |
},
|
| 194 |
{
|
| 195 |
-
"epoch": 3.
|
| 196 |
-
"learning_rate": 2.
|
| 197 |
-
"loss": 1.
|
| 198 |
"step": 290
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 3.
|
| 202 |
-
"learning_rate": 2.
|
| 203 |
-
"loss": 1.
|
| 204 |
"step": 300
|
| 205 |
},
|
| 206 |
{
|
| 207 |
-
"epoch": 3.
|
| 208 |
-
"eval_loss":
|
| 209 |
-
"eval_runtime":
|
| 210 |
-
"eval_samples_per_second":
|
| 211 |
-
"eval_steps_per_second": 0.
|
| 212 |
"step": 300
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 3.
|
| 216 |
-
"learning_rate": 2.
|
| 217 |
-
"loss": 1.
|
| 218 |
"step": 310
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch": 3.
|
| 222 |
-
"learning_rate": 2.
|
| 223 |
-
"loss": 1.
|
| 224 |
"step": 320
|
| 225 |
},
|
| 226 |
{
|
| 227 |
-
"epoch": 3.
|
| 228 |
-
"learning_rate": 2.
|
| 229 |
-
"loss": 1.
|
| 230 |
"step": 330
|
| 231 |
},
|
| 232 |
{
|
| 233 |
-
"epoch": 3.
|
| 234 |
-
"learning_rate": 2.
|
| 235 |
-
"loss": 1.
|
| 236 |
"step": 340
|
| 237 |
},
|
| 238 |
{
|
| 239 |
-
"epoch": 3.
|
| 240 |
-
"learning_rate": 2.
|
| 241 |
-
"loss": 1.
|
| 242 |
"step": 350
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch": 3.
|
| 246 |
-
"learning_rate": 2.
|
| 247 |
-
"loss": 1.
|
| 248 |
"step": 360
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch":
|
| 252 |
-
"learning_rate":
|
| 253 |
-
"loss":
|
| 254 |
"step": 370
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch": 4.
|
| 258 |
-
"learning_rate": 1.
|
| 259 |
-
"loss": 1.
|
| 260 |
"step": 380
|
| 261 |
},
|
| 262 |
{
|
| 263 |
-
"epoch": 4.
|
| 264 |
-
"learning_rate": 1.
|
| 265 |
-
"loss":
|
| 266 |
"step": 390
|
| 267 |
},
|
| 268 |
{
|
| 269 |
-
"epoch": 4.
|
| 270 |
-
"learning_rate": 1.
|
| 271 |
-
"loss":
|
| 272 |
"step": 400
|
| 273 |
},
|
| 274 |
{
|
| 275 |
-
"epoch": 4.
|
| 276 |
-
"eval_loss": 1.
|
| 277 |
-
"eval_runtime":
|
| 278 |
-
"eval_samples_per_second":
|
| 279 |
-
"eval_steps_per_second": 0.
|
| 280 |
"step": 400
|
| 281 |
},
|
| 282 |
{
|
| 283 |
-
"epoch": 4.
|
| 284 |
-
"learning_rate": 1.
|
| 285 |
-
"loss":
|
| 286 |
"step": 410
|
| 287 |
},
|
| 288 |
{
|
| 289 |
-
"epoch": 4.
|
| 290 |
-
"learning_rate": 1.
|
| 291 |
-
"loss":
|
| 292 |
"step": 420
|
| 293 |
},
|
| 294 |
{
|
| 295 |
-
"epoch": 4.
|
| 296 |
-
"learning_rate": 1.
|
| 297 |
-
"loss":
|
| 298 |
"step": 430
|
| 299 |
},
|
| 300 |
{
|
| 301 |
-
"epoch": 4.
|
| 302 |
-
"learning_rate": 1.
|
| 303 |
-
"loss":
|
| 304 |
"step": 440
|
| 305 |
},
|
| 306 |
{
|
| 307 |
-
"epoch": 4.
|
| 308 |
-
"learning_rate": 1.
|
| 309 |
-
"loss":
|
| 310 |
"step": 450
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch":
|
| 314 |
-
"learning_rate": 1.
|
| 315 |
-
"loss":
|
| 316 |
"step": 460
|
| 317 |
},
|
| 318 |
{
|
| 319 |
-
"epoch": 5.
|
| 320 |
-
"learning_rate": 1.
