Training in progress, step 150, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 205573472
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae644db6dced204b3d4a7fc41f2488b972585d9fe6da1b714504390dc3f3276c
|
| 3 |
size 205573472
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 411372650
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2763dbdc475b39f3629e485ec0f7fe8dd8f62d1750d0a500141831377c4ebbc9
|
| 3 |
size 411372650
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14308
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1af5cedfa33a695e8e556db1ebc9eecca4a98e863bc6bdc732784653406a6d13
|
| 3 |
size 14308
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5f37e21fbe85e09b136734aed2deb2ce642b5bd3d64c65196a2c110d8c5ff3a
|
| 3 |
size 1256
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,2046 +1,125 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./output/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 150,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch":
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 5.500000000000001e-06,
|
| 15 |
-
"loss":
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch":
|
| 20 |
-
"grad_norm":
|
| 21 |
"learning_rate": 1.1000000000000001e-05,
|
| 22 |
-
"loss":
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch":
|
| 27 |
-
"grad_norm": 11.
|
| 28 |
"learning_rate": 1.65e-05,
|
| 29 |
-
"loss":
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
"learning_rate": 2.2000000000000003e-05,
|
| 36 |
-
"loss":
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
"learning_rate": 2.75e-05,
|
| 43 |
-
"loss":
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
"learning_rate": 3.3e-05,
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm":
|
| 56 |
"learning_rate": 3.85e-05,
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
"learning_rate": 4.4000000000000006e-05,
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm":
|
| 70 |
"learning_rate": 4.9500000000000004e-05,
|
| 71 |
-
"loss":
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
"learning_rate": 5.5e-05,
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
"learning_rate": 5.4999434791355066e-05,
|
| 85 |
-
"loss":
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
"learning_rate": 5.4997739188653784e-05,
|
| 92 |
-
"loss":
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm":
|
| 98 |
"learning_rate": 5.4994913261595724e-05,
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm":
|
| 105 |
"learning_rate": 5.49909571263437e-05,
|
| 106 |
-
"loss":
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
"learning_rate": 5.498587094551892e-05,
|
| 113 |
-
"loss":
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"eval_loss": 0.
|
| 119 |
-
"eval_runtime":
|
| 120 |
-
"eval_samples_per_second":
|
| 121 |
-
"eval_steps_per_second":
|
| 122 |
"step": 150
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.001010496532733772,
|
| 126 |
-
"grad_norm": 14.477323532104492,
|
| 127 |
-
"learning_rate": 5.497965492819436e-05,
|
| 128 |
-
"loss": 1.0328,
|
| 129 |
-
"step": 160
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.001073652566029633,
|
| 133 |
-
"grad_norm": 15.670412063598633,
|
| 134 |
-
"learning_rate": 5.4972309329886156e-05,
|
| 135 |
-
"loss": 1.0667,
|
| 136 |
-
"step": 170
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.0011368085993254935,
|
| 140 |
-
"grad_norm": 12.536147117614746,
|
| 141 |
-
"learning_rate": 5.496383445254307e-05,
|
| 142 |
-
"loss": 1.0397,
|
| 143 |
-
"step": 180
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.0011999646326213544,
|
| 147 |
-
"grad_norm": 16.962190628051758,
|
| 148 |
-
"learning_rate": 5.495423064453413e-05,
|
| 149 |
-
"loss": 1.084,
|
| 150 |
-
"step": 190
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.001263120665917215,
|
| 154 |
-
"grad_norm": 9.591404914855957,
|
| 155 |
-
"learning_rate": 5.4943498300634254e-05,
|
| 156 |
-
"loss": 1.0233,
|
| 157 |
-
"step": 200
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.0013262766992130758,
|
| 161 |
-
"grad_norm": 12.136938095092773,
|
| 162 |
-
"learning_rate": 5.493163786200807e-05,
|
| 163 |
-
"loss": 1.0611,
|
| 164 |
-
"step": 210
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.0013894327325089367,
|
| 168 |
-
"grad_norm": 10.48437786102295,
|
| 169 |
-
"learning_rate": 5.491864981619175e-05,
|
| 170 |
-
"loss": 1.0624,
|
| 171 |
-
"step": 220
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.0014525887658047973,
|
| 175 |
-
"grad_norm": 11.525893211364746,
|
| 176 |
-
"learning_rate": 5.4904534697073e-05,
|
| 177 |
-
"loss": 1.0334,
|
| 178 |
-
"step": 230
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.0015157447991006581,
|
| 182 |
-
"grad_norm": 8.725274085998535,
|
| 183 |
-
"learning_rate": 5.488929308486908e-05,
|
| 184 |
-
"loss": 0.9931,
|
| 185 |
-
"step": 240
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.0015789008323965187,
|
| 189 |
-
"grad_norm": 11.628905296325684,
|
| 190 |
-
"learning_rate": 5.487292560610295e-05,
|
| 191 |
-
"loss": 0.963,
|
| 192 |
-
"step": 250
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.0016420568656923796,
|
| 196 |
-
"grad_norm": 16.11012840270996,
|
| 197 |
-
"learning_rate": 5.485543293357758e-05,
|
| 198 |
-
"loss": 0.9915,
|
| 199 |
-
"step": 260
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.0017052128989882404,
|
| 203 |
-
"grad_norm": 10.494763374328613,
|
| 204 |
-
"learning_rate": 5.483681578634821e-05,
|
| 205 |
-
"loss": 1.0215,
|
| 206 |
-
"step": 270
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.001768368932284101,
|
| 210 |
-
"grad_norm": 9.800888061523438,
|
| 211 |
-
"learning_rate": 5.481707492969285e-05,
|
| 212 |
-
"loss": 1.0794,
|
| 213 |
-
"step": 280
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.0018315249655799619,
|
| 217 |
-
"grad_norm": 8.919413566589355,
|
| 218 |
-
"learning_rate": 5.479621117508079e-05,
|
| 219 |
-
"loss": 1.0463,
|
| 220 |
-
"step": 290
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.0018946809988758227,
|
| 224 |
-
"grad_norm": 14.154871940612793,
|
| 225 |
-
"learning_rate": 5.477422538013927e-05,
|
| 226 |
-
"loss": 1.0916,
|
| 227 |
-
"step": 300
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.0018946809988758227,
|
| 231 |
-
"eval_loss": 0.8587029576301575,
|
| 232 |
-
"eval_runtime": 47.5564,
|
| 233 |
-
"eval_samples_per_second": 10.514,
|
| 234 |
-
"eval_steps_per_second": 10.514,
|
| 235 |
-
"step": 300
|
| 236 |
-
},
|
| 237 |
-
{
|
| 238 |
-
"epoch": 0.0019578370321716833,
|
| 239 |
-
"grad_norm": 9.973740577697754,
|
| 240 |
-
"learning_rate": 5.475111844861821e-05,
|
| 241 |
-
"loss": 1.024,
|
| 242 |
-
"step": 310
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.002020993065467544,
|
| 246 |
-
"grad_norm": 15.102771759033203,
|
| 247 |
-
"learning_rate": 5.4726891330353056e-05,
|
| 248 |
-
"loss": 1.0394,
|
| 249 |
-
"step": 320
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.002084149098763405,
|
| 253 |
-
"grad_norm": 9.678414344787598,
|
| 254 |
-
"learning_rate": 5.4701545021225746e-05,
|
| 255 |
-
"loss": 1.0355,
|
| 256 |
-
"step": 330
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 0.002147305132059266,
|
| 260 |
-
"grad_norm": 16.077899932861328,
|
| 261 |
-
"learning_rate": 5.4675080563123786e-05,
|
| 262 |
-
"loss": 0.9778,
|
| 263 |
-
"step": 340
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 0.0022104611653551262,
|
| 267 |
-
"grad_norm": 11.449134826660156,
|
| 268 |
-
"learning_rate": 5.4647499043897386e-05,
|
| 269 |
-
"loss": 1.0076,
|
| 270 |
-
"step": 350
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 0.002273617198650987,
|
| 274 |
-
"grad_norm": 9.525121688842773,
|
| 275 |
-
"learning_rate": 5.461880159731476e-05,
|
| 276 |
-
"loss": 0.9715,
|
| 277 |
-
"step": 360
|
| 278 |
-
},
|
| 279 |
-
{
|
| 280 |
-
"epoch": 0.002336773231946848,
|
| 281 |
-
"grad_norm": 8.877346992492676,
|
| 282 |
-
"learning_rate": 5.4588989403015564e-05,
|
| 283 |
-
"loss": 1.0915,
|
| 284 |
-
"step": 370
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.0023999292652427087,
|
| 288 |
-
"grad_norm": 9.848546981811523,
|
| 289 |
-
"learning_rate": 5.4558063686462315e-05,
|
| 290 |
-
"loss": 1.0043,
|
| 291 |
-
"step": 380
|
| 292 |
-
},
|
| 293 |
-
{
|
| 294 |
-
"epoch": 0.0024630852985385696,
|
| 295 |
-
"grad_norm": 14.764713287353516,
|
| 296 |
-
"learning_rate": 5.4526025718890104e-05,
|
| 297 |
-
"loss": 1.0622,
|
| 298 |
-
"step": 390
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 0.00252624133183443,
|
| 302 |
-
"grad_norm": 9.830604553222656,
|
| 303 |
-
"learning_rate": 5.44928768172543e-05,
|
| 304 |
-
"loss": 1.0403,
|
| 305 |
-
"step": 400
|
| 306 |
-
},
|
| 307 |
-
{
|
| 308 |
-
"epoch": 0.002589397365130291,
|
| 309 |
-
"grad_norm": 11.697220802307129,
|
| 310 |
-
"learning_rate": 5.44586183441764e-05,
|
| 311 |
-
"loss": 1.0231,
|
| 312 |
-
"step": 410
|
| 313 |
-
},
|
| 314 |
-
{
|
| 315 |
-
"epoch": 0.0026525533984261516,
|
| 316 |
-
"grad_norm": 11.017810821533203,
|
| 317 |
-
"learning_rate": 5.442325170788806e-05,
|
| 318 |
-
"loss": 0.9498,
|
| 319 |
-
"step": 420
|
| 320 |
-
},
|
| 321 |
-
{
|
| 322 |
-
"epoch": 0.0027157094317220125,
|
| 323 |
-
"grad_norm": 9.533434867858887,
|
| 324 |
-
"learning_rate": 5.438677836217317e-05,
|
| 325 |
-
"loss": 1.0669,
|
| 326 |
-
"step": 430
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.0027788654650178733,
|
| 330 |
-
"grad_norm": 8.570756912231445,
|
| 331 |
-
"learning_rate": 5.434919980630811e-05,
|
| 332 |
-
"loss": 1.0319,
|
| 333 |
-
"step": 440
|
| 334 |
-
},
|
| 335 |
-
{
|
| 336 |
-
"epoch": 0.0028420214983137337,
|
| 337 |
-
"grad_norm": 10.60665512084961,
|
| 338 |
-
"learning_rate": 5.431051758500015e-05,
|
| 339 |
-
"loss": 0.9454,
|
| 340 |
-
"step": 450
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 0.0028420214983137337,
|
| 344 |
-
"eval_loss": 0.8439372777938843,
|
| 345 |
-
"eval_runtime": 46.3858,
|
| 346 |
-
"eval_samples_per_second": 10.779,
|
| 347 |
-
"eval_steps_per_second": 10.779,
|
| 348 |
-
"step": 450
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 0.0029051775316095946,
|
| 352 |
-
"grad_norm": 11.22940731048584,
|
| 353 |
-
"learning_rate": 5.427073328832388e-05,
|
| 354 |
-
"loss": 1.0539,
|
| 355 |
-
"step": 460
|
| 356 |
-
},
|
| 357 |
-
{
|
| 358 |
-
"epoch": 0.0029683335649054554,
|
| 359 |
-
"grad_norm": 10.116870880126953,
|
| 360 |
-
"learning_rate": 5.422984855165592e-05,
|
| 361 |
-
"loss": 1.0274,
|
| 362 |
-
"step": 470
|
| 363 |
-
},
|
| 364 |
-
{
|
| 365 |
-
"epoch": 0.0030314895982013162,
|
| 366 |
-
"grad_norm": 12.770539283752441,
|
| 367 |
-
"learning_rate": 5.418786505560766e-05,
|
| 368 |
-
"loss": 1.0253,
|
| 369 |
-
"step": 480
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.003094645631497177,
|
| 373 |
-
"grad_norm": 11.039621353149414,
|
| 374 |
-
"learning_rate": 5.414478452595617e-05,
|
| 375 |
-
"loss": 1.0877,
|
| 376 |
-
"step": 490
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"epoch": 0.0031578016647930375,
|
| 380 |
-
"grad_norm": 13.608321189880371,
|
| 381 |
-
"learning_rate": 5.4100608733573315e-05,
|
| 382 |
-
"loss": 0.965,
|
| 383 |
-
"step": 500
|
| 384 |
-
},
|
| 385 |
-
{
|
| 386 |
-
"epoch": 0.0032209576980888983,
|
| 387 |
-
"grad_norm": 10.158246994018555,
|
| 388 |
-
"learning_rate": 5.4055339494352874e-05,
|
| 389 |
-
"loss": 0.9643,
|
| 390 |
-
"step": 510
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 0.003284113731384759,
|
| 394 |
-
"grad_norm": 9.061509132385254,
|
| 395 |
-
"learning_rate": 5.400897866913597e-05,
|
| 396 |
-
"loss": 0.9278,
|
| 397 |
-
"step": 520
|
| 398 |
-
},
|
| 399 |
-
{
|
| 400 |
-
"epoch": 0.00334726976468062,
|
| 401 |
-
"grad_norm": 9.303275108337402,
|
| 402 |
-
"learning_rate": 5.3961528163634546e-05,
|
| 403 |
-
"loss": 1.0179,
|
| 404 |
-
"step": 530
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 0.003410425797976481,
|
| 408 |
-
"grad_norm": 12.215669631958008,
|
| 409 |
-
"learning_rate": 5.391298992835303e-05,
|
| 410 |
-
"loss": 0.982,
|
| 411 |
-
"step": 540
|
| 412 |
-
},
|
| 413 |
-
{
|
| 414 |
-
"epoch": 0.0034735818312723417,
|
| 415 |
-
"grad_norm": 11.75090217590332,
|
| 416 |
-
"learning_rate": 5.386336595850817e-05,
|
| 417 |
-
"loss": 1.0544,
|
| 418 |
-
"step": 550
|
| 419 |
-
},
|
| 420 |
-
{
|
| 421 |
-
"epoch": 0.003536737864568202,
|
| 422 |
-
"grad_norm": 11.43079662322998,
|
| 423 |
-
"learning_rate": 5.3812658293946995e-05,
|
| 424 |
-
"loss": 1.0241,
|
| 425 |
-
"step": 560
|
| 426 |
-
},
|
| 427 |
-
{
|
| 428 |
-
"epoch": 0.003599893897864063,
|
| 429 |
-
"grad_norm": 10.177742004394531,
|
| 430 |
-
"learning_rate": 5.376086901906299e-05,
|
| 431 |
-
"loss": 0.9771,
|
| 432 |
-
"step": 570
|
| 433 |
-
},
|
| 434 |
-
{
|
| 435 |
-
"epoch": 0.0036630499311599237,
|
| 436 |
-
"grad_norm": 15.041535377502441,
|
| 437 |
-
"learning_rate": 5.37080002627104e-05,
|
| 438 |
-
"loss": 0.9757,
|
| 439 |
-
"step": 580
|
| 440 |
-
},
|
| 441 |
-
{
|
| 442 |
-
"epoch": 0.0037262059644557846,
|
| 443 |
-
"grad_norm": 11.840357780456543,
|
| 444 |
-
"learning_rate": 5.365405419811673e-05,
|
| 445 |
-
"loss": 1.0436,
|
| 446 |
-
"step": 590
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.0037893619977516454,
|
| 450 |
-
"grad_norm": 17.893346786499023,
|
| 451 |
-
"learning_rate": 5.359903304279339e-05,
|
| 452 |
-
"loss": 1.0265,
|
| 453 |
-
"step": 600
|
| 454 |
-
},
|
| 455 |
-
{
|
| 456 |
-
"epoch": 0.0037893619977516454,
|
| 457 |
-
"eval_loss": 0.8398363590240479,
|
| 458 |
-
"eval_runtime": 47.7667,
|
| 459 |
-
"eval_samples_per_second": 10.468,
|
| 460 |
-
"eval_steps_per_second": 10.468,
|
| 461 |
-
"step": 600
|
| 462 |
-
},
|
| 463 |
-
{
|
| 464 |
-
"epoch": 0.003852518031047506,
|
| 465 |
-
"grad_norm": 9.05345630645752,
|
| 466 |
-
"learning_rate": 5.354293905844459e-05,
|
| 467 |
-
"loss": 1.0507,
|
| 468 |
-
"step": 610
|
| 469 |
-
},
|
| 470 |
-
{
|
| 471 |
-
"epoch": 0.003915674064343367,
|
| 472 |
-
"grad_norm": 7.9775872230529785,
|
| 473 |
-
"learning_rate": 5.3485774550874306e-05,
|
| 474 |
-
"loss": 0.995,
|
| 475 |
-
"step": 620
|
| 476 |
-
},
|
| 477 |
-
{
|
| 478 |
-
"epoch": 0.0039788300976392275,
|
| 479 |
-
"grad_norm": 15.064658164978027,
|
| 480 |
-
"learning_rate": 5.3427541869891556e-05,
|
| 481 |
-
"loss": 1.017,
|
| 482 |
-
"step": 630
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.004041986130935088,
|
| 486 |
-
"grad_norm": 10.115464210510254,
|
| 487 |
-
"learning_rate": 5.336824340921377e-05,
|
| 488 |
-
"loss": 0.9997,
|
| 489 |
-
"step": 640
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.004105142164230949,
|
| 493 |
-
"grad_norm": 11.037529945373535,
|
| 494 |
-
"learning_rate": 5.330788160636841e-05,
|
| 495 |
-
"loss": 1.0244,
|
| 496 |
-
"step": 650
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 0.00416829819752681,
|
| 500 |
-
"grad_norm": 14.279242515563965,
|
| 501 |
-
"learning_rate": 5.3246458942592776e-05,
|
| 502 |
-
"loss": 0.9978,
|
| 503 |
-
"step": 660
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 0.004231454230822671,
|
| 507 |
-
"grad_norm": 12.71838092803955,
|
| 508 |
-
"learning_rate": 5.318397794273199e-05,
|
| 509 |
-
"loss": 0.9989,
|
| 510 |
-
"step": 670
|
| 511 |
-
},
|
| 512 |
-
{
|
| 513 |
-
"epoch": 0.004294610264118532,
|
| 514 |
-
"grad_norm": 12.012393951416016,
|
| 515 |
-
"learning_rate": 5.312044117513524e-05,
|
| 516 |
-
"loss": 1.0078,
|
| 517 |
-
"step": 680
|
| 518 |
-
},
|
| 519 |
-
{
|
| 520 |
-
"epoch": 0.004357766297414392,
|
| 521 |
-
"grad_norm": 15.37903118133545,
|
| 522 |
-
"learning_rate": 5.305585125155018e-05,
|
| 523 |
-
"loss": 1.0038,
|
| 524 |
-
"step": 690
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.0044209223307102525,
|
| 528 |