|
| 321 |
-
"loss":
|
| 322 |
"step": 470
|
| 323 |
},
|
| 324 |
{
|
| 325 |
-
"epoch": 5.
|
| 326 |
-
"learning_rate": 1.
|
| 327 |
-
"loss": 0.
|
| 328 |
"step": 480
|
| 329 |
},
|
| 330 |
{
|
| 331 |
-
"epoch": 5.
|
| 332 |
-
"learning_rate": 1.
|
| 333 |
-
"loss": 0.
|
| 334 |
"step": 490
|
| 335 |
},
|
| 336 |
{
|
| 337 |
-
"epoch": 5.
|
| 338 |
-
"learning_rate": 1.
|
| 339 |
-
"loss": 0.
|
| 340 |
"step": 500
|
| 341 |
},
|
| 342 |
{
|
| 343 |
-
"epoch": 5.
|
| 344 |
-
"eval_loss": 1.
|
| 345 |
-
"eval_runtime":
|
| 346 |
-
"eval_samples_per_second": 5.
|
| 347 |
-
"eval_steps_per_second": 0.
|
| 348 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 349 |
}
|
| 350 |
],
|
| 351 |
-
"max_steps":
|
| 352 |
"num_train_epochs": 10,
|
| 353 |
-
"total_flos":
|
| 354 |
"trial_name": null,
|
| 355 |
"trial_params": null
|
| 356 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.846153846153847,
|
| 5 |
+
"global_step": 896,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 10 |
{
|
| 11 |
"epoch": 0.11,
|
| 12 |
"learning_rate": 3e-06,
|
| 13 |
+
"loss": 4.0528,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 0.22,
|
| 18 |
"learning_rate": 6e-06,
|
| 19 |
+
"loss": 4.0182,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"epoch": 0.33,
|
| 24 |
"learning_rate": 9e-06,
|
| 25 |
+
"loss": 3.9239,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.44,
|
| 30 |
"learning_rate": 1.2e-05,
|
| 31 |
+
"loss": 3.5011,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
+
"epoch": 0.55,
|
| 36 |
"learning_rate": 1.5e-05,
|
| 37 |
+
"loss": 3.2438,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.66,
|
| 42 |
"learning_rate": 1.8e-05,
|
| 43 |
+
"loss": 3.1774,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.77,
|
| 48 |
"learning_rate": 2.1e-05,
|
| 49 |
+
"loss": 3.1056,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.88,
|
| 54 |
"learning_rate": 2.4e-05,
|
| 55 |
+
"loss": 3.0254,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
+
"epoch": 0.99,
|
| 60 |
"learning_rate": 2.7000000000000002e-05,
|
| 61 |
+
"loss": 2.941,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 1.1,
|
| 66 |
"learning_rate": 3e-05,
|
| 67 |
+
"loss": 2.8396,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 1.1,
|
| 72 |
+
"eval_loss": 2.859715223312378,
|
| 73 |
+
"eval_runtime": 56.6611,
|
| 74 |
+
"eval_samples_per_second": 4.73,
|
| 75 |
+
"eval_steps_per_second": 0.088,
|
| 76 |
"step": 100
|
| 77 |
},
|
| 78 |
{
|
| 79 |
+
"epoch": 1.21,
|
| 80 |
+
"learning_rate": 2.962962962962963e-05,
|
| 81 |
+
"loss": 2.7745,
|
| 82 |
"step": 110
|
| 83 |
},
|
| 84 |
{
|
| 85 |
+
"epoch": 1.32,
|
| 86 |
+
"learning_rate": 2.925925925925926e-05,
|
| 87 |
+
"loss": 2.6745,
|
| 88 |
"step": 120
|
| 89 |
},
|
| 90 |
{
|
| 91 |
+
"epoch": 1.43,
|
| 92 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 93 |
+
"loss": 2.6359,
|
| 94 |
"step": 130
|
| 95 |
},
|
| 96 |
{
|
| 97 |
+
"epoch": 1.54,
|
| 98 |
+
"learning_rate": 2.851851851851852e-05,
|
| 99 |
+
"loss": 2.5991,
|
| 100 |
"step": 140
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 1.65,
|
| 104 |
+
"learning_rate": 2.8148148148148147e-05,
|
| 105 |
+
"loss": 2.6046,
|
| 106 |
"step": 150
|
| 107 |
},
|
| 108 |
{
|
| 109 |
+
"epoch": 1.76,
|
| 110 |
+
"learning_rate": 2.777777777777778e-05,
|
| 111 |
+
"loss": 2.4419,
|
| 112 |
"step": 160
|
| 113 |
},
|
| 114 |
{
|
| 115 |
+
"epoch": 1.87,
|
| 116 |
+
"learning_rate": 2.7407407407407408e-05,
|
| 117 |
+
"loss": 2.4514,
|
| 118 |
"step": 170
|
| 119 |
},
|
| 120 |
{
|
| 121 |
+
"epoch": 1.98,
|
| 122 |
+
"learning_rate": 2.7037037037037037e-05,
|
| 123 |
+
"loss": 2.4557,
|
| 124 |
"step": 180
|
| 125 |
},
|
| 126 |
{
|
| 127 |
+
"epoch": 2.09,
|
| 128 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 129 |
+
"loss": 2.1795,
|
| 130 |
"step": 190
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"epoch": 2.2,
|
| 134 |
+
"learning_rate": 2.6296296296296296e-05,
|
| 135 |
+
"loss": 2.1014,
|
| 136 |
"step": 200
|
| 137 |
},
|
| 138 |
{
|
| 139 |
+
"epoch": 2.2,
|
| 140 |
+
"eval_loss": 2.2379915714263916,
|
| 141 |
+
"eval_runtime": 62.1156,
|
| 142 |
+
"eval_samples_per_second": 4.315,
|
| 143 |
+
"eval_steps_per_second": 0.08,
|
| 144 |
"step": 200
|
| 145 |
},
|
| 146 |
{
|
| 147 |
+
"epoch": 2.31,
|
| 148 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 149 |
+
"loss": 2.0129,
|
| 150 |
"step": 210
|
| 151 |
},
|
| 152 |
{
|
| 153 |
+
"epoch": 2.42,
|
| 154 |
+
"learning_rate": 2.5555555555555557e-05,
|
| 155 |
+
"loss": 2.0986,
|
| 156 |
"step": 220
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 2.53,
|
| 160 |
+
"learning_rate": 2.5185185185185183e-05,
|
| 161 |
+
"loss": 2.0814,
|
| 162 |
"step": 230
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 2.64,
|
| 166 |
+
"learning_rate": 2.4814814814814816e-05,
|
| 167 |
+
"loss": 1.9973,
|
| 168 |
"step": 240
|
| 169 |
},
|
| 170 |
{
|
| 171 |
+
"epoch": 2.75,
|
| 172 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 173 |
+
"loss": 1.959,
|
| 174 |
"step": 250
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"epoch": 2.86,
|
| 178 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 179 |
+
"loss": 1.9774,
|
| 180 |
"step": 260
|
| 181 |
},
|
| 182 |
{
|
| 183 |
+
"epoch": 2.97,
|
| 184 |
+
"learning_rate": 2.3703703703703703e-05,
|
| 185 |
+
"loss": 1.9267,
|
| 186 |
"step": 270
|
| 187 |
},
|
| 188 |
{
|
| 189 |
+
"epoch": 3.08,
|
| 190 |
+
"learning_rate": 2.3333333333333336e-05,
|
| 191 |
+
"loss": 1.8192,
|
| 192 |
"step": 280
|
| 193 |
},
|
| 194 |
{
|
| 195 |
+
"epoch": 3.19,
|
| 196 |
+
"learning_rate": 2.296296296296296e-05,
|
| 197 |
+
"loss": 1.6512,
|
| 198 |
"step": 290
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 3.3,
|
| 202 |
+
"learning_rate": 2.2592592592592594e-05,
|
| 203 |
+
"loss": 1.5577,
|
| 204 |
"step": 300
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 3.3,
|
| 208 |
+
"eval_loss": 2.048072576522827,
|
| 209 |
+
"eval_runtime": 55.3343,
|
| 210 |
+
"eval_samples_per_second": 4.843,
|
| 211 |
+
"eval_steps_per_second": 0.09,
|
| 212 |
"step": 300
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 3.41,
|
| 216 |
+
"learning_rate": 2.222222222222222e-05,
|
| 217 |
+
"loss": 1.5442,
|
| 218 |
"step": 310
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 3.52,
|
| 222 |
+
"learning_rate": 2.1851851851851852e-05,
|
| 223 |
+
"loss": 1.5807,
|
| 224 |
"step": 320
|
| 225 |
},
|
| 226 |
{
|
| 227 |
+
"epoch": 3.63,
|
| 228 |
+
"learning_rate": 2.148148148148148e-05,
|
| 229 |
+
"loss": 1.6009,
|
| 230 |
"step": 330
|
| 231 |
},
|
| 232 |
{
|
| 233 |
+
"epoch": 3.74,
|
| 234 |
+
"learning_rate": 2.111111111111111e-05,
|
| 235 |
+
"loss": 1.5485,
|
| 236 |
"step": 340
|
| 237 |
},
|
| 238 |
{
|
| 239 |
+
"epoch": 3.85,
|
| 240 |
+
"learning_rate": 2.074074074074074e-05,
|
| 241 |
+
"loss": 1.6185,
|
| 242 |
"step": 350