-
"grad_norm": 9.483263969421387,
|
| 529 |
-
"learning_rate": 5.29902108270156e-05,
|
| 530 |
-
"loss": 1.015,
|
| 531 |
-
"step": 700
|
| 532 |
-
},
|
| 533 |
-
{
|
| 534 |
-
"epoch": 0.004484078364006113,
|
| 535 |
-
"grad_norm": 10.519631385803223,
|
| 536 |
-
"learning_rate": 5.2923522599752245e-05,
|
| 537 |
-
"loss": 1.0532,
|
| 538 |
-
"step": 710
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 0.004547234397301974,
|
| 542 |
-
"grad_norm": 10.06659984588623,
|
| 543 |
-
"learning_rate": 5.2855789311051945e-05,
|
| 544 |
-
"loss": 0.9932,
|
| 545 |
-
"step": 720
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 0.004610390430597835,
|
| 549 |
-
"grad_norm": 10.307997703552246,
|
| 550 |
-
"learning_rate": 5.27870137451649e-05,
|
| 551 |
-
"loss": 1.0364,
|
| 552 |
-
"step": 730
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 0.004673546463893696,
|
| 556 |
-
"grad_norm": 10.062649726867676,
|
| 557 |
-
"learning_rate": 5.2717198729185245e-05,
|
| 558 |
-
"loss": 0.9269,
|
| 559 |
-
"step": 740
|
| 560 |
-
},
|
| 561 |
-
{
|
| 562 |
-
"epoch": 0.004736702497189557,
|
| 563 |
-
"grad_norm": 13.508529663085938,
|
| 564 |
-
"learning_rate": 5.264634713293485e-05,
|
| 565 |
-
"loss": 0.9577,
|
| 566 |
-
"step": 750
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.004736702497189557,
|
| 570 |
-
"eval_loss": 0.8395355343818665,
|
| 571 |
-
"eval_runtime": 47.7365,
|
| 572 |
-
"eval_samples_per_second": 10.474,
|
| 573 |
-
"eval_steps_per_second": 10.474,
|
| 574 |
-
"step": 750
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 0.0047998585304854175,
|
| 578 |
-
"grad_norm": 12.114377975463867,
|
| 579 |
-
"learning_rate": 5.2574461868845316e-05,
|
| 580 |
-
"loss": 1.0364,
|
| 581 |
-
"step": 760
|
| 582 |
-
},
|
| 583 |
-
{
|
| 584 |
-
"epoch": 0.004863014563781278,
|
| 585 |
-
"grad_norm": 9.41462516784668,
|
| 586 |
-
"learning_rate": 5.2501545891838315e-05,
|
| 587 |
-
"loss": 1.0971,
|
| 588 |
-
"step": 770
|
| 589 |
-
},
|
| 590 |
-
{
|
| 591 |
-
"epoch": 0.004926170597077139,
|
| 592 |
-
"grad_norm": 12.119346618652344,
|
| 593 |
-
"learning_rate": 5.242760219920405e-05,
|
| 594 |
-
"loss": 1.0022,
|
| 595 |
-
"step": 780
|
| 596 |
-
},
|
| 597 |
-
{
|
| 598 |
-
"epoch": 0.004989326630372999,
|
| 599 |
-
"grad_norm": 11.057748794555664,
|
| 600 |
-
"learning_rate": 5.235263383047812e-05,
|
| 601 |
-
"loss": 1.0155,
|
| 602 |
-
"step": 790
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 0.00505248266366886,
|
| 606 |
-
"grad_norm": 13.1547212600708,
|
| 607 |
-
"learning_rate": 5.2276643867316525e-05,
|
| 608 |
-
"loss": 1.0677,
|
| 609 |
-
"step": 800
|
| 610 |
-
},
|
| 611 |
-
{
|
| 612 |
-
"epoch": 0.005115638696964721,
|
| 613 |
-
"grad_norm": 10.418506622314453,
|
| 614 |
-
"learning_rate": 5.219963543336902e-05,
|
| 615 |
-
"loss": 0.9545,
|
| 616 |
-
"step": 810
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 0.005178794730260582,
|
| 620 |
-
"grad_norm": 11.536872863769531,
|
| 621 |
-
"learning_rate": 5.212161169415071e-05,
|
| 622 |
-
"loss": 0.9849,
|
| 623 |
-
"step": 820
|
| 624 |
-
},
|
| 625 |
-
{
|
| 626 |
-
"epoch": 0.0052419507635564425,
|
| 627 |
-
"grad_norm": 8.616244316101074,
|
| 628 |
-
"learning_rate": 5.204257585691191e-05,
|
| 629 |
-
"loss": 0.97,
|
| 630 |
-
"step": 830
|
| 631 |
-
},
|
| 632 |
-
{
|
| 633 |
-
"epoch": 0.005305106796852303,
|
| 634 |
-
"grad_norm": 9.628042221069336,
|
| 635 |
-
"learning_rate": 5.196253117050633e-05,
|
| 636 |
-
"loss": 1.0998,
|
| 637 |
-
"step": 840
|
| 638 |
-
},
|
| 639 |
-
{
|
| 640 |
-
"epoch": 0.005368262830148164,
|
| 641 |
-
"grad_norm": 13.618757247924805,
|
| 642 |
-
"learning_rate": 5.188148092525751e-05,
|
| 643 |
-
"loss": 1.143,
|
| 644 |
-
"step": 850
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.005431418863444025,
|
| 648 |
-
"grad_norm": 11.812853813171387,
|
| 649 |
-
"learning_rate": 5.179942845282357e-05,
|
| 650 |
-
"loss": 0.9219,
|
| 651 |
-
"step": 860
|
| 652 |
-
},
|
| 653 |
-
{
|
| 654 |
-
"epoch": 0.005494574896739886,
|
| 655 |
-
"grad_norm": 11.771102905273438,
|
| 656 |
-
"learning_rate": 5.17163771260603e-05,
|
| 657 |
-
"loss": 1.0431,
|
| 658 |
-
"step": 870
|
| 659 |
-
},
|
| 660 |
-
{
|
| 661 |
-
"epoch": 0.005557730930035747,
|
| 662 |
-
"grad_norm": 11.519074440002441,
|
| 663 |
-
"learning_rate": 5.163233035888244e-05,
|
| 664 |
-
"loss": 1.0038,
|
| 665 |
-
"step": 880
|
| 666 |
-
},
|
| 667 |
-
{
|
| 668 |
-
"epoch": 0.0056208869633316075,
|
| 669 |
-
"grad_norm": 13.609795570373535,
|
| 670 |
-
"learning_rate": 5.154729160612338e-05,
|
| 671 |
-
"loss": 0.9865,
|
| 672 |
-
"step": 890
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 0.0056840429966274674,
|
| 676 |
-
"grad_norm": 9.950550079345703,
|
| 677 |
-
"learning_rate": 5.146126436339321e-05,
|
| 678 |
-
"loss": 0.9997,
|
| 679 |
-
"step": 900
|
| 680 |
-
},
|
| 681 |
-
{
|
| 682 |
-
"epoch": 0.0056840429966274674,
|
| 683 |
-
"eval_loss": 0.8275522589683533,
|
| 684 |
-
"eval_runtime": 46.5445,
|
| 685 |
-
"eval_samples_per_second": 10.742,
|
| 686 |
-
"eval_steps_per_second": 10.742,
|
| 687 |
-
"step": 900
|
| 688 |
-
},
|
| 689 |
-
{
|
| 690 |
-
"epoch": 0.005747199029923328,
|
| 691 |
-
"grad_norm": 10.390827178955078,
|
| 692 |
-
"learning_rate": 5.137425216693491e-05,
|
| 693 |
-
"loss": 1.0279,
|
| 694 |
-
"step": 910
|
| 695 |
-
},
|
| 696 |
-
{
|
| 697 |
-
"epoch": 0.005810355063219189,
|
| 698 |
-
"grad_norm": 11.740073204040527,
|
| 699 |
-
"learning_rate": 5.128625859347907e-05,
|
| 700 |
-
"loss": 1.059,
|
| 701 |
-
"step": 920
|
| 702 |
-
},
|
| 703 |
-
{
|
| 704 |
-
"epoch": 0.00587351109651505,
|
| 705 |
-
"grad_norm": 10.699241638183594,
|
| 706 |
-
"learning_rate": 5.1197287260096865e-05,
|
| 707 |
-
"loss": 1.0401,
|
| 708 |
-
"step": 930
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 0.005936667129810911,
|
| 712 |
-
"grad_norm": 10.624009132385254,
|
| 713 |
-
"learning_rate": 5.110734182405132e-05,
|
| 714 |
-
"loss": 1.0079,
|
| 715 |
-
"step": 940
|
| 716 |
-
},
|
| 717 |
-
{
|
| 718 |
-
"epoch": 0.005999823163106772,
|
| 719 |
-
"grad_norm": 10.221404075622559,
|
| 720 |
-
"learning_rate": 5.1016425982647025e-05,
|
| 721 |
-
"loss": 0.9439,
|
| 722 |
-
"step": 950
|
| 723 |
-
},
|
| 724 |
-
{
|
| 725 |
-
"epoch": 0.0060629791964026325,
|
| 726 |
-
"grad_norm": 10.964374542236328,
|
| 727 |
-
"learning_rate": 5.092454347307812e-05,
|
| 728 |
-
"loss": 1.0385,
|
| 729 |
-
"step": 960
|
| 730 |
-
},
|
| 731 |
-
{
|
| 732 |
-
"epoch": 0.006126135229698493,
|
| 733 |
-
"grad_norm": 11.145320892333984,
|
| 734 |
-
"learning_rate": 5.08316980722747e-05,
|
| 735 |
-
"loss": 1.0424,
|
| 736 |
-
"step": 970
|
| 737 |
-
},
|
| 738 |
-
{
|
| 739 |
-
"epoch": 0.006189291262994354,
|
| 740 |
-
"grad_norm": 14.886443138122559,
|
| 741 |
-
"learning_rate": 5.0737893596747534e-05,
|
| 742 |
-
"loss": 0.9933,
|
| 743 |
-
"step": 980
|
| 744 |
-
},
|
| 745 |
-
{
|
| 746 |
-
"epoch": 0.006252447296290215,
|
| 747 |
-
"grad_norm": 11.04690933227539,
|
| 748 |
-
"learning_rate": 5.064313390243121e-05,
|
| 749 |
-
"loss": 1.1194,
|
| 750 |
-
"step": 990
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 0.006315603329586075,
|
| 754 |
-
"grad_norm": 8.777559280395508,
|
| 755 |
-
"learning_rate": 5.054742288452562e-05,
|
| 756 |
-
"loss": 1.0695,
|
| 757 |
-
"step": 1000
|
| 758 |
-
},
|
| 759 |
-
{
|
| 760 |
-
"epoch": 0.006378759362881936,
|
| 761 |
-
"grad_norm": 11.905736923217773,
|
| 762 |
-
"learning_rate": 5.0450764477335825e-05,
|
| 763 |
-
"loss": 0.9496,
|
| 764 |
-
"step": 1010
|
| 765 |
-
},
|
| 766 |
-
{
|
| 767 |
-
"epoch": 0.006441915396177797,
|
| 768 |
-
"grad_norm": 10.9399995803833,
|
| 769 |
-
"learning_rate": 5.035316265411036e-05,
|
| 770 |
-
"loss": 0.9413,
|
| 771 |
-
"step": 1020
|
| 772 |
-
},
|
| 773 |
-
{
|
| 774 |
-
"epoch": 0.0065050714294736574,
|
| 775 |
-
"grad_norm": 10.541971206665039,