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 3.96,
|
| 246 |
+
"learning_rate": 2.0370370370370372e-05,
|
| 247 |
+
"loss": 1.5032,
|
| 248 |
"step": 360
|
| 249 |
},
|
| 250 |
{
|
| 251 |
+
"epoch": 4.07,
|
| 252 |
+
"learning_rate": 1.9999999999999998e-05,
|
| 253 |
+
"loss": 1.2841,
|
| 254 |
"step": 370
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 4.18,
|
| 258 |
+
"learning_rate": 1.962962962962963e-05,
|
| 259 |
+
"loss": 1.2237,
|
| 260 |
"step": 380
|
| 261 |
},
|
| 262 |
{
|
| 263 |
+
"epoch": 4.29,
|
| 264 |
+
"learning_rate": 1.925925925925926e-05,
|
| 265 |
+
"loss": 1.2177,
|
| 266 |
"step": 390
|
| 267 |
},
|
| 268 |
{
|
| 269 |
+
"epoch": 4.4,
|
| 270 |
+
"learning_rate": 1.888888888888889e-05,
|
| 271 |
+
"loss": 1.2009,
|
| 272 |
"step": 400
|
| 273 |
},
|
| 274 |
{
|
| 275 |
+
"epoch": 4.4,
|
| 276 |
+
"eval_loss": 1.8517041206359863,
|
| 277 |
+
"eval_runtime": 56.0224,
|
| 278 |
+
"eval_samples_per_second": 4.784,
|
| 279 |
+
"eval_steps_per_second": 0.089,
|
| 280 |
"step": 400
|
| 281 |
},
|
| 282 |
{
|
| 283 |
+
"epoch": 4.51,
|
| 284 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 285 |
+
"loss": 1.1844,
|
| 286 |
"step": 410
|
| 287 |
},
|
| 288 |
{
|
| 289 |
+
"epoch": 4.62,
|
| 290 |
+
"learning_rate": 1.814814814814815e-05,
|
| 291 |
+
"loss": 1.2252,
|
| 292 |
"step": 420
|
| 293 |
},
|
| 294 |
{
|
| 295 |
+
"epoch": 4.73,
|
| 296 |
+
"learning_rate": 1.7777777777777777e-05,
|
| 297 |
+
"loss": 1.1829,
|
| 298 |
"step": 430
|
| 299 |
},
|
| 300 |
{
|
| 301 |
+
"epoch": 4.84,
|
| 302 |
+
"learning_rate": 1.740740740740741e-05,
|
| 303 |
+
"loss": 1.177,
|
| 304 |
"step": 440
|
| 305 |
},
|
| 306 |
{
|
| 307 |
+
"epoch": 4.95,
|
| 308 |
+
"learning_rate": 1.7037037037037035e-05,
|
| 309 |
+
"loss": 1.194,
|
| 310 |
"step": 450
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 5.05,
|
| 314 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 315 |
+
"loss": 1.0947,
|
| 316 |
"step": 460
|
| 317 |
},
|
| 318 |
{
|
| 319 |
+
"epoch": 5.16,
|
| 320 |
+
"learning_rate": 1.6296296296296297e-05,
|
| 321 |
+
"loss": 1.0177,
|
| 322 |
"step": 470
|
| 323 |
},
|
| 324 |
{
|
| 325 |
+
"epoch": 5.27,
|
| 326 |
+
"learning_rate": 1.5925925925925926e-05,
|
| 327 |
+
"loss": 0.9692,
|
| 328 |
"step": 480
|
| 329 |
},
|
| 330 |
{
|
| 331 |
+
"epoch": 5.38,
|
| 332 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 333 |
+
"loss": 0.9933,
|
| 334 |
"step": 490
|
| 335 |
},
|
| 336 |
{
|
| 337 |
+
"epoch": 5.49,
|
| 338 |
+
"learning_rate": 1.5185185185185186e-05,
|
| 339 |
+
"loss": 0.9451,
|
| 340 |
"step": 500
|
| 341 |
},
|
| 342 |
{
|
| 343 |
+
"epoch": 5.49,
|
| 344 |
+
"eval_loss": 1.683161973953247,
|
| 345 |
+
"eval_runtime": 51.3302,
|
| 346 |
+
"eval_samples_per_second": 5.221,
|
| 347 |
+
"eval_steps_per_second": 0.097,
|
| 348 |
"step": 500
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 5.6,
|
| 352 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 353 |
+
"loss": 0.9674,
|
| 354 |
+
"step": 510
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 5.71,
|
| 358 |
+
"learning_rate": 1.4444444444444444e-05,
|
| 359 |
+
"loss": 0.9519,
|
| 360 |
+
"step": 520
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 5.82,
|
| 364 |
+
"learning_rate": 1.4074074074074073e-05,
|
| 365 |
+
"loss": 0.9108,
|
| 366 |
+
"step": 530
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 5.93,
|
| 370 |
+
"learning_rate": 1.3703703703703704e-05,
|
| 371 |
+
"loss": 0.8963,
|
| 372 |
+
"step": 540
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 6.04,
|
| 376 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 377 |
+
"loss": 0.8255,
|
| 378 |
+
"step": 550
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 6.15,
|
| 382 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 383 |
+
"loss": 0.773,
|
| 384 |
+
"step": 560
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 6.26,
|
| 388 |
+
"learning_rate": 1.2592592592592592e-05,
|
| 389 |
+
"loss": 0.7945,
|
| 390 |
+
"step": 570
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 6.37,
|
| 394 |
+
"learning_rate": 1.2222222222222222e-05,
|
| 395 |
+
"loss": 0.7398,
|
| 396 |
+
"step": 580
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 6.48,
|
| 400 |
+
"learning_rate": 1.1851851851851852e-05,
|
| 401 |
+
"loss": 0.7495,
|
| 402 |
+
"step": 590
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 6.59,
|
| 406 |
+
"learning_rate": 1.148148148148148e-05,
|
| 407 |
+
"loss": 0.7491,
|
| 408 |
+
"step": 600
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 6.59,
|
| 412 |
+
"eval_loss": 1.6291619539260864,
|
| 413 |
+
"eval_runtime": 56.8018,
|
| 414 |
+
"eval_samples_per_second": 4.718,
|
| 415 |
+
"eval_steps_per_second": 0.088,
|
| 416 |
+
"step": 600
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 6.7,
|
| 420 |
+
"learning_rate": 1.111111111111111e-05,
|
| 421 |
+
"loss": 0.7644,
|
| 422 |
+
"step": 610
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 6.81,
|
| 426 |
+
"learning_rate": 1.074074074074074e-05,
|
| 427 |
+
"loss": 0.7328,
|
| 428 |
+
"step": 620
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 6.92,
|
| 432 |
+
"learning_rate": 1.037037037037037e-05,
|
| 433 |
+
"loss": 0.7294,
|
| 434 |
+
"step": 630
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 7.03,
|
| 438 |
+
"learning_rate": 9.999999999999999e-06,
|
| 439 |
+
"loss": 0.6665,
|
| 440 |
+
"step": 640
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 7.14,
|
| 444 |
+
"learning_rate": 9.62962962962963e-06,
|
| 445 |
+
"loss": 0.6109,
|
| 446 |
+
"step": 650
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 7.25,
|
| 450 |
+
"learning_rate": 9.259259259259259e-06,
|
| 451 |
+
"loss": 0.6219,
|
| 452 |
+
"step": 660
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 7.36,
|
| 456 |
+
"learning_rate": 8.888888888888888e-06,
|
| 457 |
+
"loss": 0.592,
|
| 458 |
+
"step": 670
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 7.47,
|
| 462 |
+
"learning_rate": 8.518518518518517e-06,
|
| 463 |
+
"loss": 0.5722,
|
| 464 |
+
"step": 680
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 7.58,
|
| 468 |
+
"learning_rate": 8.148148148148148e-06,
|
| 469 |
+
"loss": 0.5538,
|
| 470 |
+
"step": 690
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 7.69,
|
| 474 |
+
"learning_rate": 7.777777777777777e-06,
|
| 475 |
+
"loss": 0.5546,
|
| 476 |
+
"step": 700
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 7.69,
|
| 480 |
+
"eval_loss": 1.5901862382888794,
|
| 481 |
+
"eval_runtime": 52.5596,
|
| 482 |
+
"eval_samples_per_second": 5.099,
|
| 483 |
+
"eval_steps_per_second": 0.095,
|
| 484 |
+
"step": 700
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 7.8,
|
| 488 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 489 |
+
"loss": 0.546,
|
| 490 |
+
"step": 710
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 7.91,
|
| 494 |
+
"learning_rate": 7.037037037037037e-06,