|
| 776 |
-
"learning_rate": 5.02546214268779e-05,
|
| 777 |
-
"loss": 0.9718,
|
| 778 |
-
"step": 1030
|
| 779 |
-
},
|
| 780 |
-
{
|
| 781 |
-
"epoch": 0.006568227462769518,
|
| 782 |
-
"grad_norm": 11.16507625579834,
|
| 783 |
-
"learning_rate": 5.0155144846282345e-05,
|
| 784 |
-
"loss": 1.0197,
|
| 785 |
-
"step": 1040
|
| 786 |
-
},
|
| 787 |
-
{
|
| 788 |
-
"epoch": 0.006631383496065379,
|
| 789 |
-
"grad_norm": 10.459815979003906,
|
| 790 |
-
"learning_rate": 5.005473700141629e-05,
|
| 791 |
-
"loss": 1.0872,
|
| 792 |
-
"step": 1050
|
| 793 |
-
},
|
| 794 |
-
{
|
| 795 |
-
"epoch": 0.006631383496065379,
|
| 796 |
-
"eval_loss": 0.8140906691551208,
|
| 797 |
-
"eval_runtime": 47.6038,
|
| 798 |
-
"eval_samples_per_second": 10.503,
|
| 799 |
-
"eval_steps_per_second": 10.503,
|
| 800 |
-
"step": 1050
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.00669453952936124,
|
| 804 |
-
"grad_norm": 12.183165550231934,
|
| 805 |
-
"learning_rate": 4.995340201965296e-05,
|
| 806 |
-
"loss": 0.9901,
|
| 807 |
-
"step": 1060
|
| 808 |
-
},
|
| 809 |
-
{
|
| 810 |
-
"epoch": 0.006757695562657101,
|
| 811 |
-
"grad_norm": 9.4849853515625,
|
| 812 |
-
"learning_rate": 4.985114406647658e-05,
|
| 813 |
-
"loss": 0.9736,
|
| 814 |
-
"step": 1070
|
| 815 |
-
},
|
| 816 |
-
{
|
| 817 |
-
"epoch": 0.006820851595952962,
|
| 818 |
-
"grad_norm": 10.753118515014648,
|
| 819 |
-
"learning_rate": 4.9747967345311055e-05,
|
| 820 |
-
"loss": 1.0449,
|
| 821 |
-
"step": 1080
|
| 822 |
-
},
|
| 823 |
-
{
|
| 824 |
-
"epoch": 0.0068840076292488225,
|
| 825 |
-
"grad_norm": 13.308238983154297,
|
| 826 |
-
"learning_rate": 4.9643876097347296e-05,
|
| 827 |
-
"loss": 0.9135,
|
| 828 |
-
"step": 1090
|
| 829 |
-
},
|
| 830 |
-
{
|
| 831 |
-
"epoch": 0.006947163662544683,
|
| 832 |
-
"grad_norm": 7.950455665588379,
|
| 833 |
-
"learning_rate": 4.953887460136881e-05,
|
| 834 |
-
"loss": 1.007,
|
| 835 |
-
"step": 1100
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"epoch": 0.007010319695840543,
|
| 839 |
-
"grad_norm": 11.71200942993164,
|
| 840 |
-
"learning_rate": 4.943296717357583e-05,
|
| 841 |
-
"loss": 1.0232,
|
| 842 |
-
"step": 1110
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 0.007073475729136404,
|
| 846 |
-
"grad_norm": 11.027750968933105,
|
| 847 |
-
"learning_rate": 4.93261581674079e-05,
|
| 848 |
-
"loss": 1.0357,
|
| 849 |
-
"step": 1120
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 0.007136631762432265,
|
| 853 |
-
"grad_norm": 11.876934051513672,
|
| 854 |
-
"learning_rate": 4.921845197336491e-05,
|
| 855 |
-
"loss": 1.0203,
|
| 856 |
-
"step": 1130
|
| 857 |
-
},
|
| 858 |
-
{
|
| 859 |
-
"epoch": 0.007199787795728126,
|
| 860 |
-
"grad_norm": 9.920092582702637,
|
| 861 |
-
"learning_rate": 4.910985301882667e-05,
|
| 862 |
-
"loss": 1.0059,
|
| 863 |
-
"step": 1140
|
| 864 |
-
},
|
| 865 |
-
{
|
| 866 |
-
"epoch": 0.007262943829023987,
|
| 867 |
-
"grad_norm": 15.140681266784668,
|
| 868 |
-
"learning_rate": 4.9000365767870824e-05,
|
| 869 |
-
"loss": 0.9476,
|
| 870 |
-
"step": 1150
|
| 871 |
-
},
|
| 872 |
-
{
|
| 873 |
-
"epoch": 0.0073260998623198475,
|
| 874 |
-
"grad_norm": 13.791475296020508,
|
| 875 |
-
"learning_rate": 4.8889994721089426e-05,
|
| 876 |
-
"loss": 1.0171,
|
| 877 |
-
"step": 1160
|
| 878 |
-
},
|
| 879 |
-
{
|
| 880 |
-
"epoch": 0.007389255895615708,
|
| 881 |
-
"grad_norm": 11.235550880432129,
|
| 882 |
-
"learning_rate": 4.877874441540394e-05,
|
| 883 |
-
"loss": 1.0835,
|
| 884 |
-
"step": 1170
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.007452411928911569,
|
| 888 |
-
"grad_norm": 10.311321258544922,
|
| 889 |
-
"learning_rate": 4.866661942387867e-05,
|
| 890 |
-
"loss": 0.9844,
|
| 891 |
-
"step": 1180
|
| 892 |
-
},
|
| 893 |
-
{
|
| 894 |
-
"epoch": 0.00751556796220743,
|
| 895 |
-
"grad_norm": 10.934793472290039,
|
| 896 |
-
"learning_rate": 4.855362435553285e-05,
|
| 897 |
-
"loss": 0.9778,
|
| 898 |
-
"step": 1190
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 0.007578723995503291,
|
| 902 |
-
"grad_norm": 10.166644096374512,
|
| 903 |
-
"learning_rate": 4.84397638551512e-05,
|
| 904 |
-
"loss": 0.9602,
|
| 905 |
-
"step": 1200
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 0.007578723995503291,
|
| 909 |
-
"eval_loss": 0.8132573366165161,
|
| 910 |
-
"eval_runtime": 46.9842,
|
| 911 |
-
"eval_samples_per_second": 10.642,
|
| 912 |
-
"eval_steps_per_second": 10.642,
|
| 913 |
-
"step": 1200
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 0.007641880028799151,
|
| 917 |
-
"grad_norm": 8.92570686340332,
|
| 918 |
-
"learning_rate": 4.83250426030929e-05,
|
| 919 |
-
"loss": 1.0157,
|
| 920 |
-
"step": 1210
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.007705036062095012,
|
| 924 |
-
"grad_norm": 17.16946792602539,
|
| 925 |
-
"learning_rate": 4.82094653150993e-05,
|
| 926 |
-
"loss": 1.1113,
|
| 927 |
-
"step": 1220
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 0.0077681920953908724,
|
| 931 |
-
"grad_norm": 10.54002571105957,
|
| 932 |
-
"learning_rate": 4.8093036742100026e-05,
|
| 933 |
-
"loss": 1.1527,
|
| 934 |
-
"step": 1230
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.007831348128686733,
|
| 938 |
-
"grad_norm": 7.984756946563721,
|
| 939 |
-
"learning_rate": 4.79757616700177e-05,
|
| 940 |
-
"loss": 1.0058,
|
| 941 |
-
"step": 1240
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 0.007894504161982594,
|
| 945 |
-
"grad_norm": 8.092245101928711,
|
| 946 |
-
"learning_rate": 4.7857644919571176e-05,
|
| 947 |
-
"loss": 1.0381,
|
| 948 |
-
"step": 1250
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.007957660195278455,
|
| 952 |
-
"grad_norm": 9.762504577636719,
|
| 953 |
-
"learning_rate": 4.773869134607747e-05,
|
| 954 |
-
"loss": 1.0143,
|
| 955 |
-
"step": 1260
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 0.008020816228574316,
|
| 959 |
-
"grad_norm": 12.095884323120117,
|
| 960 |
-
"learning_rate": 4.761890583925204e-05,
|
| 961 |
-
"loss": 0.9924,
|
| 962 |
-
"step": 1270
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.008083972261870177,
|
| 966 |
-
"grad_norm": 8.758190155029297,
|
| 967 |
-
"learning_rate": 4.749829332300792e-05,
|
| 968 |
-
"loss": 1.0306,
|
| 969 |
-
"step": 1280
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.008147128295166037,
|
| 973 |
-
"grad_norm": 10.170825958251953,
|
| 974 |
-
"learning_rate": 4.737685875525327e-05,
|
| 975 |
-
"loss": 1.0738,
|
| 976 |
-
"step": 1290
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 0.008210284328461898,
|
| 980 |
-
"grad_norm": 12.792614936828613,
|
| 981 |
-
"learning_rate": 4.725460712768751e-05,
|
| 982 |
-
"loss": 1.0438,
|
| 983 |
-
"step": 1300
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 0.008273440361757759,
|
| 987 |
-
"grad_norm": 9.970301628112793,
|
| 988 |
-
"learning_rate": 4.7131543465596236e-05,
|
| 989 |
-
"loss": 0.9807,
|
| 990 |
-
"step": 1310
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.00833659639505362,
|
| 994 |
-
"grad_norm": 14.707938194274902,
|
| 995 |
-
"learning_rate": 4.700767282764459e-05,
|
| 996 |
-
"loss": 1.0162,
|
| 997 |
-
"step": 1320
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 0.00839975242834948,
|
| 1001 |
-
"grad_norm": 7.348803520202637,
|
| 1002 |
-
"learning_rate": 4.688300030566933e-05,
|
| 1003 |
-
"loss": 1.112,
|
| 1004 |
-
"step": 1330
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 0.008462908461645342,
|
| 1008 |
-
"grad_norm": 11.421042442321777,
|
| 1009 |
-
"learning_rate": 4.6757531024469514e-05,
|
| 1010 |
-
"loss": 1.0082,
|
| 1011 |
-
"step": 1340
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 0.008526064494941202,
|
| 1015 |
-
"grad_norm": 11.286333084106445,
|
| 1016 |
-
"learning_rate": 4.663127014159588e-05,
|
| 1017 |
-
"loss": 1.037,
|
| 1018 |
-
"step": 1350
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 0.008526064494941202,
|
| 1022 |
-
"eval_loss": 0.8081769347190857,
|
| 1023 |
-
"eval_runtime": 46.791,
|
| 1024 |
-
"eval_samples_per_second": 10.686,
|
| 1025 |
-
"eval_steps_per_second": 10.686,
|
| 1026 |
-
"step": 1350
|
| 1027 |
-
},
|
| 1028 |
-
{
|
| 1029 |
-
"epoch": 0.008589220528237063,
|
| 1030 |
-
"grad_norm": 12.930586814880371,
|
| 1031 |
-
"learning_rate": 4.650422284713878e-05,
|
| 1032 |
-
"loss": 0.9875,
|
| 1033 |
-
"step": 1360
|
| 1034 |
-
},
|
| 1035 |
-
{
|
| 1036 |
-
"epoch": 0.008652376561532922,
|
| 1037 |
-
"grad_norm": 10.3340482711792,
|
| 1038 |
-
"learning_rate": 4.637639436351489e-05,
|
| 1039 |
-
"loss": 0.9905,
|
| 1040 |
-
"step": 1370
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 0.008715532594828783,
|
| 1044 |
-
"grad_norm": 16.0806941986084,
|
| 1045 |
-
"learning_rate": 4.624778994525249e-05,
|
| 1046 |
-
"loss": 1.0658,
|
| 1047 |
-
"step": 1380
|
| 1048 |
-
},
|
| 1049 |
-
{
|
| 1050 |
-
"epoch": 0.008778688628124644,
|
| 1051 |
-
"grad_norm": 15.543879508972168,
|
| 1052 |
-
"learning_rate": 4.6118414878775514e-05,
|
| 1053 |
-
"loss": 0.9815,
|
| 1054 |
-
"step": 1390
|
| 1055 |
-
},
|
| 1056 |
-
{
|
| 1057 |
-
"epoch": 0.008841844661420505,
|
| 1058 |
-
"grad_norm": 8.364940643310547,
|
| 1059 |
-
"learning_rate": 4.5988274482186214e-05,
|
| 1060 |
-
"loss": 0.9831,
|
| 1061 |
-
"step": 1400
|
| 1062 |
-
},
|
| 1063 |
-
{
|
| 1064 |
-
"epoch": 0.008905000694716366,
|
| 1065 |
-
"grad_norm": 10.709568977355957,
|
| 1066 |
-
"learning_rate": 4.5857374105046574e-05,
|
| 1067 |
-
"loss": 0.9679,
|
| 1068 |
-
"step": 1410
|
| 1069 |
-
},
|
| 1070 |
-
{
|
| 1071 |
-
"epoch": 0.008968156728012227,
|
| 1072 |
-
"grad_norm": 12.81725788116455,
|
| 1073 |
-
"learning_rate": 4.572571912815838e-05,
|
| 1074 |
-
"loss": 1.0276,
|
| 1075 |
-
"step": 1420
|
| 1076 |
-
},
|
| 1077 |
-
{
|
| 1078 |
-
"epoch": 0.009031312761308087,
|
| 1079 |
-
"grad_norm": 10.538129806518555,
|
| 1080 |
-
"learning_rate": 4.55933149633421e-05,
|
| 1081 |
-
"loss": 1.0586,
|
| 1082 |
-
"step": 1430
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 0.009094468794603948,
|
| 1086 |
-
"grad_norm": 14.130047798156738,
|
| 1087 |
-
"learning_rate": 4.5460167053214335e-05,
|
| 1088 |
-
"loss": 1.0164,
|
| 1089 |
-
"step": 1440
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 0.009157624827899809,
|
| 1093 |
-
"grad_norm": 12.928654670715332,
|
| 1094 |
-
"learning_rate": 4.532628087096419e-05,
|
| 1095 |
-
"loss": 1.0102,
|
| 1096 |
-
"step": 1450
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 0.00922078086119567,
|
| 1100 |
-
"grad_norm": 13.670207023620605,
|
| 1101 |
-
"learning_rate": 4.5191661920128194e-05,
|
| 1102 |
-
"loss": 0.9835,
|
| 1103 |
-
"step": 1460
|
| 1104 |
-
},
|
| 1105 |
-
{
|
| 1106 |
-
"epoch": 0.00928393689449153,
|
| 1107 |
-
"grad_norm": 13.250060081481934,
|
| 1108 |
-
"learning_rate": 4.5056315734364154e-05,
|
| 1109 |
-
"loss": 0.9599,
|
| 1110 |
-
"step": 1470
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 0.009347092927787392,
|
| 1114 |
-
"grad_norm": 13.03905963897705,
|
| 1115 |
-
"learning_rate": 4.492024787722368e-05,
|
| 1116 |
-
"loss": 1.0552,
|
| 1117 |
-
"step": 1480
|
| 1118 |
-
},
|
| 1119 |
-
{
|
| 1120 |
-
"epoch": 0.009410248961083252,
|
| 1121 |
-
"grad_norm": 11.335886001586914,
|
| 1122 |
-
"learning_rate": 4.47834639419234e-05,
|
| 1123 |
-
"loss": 0.9254,
|
| 1124 |
-
"step": 1490
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 0.009473404994379113,
|
| 1128 |
-
"grad_norm": 11.690058708190918,
|
| 1129 |
-
"learning_rate": 4.464596955111518e-05,
|
| 1130 |
-
"loss": 1.0431,
|
| 1131 |
-
"step": 1500
|
| 1132 |
-
},
|
| 1133 |
-
{
|
| 1134 |
-
"epoch": 0.009473404994379113,
|
| 1135 |
-
"eval_loss": 0.8068813681602478,
|
| 1136 |
-
"eval_runtime": 46.6502,
|
| 1137 |
-
"eval_samples_per_second": 10.718,
|
| 1138 |
-
"eval_steps_per_second": 10.718,
|
| 1139 |
-
"step": 1500
|
| 1140 |
-
},
|
| 1141 |
-
{
|
| 1142 |
-
"epoch": 0.009536561027674974,
|
| 1143 |
-
"grad_norm": 9.877214431762695,
|
| 1144 |
-
"learning_rate": 4.450777035665487e-05,
|
| 1145 |
-
"loss": 0.9958,
|
| 1146 |
-
"step": 1510
|
| 1147 |
-
},
|
| 1148 |
-
{
|
| 1149 |
-
"epoch": 0.009599717060970835,
|
| 1150 |
-
"grad_norm": 8.600367546081543,
|
| 1151 |
-
"learning_rate": 4.436887203937009e-05,
|
| 1152 |
-
"loss": 1.01,
|
| 1153 |
-
"step": 1520
|
| 1154 |
-
},
|
| 1155 |
-
{
|
| 1156 |
-
"epoch": 0.009662873094266696,
|
| 1157 |
-
"grad_norm": 10.299551963806152,
|
| 1158 |
-
"learning_rate": 4.422928030882661e-05,
|
| 1159 |
-
"loss": 0.9326,
|
| 1160 |
-
"step": 1530
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 0.009726029127562557,
|
| 1164 |
-
"grad_norm": 10.05881404876709,
|
| 1165 |
-
"learning_rate": 4.4089000903093746e-05,
|
| 1166 |
-
"loss": 0.9944,
|
| 1167 |
-
"step": 1540
|
| 1168 |
-
},
|
| 1169 |
-
{
|
| 1170 |
-
"epoch": 0.009789185160858417,
|
| 1171 |
-
"grad_norm": 9.281290054321289,
|
| 1172 |
-
"learning_rate": 4.394803958850844e-05,
|
| 1173 |
-
"loss": 1.0549,
|
| 1174 |
-
"step": 1550
|
| 1175 |
-
},
|
| 1176 |
-
{
|
| 1177 |
-
"epoch": 0.009852341194154278,
|
| 1178 |
-
"grad_norm": 12.724478721618652,
|
| 1179 |
-
"learning_rate": 4.380640215943821e-05,
|
| 1180 |
-
"loss": 1.0675,
|
| 1181 |
-
"step": 1560
|
| 1182 |
-
},
|
| 1183 |
-
{
|
| 1184 |
-
"epoch": 0.009915497227450139,
|
| 1185 |
-
"grad_norm": 10.141554832458496,
|
| 1186 |
-
"learning_rate": 4.366409443804301e-05,
|
| 1187 |
-
"loss": 1.0266,
|
| 1188 |
-
"step": 1570
|
| 1189 |
-
},
|
| 1190 |
-
{
|
| 1191 |
-
"epoch": 0.009978653260745998,
|
| 1192 |
-
"grad_norm": 10.95832347869873,
|
| 1193 |
-
"learning_rate": 4.352112227403589e-05,
|
| 1194 |
-
"loss": 1.0158,
|
| 1195 |
-
"step": 1580
|
| 1196 |
-
},
|
| 1197 |
-
{
|
| 1198 |
-
"epoch": 0.010041809294041859,
|
| 1199 |
-
"grad_norm": 12.602045059204102,
|
| 1200 |
-
"learning_rate": 4.337749154444254e-05,
|
| 1201 |
-
"loss": 0.9915,
|
| 1202 |
-
"step": 1590
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 0.01010496532733772,
|
| 1206 |
-
"grad_norm": 13.776515007019043,
|
| 1207 |
-
"learning_rate": 4.3233208153359665e-05,
|
| 1208 |
-
"loss": 1.0467,
|
| 1209 |
-
"step": 1600
|
| 1210 |
-
},
|
| 1211 |
-
{
|
| 1212 |
-
"epoch": 0.01016812136063358,
|
| 1213 |
-
"grad_norm": 11.861095428466797,
|
| 1214 |
-
"learning_rate": 4.308827803171238e-05,
|
| 1215 |
-
"loss": 0.9941,
|
| 1216 |
-
"step": 1610
|
| 1217 |
-
},
|
| 1218 |
-
{
|
| 1219 |
-
"epoch": 0.010231277393929442,
|
| 1220 |
-
"grad_norm": 10.238848686218262,
|
| 1221 |
-
"learning_rate": 4.294270713701031e-05,
|
| 1222 |
-
"loss": 1.0242,
|
| 1223 |
-
"step": 1620
|
| 1224 |
-
},
|
| 1225 |
-
{
|
| 1226 |
-
"epoch": 0.010294433427225302,
|
| 1227 |
-
"grad_norm": 10.419410705566406,
|
| 1228 |
-
"learning_rate": 4.2796501453102784e-05,
|
| 1229 |
-
"loss": 0.9683,
|
| 1230 |
-
"step": 1630
|
| 1231 |
-
},
|
| 1232 |
-
{
|
| 1233 |
-
"epoch": 0.010357589460521163,
|
| 1234 |
-
"grad_norm": 10.017350196838379,
|
| 1235 |
-
"learning_rate": 4.264966698993282e-05,
|
| 1236 |
-
"loss": 1.0742,
|
| 1237 |
-
"step": 1640
|
| 1238 |
-
},
|
| 1239 |
-
{
|
| 1240 |
-
"epoch": 0.010420745493817024,
|
| 1241 |
-
"grad_norm": 9.052342414855957,
|
| 1242 |
-
"learning_rate": 4.2502209783290085e-05,
|
| 1243 |
-
"loss": 0.9827,
|
| 1244 |
-
"step": 1650
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 0.010420745493817024,
|
| 1248 |
-
"eval_loss": 0.7950078248977661,
|
| 1249 |
-
"eval_runtime": 47.5324,
|
| 1250 |
-
"eval_samples_per_second": 10.519,
|
| 1251 |
-
"eval_steps_per_second": 10.519,
|
| 1252 |
-
"step": 1650
|
| 1253 |
-
},
|
| 1254 |
-
{
|
| 1255 |
-
"epoch": 0.010483901527112885,
|
| 1256 |
-
"grad_norm": 9.733206748962402,
|
| 1257 |
-
"learning_rate": 4.235413589456281e-05,
|
| 1258 |
-
"loss": 1.0076,
|
| 1259 |
-
"step": 1660
|
| 1260 |
-
},
|
| 1261 |
-
{
|