|
| 495 |
+
"loss": 0.5322,
|
| 496 |
+
"step": 720
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 8.02,
|
| 500 |
+
"learning_rate": 6.666666666666667e-06,
|
| 501 |
+
"loss": 0.588,
|
| 502 |
+
"step": 730
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 8.13,
|
| 506 |
+
"learning_rate": 6.296296296296296e-06,
|
| 507 |
+
"loss": 0.4936,
|
| 508 |
+
"step": 740
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 8.24,
|
| 512 |
+
"learning_rate": 5.925925925925926e-06,
|
| 513 |
+
"loss": 0.4625,
|
| 514 |
+
"step": 750
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 8.35,
|
| 518 |
+
"learning_rate": 5.555555555555555e-06,
|
| 519 |
+
"loss": 0.4577,
|
| 520 |
+
"step": 760
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 8.46,
|
| 524 |
+
"learning_rate": 5.185185185185185e-06,
|
| 525 |
+
"loss": 0.4619,
|
| 526 |
+
"step": 770
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 8.57,
|
| 530 |
+
"learning_rate": 4.814814814814815e-06,
|
| 531 |
+
"loss": 0.5216,
|
| 532 |
+
"step": 780
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 8.68,
|
| 536 |
+
"learning_rate": 4.444444444444444e-06,
|
| 537 |
+
"loss": 0.4771,
|
| 538 |
+
"step": 790
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 8.79,
|
| 542 |
+
"learning_rate": 4.074074074074074e-06,
|
| 543 |
+
"loss": 0.452,
|
| 544 |
+
"step": 800
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 8.79,
|
| 548 |
+
"eval_loss": 1.4902839660644531,
|
| 549 |
+
"eval_runtime": 54.4822,
|
| 550 |
+
"eval_samples_per_second": 4.919,
|
| 551 |
+
"eval_steps_per_second": 0.092,
|
| 552 |
+
"step": 800
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 8.9,
|
| 556 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 557 |
+
"loss": 0.4645,
|
| 558 |
+
"step": 810
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 9.01,
|
| 562 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 563 |
+
"loss": 0.4459,
|
| 564 |
+
"step": 820
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 9.12,
|
| 568 |
+
"learning_rate": 2.962962962962963e-06,
|
| 569 |
+
"loss": 0.4157,
|
| 570 |
+
"step": 830
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 9.23,
|
| 574 |
+
"learning_rate": 2.5925925925925925e-06,
|
| 575 |
+
"loss": 0.3948,
|
| 576 |
+
"step": 840
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 9.34,
|
| 580 |
+
"learning_rate": 2.222222222222222e-06,
|
| 581 |
+
"loss": 0.4186,
|
| 582 |
+
"step": 850
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 9.45,
|
| 586 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 587 |
+
"loss": 0.3841,
|
| 588 |
+
"step": 860
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 9.56,
|
| 592 |
+
"learning_rate": 1.4814814814814815e-06,
|
| 593 |
+
"loss": 0.4086,
|
| 594 |
+
"step": 870
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 9.67,
|
| 598 |
+
"learning_rate": 1.111111111111111e-06,
|
| 599 |
+
"loss": 0.4018,
|
| 600 |
+
"step": 880
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 9.78,
|
| 604 |
+
"learning_rate": 7.407407407407407e-07,
|
| 605 |
+
"loss": 0.405,
|
| 606 |
+
"step": 890
|
| 607 |
}
|
| 608 |
],
|
| 609 |
+
"max_steps": 910,
|
| 610 |
"num_train_epochs": 10,
|
| 611 |
+
"total_flos": 4163675271142080.0,
|
| 612 |
"trial_name": null,
|
| 613 |
"trial_params": null
|
| 614 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f5cae42c017fd963e96f9cfd9dfe587553a09e551e3e4633e4e56493842b807
|
| 3 |
+
size 3311
|