| 1262 |
-
"epoch": 0.010547057560408746,
|
| 1263 |
-
"grad_norm": 13.072066307067871,
|
| 1264 |
-
"learning_rate": 4.2205451410488565e-05,
|
| 1265 |
-
"loss": 0.9684,
|
| 1266 |
-
"step": 1670
|
| 1267 |
-
},
|
| 1268 |
-
{
|
| 1269 |
-
"epoch": 0.010610213593704607,
|
| 1270 |
-
"grad_norm": 10.068881034851074,
|
| 1271 |
-
"learning_rate": 4.205616244290416e-05,
|
| 1272 |
-
"loss": 1.0199,
|
| 1273 |
-
"step": 1680
|
| 1274 |
-
},
|
| 1275 |
-
{
|
| 1276 |
-
"epoch": 0.010673369627000467,
|
| 1277 |
-
"grad_norm": 13.202546119689941,
|
| 1278 |
-
"learning_rate": 4.1906275128494296e-05,
|
| 1279 |
-
"loss": 1.0349,
|
| 1280 |
-
"step": 1690
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 0.010736525660296328,
|
| 1284 |
-
"grad_norm": 9.893623352050781,
|
| 1285 |
-
"learning_rate": 4.175579562853945e-05,
|
| 1286 |
-
"loss": 0.99,
|
| 1287 |
-
"step": 1700
|
| 1288 |
-
},
|
| 1289 |
-
{
|
| 1290 |
-
"epoch": 0.010799681693592189,
|
| 1291 |
-
"grad_norm": 7.337754726409912,
|
| 1292 |
-
"learning_rate": 4.160473012866242e-05,
|
| 1293 |
-
"loss": 1.0124,
|
| 1294 |
-
"step": 1710
|
| 1295 |
-
},
|
| 1296 |
-
{
|
| 1297 |
-
"epoch": 0.01086283772688805,
|
| 1298 |
-
"grad_norm": 8.579658508300781,
|
| 1299 |
-
"learning_rate": 4.145308483857426e-05,
|
| 1300 |
-
"loss": 1.0585,
|
| 1301 |
-
"step": 1720
|
| 1302 |
-
},
|
| 1303 |
-
{
|
| 1304 |
-
"epoch": 0.01092599376018391,
|
| 1305 |
-
"grad_norm": 9.737414360046387,
|
| 1306 |
-
"learning_rate": 4.1300865991818885e-05,
|
| 1307 |
-
"loss": 1.0679,
|
| 1308 |
-
"step": 1730
|
| 1309 |
-
},
|
| 1310 |
-
{
|
| 1311 |
-
"epoch": 0.010989149793479772,
|
| 1312 |
-
"grad_norm": 11.66639518737793,
|
| 1313 |
-
"learning_rate": 4.114807984551688e-05,
|
| 1314 |
-
"loss": 1.0104,
|
| 1315 |
-
"step": 1740
|
| 1316 |
-
},
|
| 1317 |
-
{
|
| 1318 |
-
"epoch": 0.011052305826775632,
|
| 1319 |
-
"grad_norm": 8.344025611877441,
|
| 1320 |
-
"learning_rate": 4.0994732680108296e-05,
|
| 1321 |
-
"loss": 1.0121,
|
| 1322 |
-
"step": 1750
|
| 1323 |
-
},
|
| 1324 |
-
{
|
| 1325 |
-
"epoch": 0.011115461860071493,
|
| 1326 |
-
"grad_norm": 9.461050987243652,
|
| 1327 |
-
"learning_rate": 4.084083079909448e-05,
|
| 1328 |
-
"loss": 0.9293,
|
| 1329 |
-
"step": 1760
|
| 1330 |
-
},
|
| 1331 |
-
{
|
| 1332 |
-
"epoch": 0.011178617893367354,
|
| 1333 |
-
"grad_norm": 11.834268569946289,
|
| 1334 |
-
"learning_rate": 4.068638052877899e-05,
|
| 1335 |
-
"loss": 1.031,
|
| 1336 |
-
"step": 1770
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 0.011241773926663215,
|
| 1340 |
-
"grad_norm": 11.639049530029297,
|
| 1341 |
-
"learning_rate": 4.0531388218007466e-05,
|
| 1342 |
-
"loss": 1.0569,
|
| 1343 |
-
"step": 1780
|
| 1344 |
-
},
|
| 1345 |
-
{
|
| 1346 |
-
"epoch": 0.011304929959959074,
|
| 1347 |
-
"grad_norm": 11.35422134399414,
|
| 1348 |
-
"learning_rate": 4.037586023790676e-05,
|
| 1349 |
-
"loss": 1.003,
|
| 1350 |
-
"step": 1790
|
| 1351 |
-
},
|
| 1352 |
-
{
|
| 1353 |
-
"epoch": 0.011368085993254935,
|
| 1354 |
-
"grad_norm": 7.621947288513184,
|
| 1355 |
-
"learning_rate": 4.0219802981622975e-05,
|
| 1356 |
-
"loss": 1.0169,
|
| 1357 |
-
"step": 1800
|
| 1358 |
-
},
|
| 1359 |
-
{
|
| 1360 |
-
"epoch": 0.011368085993254935,
|
| 1361 |
-
"eval_loss": 0.7995550632476807,
|
| 1362 |
-
"eval_runtime": 47.0267,
|
| 1363 |
-
"eval_samples_per_second": 10.632,
|
| 1364 |
-
"eval_steps_per_second": 10.632,
|
| 1365 |
-
"step": 1800
|
| 1366 |
-
},
|
| 1367 |
-
{
|
| 1368 |
-
"epoch": 0.011431242026550796,
|
| 1369 |
-
"grad_norm": 8.875388145446777,
|
| 1370 |
-
"learning_rate": 4.006322286405867e-05,
|
| 1371 |
-
"loss": 1.1443,
|
| 1372 |
-
"step": 1810
|
| 1373 |
-
},
|
| 1374 |
-
{
|
| 1375 |
-
"epoch": 0.011494398059846657,
|
| 1376 |
-
"grad_norm": 9.980118751525879,
|
| 1377 |
-
"learning_rate": 3.99061263216092e-05,
|
| 1378 |
-
"loss": 0.9522,
|
| 1379 |
-
"step": 1820
|
| 1380 |
-
},
|
| 1381 |
-
{
|
| 1382 |
-
"epoch": 0.011557554093142517,
|
| 1383 |
-
"grad_norm": 13.012611389160156,
|
| 1384 |
-
"learning_rate": 3.974851981189813e-05,
|
| 1385 |
-
"loss": 0.9845,
|
| 1386 |
-
"step": 1830
|
| 1387 |
-
},
|
| 1388 |
-
{
|
| 1389 |
-
"epoch": 0.011620710126438378,
|
| 1390 |
-
"grad_norm": 8.837371826171875,
|
| 1391 |
-
"learning_rate": 3.9590409813511765e-05,
|
| 1392 |
-
"loss": 1.0351,
|
| 1393 |
-
"step": 1840
|
| 1394 |
-
},
|
| 1395 |
-
{
|
| 1396 |
-
"epoch": 0.011683866159734239,
|
| 1397 |
-
"grad_norm": 7.165218830108643,
|
| 1398 |
-
"learning_rate": 3.943180282573285e-05,
|
| 1399 |
-
"loss": 1.0092,
|
| 1400 |
-
"step": 1850
|
| 1401 |
-
},
|
| 1402 |
-
{
|
| 1403 |
-
"epoch": 0.0117470221930301,
|
| 1404 |
-
"grad_norm": 8.621559143066406,
|
| 1405 |
-
"learning_rate": 3.927270536827346e-05,
|
| 1406 |
-
"loss": 0.9187,
|
| 1407 |
-
"step": 1860
|
| 1408 |
-
},
|
| 1409 |
-
{
|
| 1410 |
-
"epoch": 0.01181017822632596,
|
| 1411 |
-
"grad_norm": 11.009320259094238,
|
| 1412 |
-
"learning_rate": 3.91131239810069e-05,
|
| 1413 |
-
"loss": 0.9718,
|
| 1414 |
-
"step": 1870
|
| 1415 |
-
},
|
| 1416 |
-
{
|
| 1417 |
-
"epoch": 0.011873334259621822,
|
| 1418 |
-
"grad_norm": 11.635251998901367,
|
| 1419 |
-
"learning_rate": 3.895306522369898e-05,
|
| 1420 |
-
"loss": 0.99,
|
| 1421 |
-
"step": 1880
|
| 1422 |
-
},
|
| 1423 |
-
{
|
| 1424 |
-
"epoch": 0.011936490292917682,
|
| 1425 |
-
"grad_norm": 13.04053783416748,
|
| 1426 |
-
"learning_rate": 3.87925356757383e-05,
|
| 1427 |
-
"loss": 0.971,
|
| 1428 |
-
"step": 1890
|
| 1429 |
-
},
|
| 1430 |
-
{
|
| 1431 |
-
"epoch": 0.011999646326213543,
|
| 1432 |
-
"grad_norm": 9.825182914733887,
|
| 1433 |
-
"learning_rate": 3.863154193586583e-05,
|
| 1434 |
-
"loss": 0.9937,
|
| 1435 |
-
"step": 1900
|
| 1436 |
-
},
|
| 1437 |
-
{
|
| 1438 |
-
"epoch": 0.012062802359509404,
|
| 1439 |
-
"grad_norm": 10.418283462524414,
|
| 1440 |
-
"learning_rate": 3.847009062190365e-05,
|
| 1441 |
-
"loss": 0.9927,
|
| 1442 |
-
"step": 1910
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 0.012125958392805265,
|
| 1446 |
-
"grad_norm": 9.913063049316406,
|
| 1447 |
-
"learning_rate": 3.83081883704829e-05,
|
| 1448 |
-
"loss": 0.9518,
|
| 1449 |
-
"step": 1920
|
| 1450 |
-
},
|
| 1451 |
-
{
|
| 1452 |
-
"epoch": 0.012189114426101126,
|
| 1453 |
-
"grad_norm": 14.064604759216309,
|
| 1454 |
-
"learning_rate": 3.814584183677102e-05,
|
| 1455 |
-
"loss": 1.0422,
|
| 1456 |
-
"step": 1930
|
| 1457 |
-
},
|
| 1458 |
-
{
|
| 1459 |
-
"epoch": 0.012252270459396987,
|
| 1460 |
-
"grad_norm": 9.01578426361084,
|
| 1461 |
-
"learning_rate": 3.7983057694198145e-05,
|
| 1462 |
-
"loss": 0.928,
|
| 1463 |
-
"step": 1940
|
| 1464 |
-
},
|
| 1465 |
-
{
|
| 1466 |
-
"epoch": 0.012315426492692847,
|
| 1467 |
-
"grad_norm": 8.885491371154785,
|
| 1468 |
-
"learning_rate": 3.781984263418279e-05,
|
| 1469 |
-
"loss": 0.9318,
|
| 1470 |
-
"step": 1950
|
| 1471 |
-
},
|
| 1472 |
-
{
|
| 1473 |
-
"epoch": 0.012315426492692847,
|
| 1474 |
-
"eval_loss": 0.8084725737571716,
|
| 1475 |
-
"eval_runtime": 45.7715,
|
| 1476 |
-
"eval_samples_per_second": 10.924,
|
| 1477 |
-
"eval_steps_per_second": 10.924,
|
| 1478 |
-
"step": 1950
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 0.012378582525988708,
|
| 1482 |
-
"grad_norm": 13.71338939666748,
|
| 1483 |
-
"learning_rate": 3.76562033658568e-05,
|
| 1484 |
-
"loss": 0.9817,
|
| 1485 |
-
"step": 1960
|
| 1486 |
-
},
|
| 1487 |
-
{
|
| 1488 |
-
"epoch": 0.012441738559284569,
|
| 1489 |
-
"grad_norm": 8.36632251739502,
|
| 1490 |
-
"learning_rate": 3.749214661578957e-05,
|
| 1491 |
-
"loss": 0.9689,
|
| 1492 |
-
"step": 1970
|
| 1493 |
-
},
|
| 1494 |
-
{
|
| 1495 |
-
"epoch": 0.01250489459258043,
|
| 1496 |
-
"grad_norm": 9.292037010192871,
|
| 1497 |
-
"learning_rate": 3.732767912771153e-05,
|
| 1498 |
-
"loss": 0.876,
|
| 1499 |
-
"step": 1980
|
| 1500 |
-
},
|
| 1501 |
-
{
|
| 1502 |
-
"epoch": 0.01256805062587629,
|
| 1503 |
-
"grad_norm": 10.010106086730957,
|
| 1504 |
-
"learning_rate": 3.716280766223693e-05,
|
| 1505 |
-
"loss": 0.9178,
|
| 1506 |
-
"step": 1990
|
| 1507 |
-
},
|
| 1508 |
-
{
|
| 1509 |
-
"epoch": 0.01263120665917215,
|
| 1510 |
-
"grad_norm": 12.617682456970215,
|
| 1511 |
-
"learning_rate": 3.699753899658596e-05,
|
| 1512 |
-
"loss": 0.9502,
|
| 1513 |
-
"step": 2000
|
| 1514 |
-
},
|
| 1515 |
-
{
|
| 1516 |
-
"epoch": 0.01269436269246801,
|
| 1517 |
-
"grad_norm": 9.597683906555176,
|
| 1518 |
-
"learning_rate": 3.683187992430616e-05,
|
| 1519 |
-
"loss": 0.9964,
|
| 1520 |
-
"step": 2010
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 0.012757518725763872,
|
| 1524 |
-
"grad_norm": 11.982294082641602,
|
| 1525 |
-
"learning_rate": 3.666583725499315e-05,
|
| 1526 |
-
"loss": 0.8793,
|
| 1527 |
-
"step": 2020
|
| 1528 |
-
},
|
| 1529 |
-
{
|
| 1530 |
-
"epoch": 0.012820674759059732,
|
| 1531 |
-
"grad_norm": 12.932259559631348,
|
| 1532 |
-
"learning_rate": 3.6499417814010715e-05,
|
| 1533 |
-
"loss": 1.0322,
|
| 1534 |
-
"step": 2030
|
| 1535 |
-
},
|
| 1536 |
-
{
|
| 1537 |
-
"epoch": 0.012883830792355593,
|
| 1538 |
-
"grad_norm": 11.51771068572998,
|
| 1539 |
-
"learning_rate": 3.6332628442210255e-05,
|
| 1540 |
-
"loss": 0.9822,
|
| 1541 |
-
"step": 2040
|
| 1542 |
-
},
|
| 1543 |
-
{
|
| 1544 |
-
"epoch": 0.012946986825651454,
|
| 1545 |
-
"grad_norm": 11.403252601623535,
|
| 1546 |
-
"learning_rate": 3.616547599564958e-05,
|
| 1547 |
-
"loss": 1.0164,
|
| 1548 |
-
"step": 2050
|
| 1549 |
-
},
|
| 1550 |
-
{
|
| 1551 |
-
"epoch": 0.013010142858947315,
|
| 1552 |
-
"grad_norm": 9.606186866760254,
|
| 1553 |
-
"learning_rate": 3.599796734531105e-05,
|
| 1554 |
-
"loss": 1.0764,
|
| 1555 |
-
"step": 2060
|
| 1556 |
-
},
|
| 1557 |
-
{
|
| 1558 |
-
"epoch": 0.013073298892243176,
|
| 1559 |
-
"grad_norm": 10.46757984161377,
|
| 1560 |
-
"learning_rate": 3.5830109376819235e-05,
|
| 1561 |
-
"loss": 1.0254,
|
| 1562 |
-
"step": 2070
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 0.013136454925539037,
|
| 1566 |
-
"grad_norm": 12.479496955871582,
|
| 1567 |
-
"learning_rate": 3.566190899015774e-05,
|
| 1568 |
-
"loss": 1.0147,
|
| 1569 |
-
"step": 2080
|
| 1570 |
-
},
|
| 1571 |
-
{
|
| 1572 |
-
"epoch": 0.013199610958834897,
|
| 1573 |
-
"grad_norm": 11.903168678283691,
|
| 1574 |
-
"learning_rate": 3.5493373099385677e-05,
|
| 1575 |
-
"loss": 1.0082,
|
| 1576 |
-
"step": 2090
|
| 1577 |
-
},
|
| 1578 |
-
{
|
| 1579 |
-
"epoch": 0.013262766992130758,
|
| 1580 |
-
"grad_norm": 9.391939163208008,
|
| 1581 |
-
"learning_rate": 3.5324508632353394e-05,
|
| 1582 |
-
"loss": 1.0489,
|
| 1583 |
-
"step": 2100
|
| 1584 |
-
},
|
| 1585 |
-
{
|
| 1586 |
-
"epoch": 0.013262766992130758,
|
| 1587 |
-
"eval_loss": 0.7725129127502441,
|
| 1588 |
-
"eval_runtime": 45.2632,
|
| 1589 |
-
"eval_samples_per_second": 11.047,
|
| 1590 |
-
"eval_steps_per_second": 11.047,
|
| 1591 |
-
"step": 2100
|
| 1592 |
-
},
|
| 1593 |
-
{
|
| 1594 |
-
"epoch": 0.013325923025426619,
|
| 1595 |
-
"grad_norm": 10.084378242492676,
|
| 1596 |
-
"learning_rate": 3.515532253041774e-05,
|
| 1597 |
-
"loss": 1.0542,
|
| 1598 |
-
"step": 2110
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 0.01338907905872248,
|
| 1602 |
-
"grad_norm": 11.785043716430664,
|
| 1603 |
-
"learning_rate": 3.498582174815671e-05,
|
| 1604 |
-
"loss": 1.0488,
|
| 1605 |
-
"step": 2120
|
| 1606 |
-
},
|
| 1607 |
-
{
|
| 1608 |
-
"epoch": 0.01345223509201834,
|
| 1609 |
-
"grad_norm": 10.521591186523438,
|
| 1610 |
-
"learning_rate": 3.481601325308357e-05,
|
| 1611 |
-
"loss": 0.9893,
|
| 1612 |
-
"step": 2130
|
| 1613 |
-
},
|
| 1614 |
-
{
|
| 1615 |
-
"epoch": 0.013515391125314202,
|
| 1616 |
-
"grad_norm": 8.644886016845703,
|
| 1617 |
-
"learning_rate": 3.4645904025360455e-05,
|
| 1618 |
-
"loss": 0.9614,
|
| 1619 |
-
"step": 2140
|
| 1620 |
-
},
|
| 1621 |
-
{
|
| 1622 |
-
"epoch": 0.013578547158610062,
|
| 1623 |
-
"grad_norm": 9.69764232635498,
|
| 1624 |
-
"learning_rate": 3.447550105751145e-05,
|
| 1625 |
-
"loss": 0.9104,
|
| 1626 |
-
"step": 2150
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"epoch": 0.013641703191905923,
|
| 1630 |
-
"grad_norm": 14.97421646118164,
|
| 1631 |
-
"learning_rate": 3.4304811354135145e-05,
|
| 1632 |
-
"loss": 0.9324,
|
| 1633 |
-
"step": 2160
|
| 1634 |
-
},
|
| 1635 |
-
{
|
| 1636 |
-
"epoch": 0.013704859225201784,
|
| 1637 |
-
"grad_norm": 9.63864803314209,
|
| 1638 |
-
"learning_rate": 3.4133841931616696e-05,
|
| 1639 |
-
"loss": 1.0068,
|
| 1640 |
-
"step": 2170
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 0.013768015258497645,
|
| 1644 |
-
"grad_norm": 9.305994033813477,
|
| 1645 |
-
"learning_rate": 3.396259981783942e-05,
|
| 1646 |
-
"loss": 0.9625,
|
| 1647 |
-
"step": 2180
|
| 1648 |
-
},
|
| 1649 |
-
{
|
| 1650 |
-
"epoch": 0.013831171291793506,
|
| 1651 |
-
"grad_norm": 8.71728515625,
|
| 1652 |
-
"learning_rate": 3.37910920518959e-05,
|
| 1653 |
-
"loss": 1.0038,
|
| 1654 |
-
"step": 2190
|
| 1655 |
-
},
|
| 1656 |
-
{
|
| 1657 |
-
"epoch": 0.013894327325089367,
|
| 1658 |
-
"grad_norm": 6.916072845458984,
|
| 1659 |
-
"learning_rate": 3.3619325683798646e-05,
|
| 1660 |
-
"loss": 1.017,
|
| 1661 |
-
"step": 2200
|
| 1662 |
-
},
|
| 1663 |
-
{
|
| 1664 |
-
"epoch": 0.013957483358385226,
|
| 1665 |
-
"grad_norm": 10.644246101379395,
|
| 1666 |
-
"learning_rate": 3.3447307774190296e-05,
|
| 1667 |
-
"loss": 0.883,
|
| 1668 |
-
"step": 2210
|
| 1669 |
-
},
|
| 1670 |
-
{
|
| 1671 |
-
"epoch": 0.014020639391681087,
|
| 1672 |
-
"grad_norm": 12.687426567077637,
|
| 1673 |
-
"learning_rate": 3.327504539405335e-05,
|
| 1674 |
-
"loss": 0.9776,
|
| 1675 |
-
"step": 2220
|
| 1676 |
-
},
|
| 1677 |
-
{
|
| 1678 |
-
"epoch": 0.014083795424976947,
|
| 1679 |
-
"grad_norm": 13.301749229431152,
|
| 1680 |
-
"learning_rate": 3.3102545624419583e-05,
|
| 1681 |
-
"loss": 0.9652,
|
| 1682 |
-
"step": 2230
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 0.014146951458272808,
|
| 1686 |
-
"grad_norm": 11.929834365844727,
|
| 1687 |
-
"learning_rate": 3.292981555607884e-05,
|
| 1688 |
-
"loss": 0.9486,
|
| 1689 |
-
"step": 2240
|
| 1690 |
-
},
|
| 1691 |
-
{
|
| 1692 |
-
"epoch": 0.014210107491568669,
|
| 1693 |
-
"grad_norm": 11.27684211730957,
|
| 1694 |
-
"learning_rate": 3.2756862289287746e-05,
|
| 1695 |
-
"loss": 0.971,
|
| 1696 |
-
"step": 2250
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 0.014210107491568669,
|
| 1700 |
-
"eval_loss": 0.7654532194137573,
|
| 1701 |
-
"eval_runtime": 45.5502,
|
| 1702 |
-
"eval_samples_per_second": 10.977,
|
| 1703 |
-
"eval_steps_per_second": 10.977,
|
| 1704 |
-
"step": 2250
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 0.01427326352486453,
|
| 1708 |
-
"grad_norm": 7.769909381866455,
|
| 1709 |
-
"learning_rate": 3.258369293347764e-05,
|
| 1710 |
-
"loss": 0.9666,
|
| 1711 |
-
"step": 2260
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 0.01433641955816039,
|
| 1715 |
-
"grad_norm": 8.687183380126953,
|
| 1716 |
-
"learning_rate": 3.241031460696251e-05,
|
| 1717 |
-
"loss": 0.995,
|
| 1718 |
-
"step": 2270
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 0.014399575591456252,
|
| 1722 |
-
"grad_norm": 9.484212875366211,
|
| 1723 |
-
"learning_rate": 3.223673443664627e-05,
|
| 1724 |
-
"loss": 0.9212,
|
| 1725 |
-
"step": 2280
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 0.014462731624752112,
|
| 1729 |
-
"grad_norm": 8.085176467895508,
|
| 1730 |
-
"learning_rate": 3.206295955772987e-05,
|
| 1731 |
-
"loss": 0.9898,
|
| 1732 |
-
"step": 2290
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 0.014525887658047973,
|
| 1736 |
-
"grad_norm": 7.797507286071777,
|
| 1737 |
-
"learning_rate": 3.188899711341793e-05,
|
| 1738 |
-
"loss": 1.0636,
|
| 1739 |
-
"step": 2300
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 0.014589043691343834,
|
| 1743 |
-
"grad_norm": 9.53068733215332,
|
| 1744 |
-
"learning_rate": 3.171485425462518e-05,
|
| 1745 |
-
"loss": 0.9875,
|
| 1746 |
-
"step": 2310
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 0.014652199724639695,
|
| 1750 |
-
"grad_norm": 8.484146118164062,
|
| 1751 |
-
"learning_rate": 3.15405381396825e-05,
|
| 1752 |
-
"loss": 0.9965,
|
| 1753 |
-
"step": 2320
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 0.014715355757935556,
|
| 1757 |
-
"grad_norm": 9.026712417602539,
|
| 1758 |
-
"learning_rate": 3.136605593404258e-05,
|
| 1759 |
-
"loss": 1.0284,
|
| 1760 |
-
"step": 2330
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 0.014778511791231417,
|
| 1764 |
-
"grad_norm": 8.642735481262207,
|
| 1765 |
-
"learning_rate": 3.119141480998553e-05,
|
| 1766 |
-
"loss": 0.9835,
|
| 1767 |
-
"step": 2340
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 0.014841667824527277,
|
| 1771 |
-
"grad_norm": 9.15679931640625,
|
| 1772 |
-
"learning_rate": 3.101662194632392e-05,
|
| 1773 |
-
"loss": 1.065,
|
| 1774 |
-
"step": 2350
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 0.014904823857823138,
|
| 1778 |
-
"grad_norm": 12.24231243133545,
|
| 1779 |
-
"learning_rate": 3.0841684528107766e-05,
|
| 1780 |
-
"loss": 0.9253,
|
| 1781 |
-
"step": 2360
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 0.014967979891118999,
|
| 1785 |
-
"grad_norm": 10.09658432006836,
|
| 1786 |
-
"learning_rate": 3.066660974632914e-05,
|
| 1787 |
-
"loss": 1.0556,
|
| 1788 |
-
"step": 2370
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 0.01503113592441486,
|
| 1792 |
-
"grad_norm": 9.404873847961426,
|
| 1793 |
-
"learning_rate": 3.0491404797626605e-05,
|
| 1794 |
-
"loss": 0.9384,
|
| 1795 |
-
"step": 2380
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 0.01509429195771072,
|
| 1799 |
-
"grad_norm": 10.99118423461914,
|
| 1800 |
-
"learning_rate": 3.031607688398936e-05,
|
| 1801 |
-
"loss": 0.8505,
|
| 1802 |
-
"step": 2390
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 0.015157447991006582,
|
| 1806 |
-
"grad_norm": 8.370269775390625,
|
| 1807 |
-
"learning_rate": 3.0140633212461248e-05,
|
| 1808 |
-
"loss": 1.0307,
|
| 1809 |
-
"step": 2400
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 0.015157447991006582,
|
| 1813 |
-
"eval_loss": 0.7730175852775574,
|
| 1814 |
-
"eval_runtime": 45.2705,
|
| 1815 |
-
"eval_samples_per_second": 11.045,
|
| 1816 |
-
"eval_steps_per_second": 11.045,
|
| 1817 |
-
"step": 2400
|
| 1818 |
-
},
|
| 1819 |
-
{
|
| 1820 |
-
"epoch": 0.015220604024302442,
|
| 1821 |
-
"grad_norm": 10.080233573913574,
|
| 1822 |
-
"learning_rate": 2.9965080994844422e-05,
|
| 1823 |
-
"loss": 0.9019,
|
| 1824 |
-
"step": 2410
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 0.015283760057598302,
|
| 1828 |
-
"grad_norm": 8.396200180053711,
|
| 1829 |
-
"learning_rate": 2.978942744740296e-05,
|
| 1830 |
-
"loss": 0.9471,
|
| 1831 |
-
"step": 2420
|
| 1832 |
-
},
|
| 1833 |
-
{
|
| 1834 |
-
"epoch": 0.015346916090894162,
|
| 1835 |
-
"grad_norm": 8.11946964263916,
|
| 1836 |
-
"learning_rate": 2.961367979056621e-05,
|
| 1837 |
-
"loss": 0.9168,
|
| 1838 |
-
"step": 2430
|
| 1839 |
-
},
|
| 1840 |
-
{
|
| 1841 |
-
"epoch": 0.015410072124190023,
|
| 1842 |
-
"grad_norm": 9.612983703613281,
|
| 1843 |
-
"learning_rate": 2.9437845248631984e-05,
|
| 1844 |
-
"loss": 0.8768,
|
| 1845 |
-
"step": 2440
|
| 1846 |
-
},
|
| 1847 |
-
{
|
| 1848 |
-
"epoch": 0.015473228157485884,
|
| 1849 |
-
"grad_norm": 8.1029634475708,
|
| 1850 |
-
"learning_rate": 2.926193104946961e-05,
|
| 1851 |
-
"loss": 0.9638,
|
| 1852 |
-
"step": 2450
|
| 1853 |
-
},
|
| 1854 |
-
{
|
| 1855 |
-
"epoch": 0.015536384190781745,
|
| 1856 |
-
"grad_norm": 7.916716575622559,
|
| 1857 |
-
"learning_rate": 2.90859444242228e-05,
|
| 1858 |
-
"loss": 1.0279,
|
| 1859 |
-
"step": 2460
|
| 1860 |
-
},
|
| 1861 |
-
{
|
| 1862 |
-
"epoch": 0.015599540224077606,
|
| 1863 |
-
"grad_norm": 9.679882049560547,
|
| 1864 |
-
"learning_rate": 2.8909892607012427e-05,
|
| 1865 |
-
"loss": 0.9395,
|
| 1866 |
-
"step": 2470
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 0.015662696257373467,
|
| 1870 |
-
"grad_norm": 10.735697746276855,
|
| 1871 |
-
"learning_rate": 2.8733782834639165e-05,
|
| 1872 |
-
"loss": 0.9673,
|
| 1873 |
-
"step": 2480
|
| 1874 |
-
},
|
| 1875 |
-
{
|
| 1876 |
-
"epoch": 0.015725852290669327,
|
| 1877 |
-
"grad_norm": 9.609941482543945,
|
| 1878 |
-
"learning_rate": 2.8557622346285957e-05,
|
| 1879 |
-
"loss": 0.9712,
|
| 1880 |
-
"step": 2490
|
| 1881 |
-
},
|
| 1882 |
-
{
|
| 1883 |
-
"epoch": 0.015789008323965188,
|
| 1884 |
-
"grad_norm": 10.74219799041748,
|
| 1885 |
-
"learning_rate": 2.8381418383220526e-05,
|
| 1886 |
-
"loss": 0.9714,
|
| 1887 |
-
"step": 2500
|
| 1888 |
-
},
|
| 1889 |
-
{
|
| 1890 |
-
"epoch": 0.01585216435726105,
|
| 1891 |
-
"grad_norm": 8.981695175170898,
|
| 1892 |
-
"learning_rate": 2.8205178188497627e-05,
|
| 1893 |
-
"loss": 0.9977,
|
| 1894 |
-
"step": 2510
|
| 1895 |
-
},
|
| 1896 |
-
{
|
| 1897 |
-
"epoch": 0.01591532039055691,
|
| 1898 |
-
"grad_norm": 9.596540451049805,
|
| 1899 |
-
"learning_rate": 2.8028909006661396e-05,
|
| 1900 |
-
"loss": 0.946,
|
| 1901 |
-
"step": 2520
|
| 1902 |
-
},
|
| 1903 |
-
{
|
| 1904 |
-
"epoch": 0.01597847642385277,
|
| 1905 |
-
"grad_norm": 7.082404613494873,
|
| 1906 |
-
"learning_rate": 2.78526180834475e-05,
|
| 1907 |
-
"loss": 0.9852,
|
| 1908 |
-
"step": 2530
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 0.01604163245714863,
|
| 1912 |
-
"grad_norm": 8.827990531921387,
|
| 1913 |
-
"learning_rate": 2.7676312665485307e-05,
|
| 1914 |
-
"loss": 1.0499,
|
| 1915 |
-
"step": 2540
|
| 1916 |
-
},
|
| 1917 |
-
{
|
| 1918 |
-
"epoch": 0.016104788490444492,
|
| 1919 |
-
"grad_norm": 10.917720794677734,
|
| 1920 |
-
"learning_rate": 2.75e-05,
|
| 1921 |
-
"loss": 0.9835,
|
| 1922 |
-
"step": 2550
|
| 1923 |
-
},
|
| 1924 |
-
{
|
| 1925 |
-
"epoch": 0.016104788490444492,
|
| 1926 |
-
"eval_loss": 0.7639342546463013,
|
| 1927 |
-
"eval_runtime": 45.562,
|
| 1928 |
-
"eval_samples_per_second": 10.974,
|
| 1929 |
-
"eval_steps_per_second": 10.974,
|
| 1930 |
-
"step": 2550
|
| 1931 |
-
},
|
| 1932 |
-
{
|
| 1933 |
-
"epoch": 0.016167944523740353,
|
| 1934 |
-
"grad_norm": 11.584785461425781,
|
| 1935 |
-
"learning_rate": 2.7323687334514695e-05,
|
| 1936 |
-
"loss": 0.8422,
|
| 1937 |
-
"step": 2560
|
| 1938 |
-
},
|
| 1939 |
-
{
|
| 1940 |
-
"epoch": 0.016231100557036214,
|
| 1941 |
-
"grad_norm": 9.246831893920898,
|
| 1942 |
-
"learning_rate": 2.71473819165525e-05,
|
| 1943 |
-
"loss": 1.0227,
|
| 1944 |
-
"step": 2570
|
| 1945 |
-
},
|
| 1946 |
-
{
|
| 1947 |
-
"epoch": 0.016294256590332075,
|
| 1948 |
-
"grad_norm": 9.19963264465332,
|
| 1949 |
-
"learning_rate": 2.6971090993338606e-05,
|
| 1950 |
-
"loss": 0.9826,
|
| 1951 |
-
"step": 2580
|
| 1952 |
-
},
|
| 1953 |
-
{
|
| 1954 |
-
"epoch": 0.016357412623627936,
|
| 1955 |
-
"grad_norm": 7.812788009643555,
|
| 1956 |
-
"learning_rate": 2.679482181150238e-05,
|
| 1957 |
-
"loss": 0.8062,
|
| 1958 |
-
"step": 2590
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 0.016420568656923797,
|
| 1962 |
-
"grad_norm": 6.75607967376709,
|
| 1963 |
-
"learning_rate": 2.6618581616779483e-05,
|
| 1964 |
-
"loss": 0.8495,
|
| 1965 |
-
"step": 2600
|
| 1966 |
-
},
|
| 1967 |
-
{
|
| 1968 |
-
"epoch": 0.016483724690219657,
|
| 1969 |
-
"grad_norm": 8.447277069091797,
|
| 1970 |
-
"learning_rate": 2.644237765371404e-05,
|
| 1971 |
-
"loss": 1.1002,
|
| 1972 |
-
"step": 2610
|
| 1973 |
-
},
|
| 1974 |
-
{
|
| 1975 |
-
"epoch": 0.016546880723515518,
|
| 1976 |
-
"grad_norm": 9.761106491088867,
|
| 1977 |
-
"learning_rate": 2.626621716536085e-05,
|
| 1978 |
-
"loss": 0.9549,
|
| 1979 |
-
"step": 2620
|
| 1980 |
-
},
|
| 1981 |
-
{
|
| 1982 |
-
"epoch": 0.01661003675681138,
|
| 1983 |
-
"grad_norm": 10.971216201782227,
|
| 1984 |
-
"learning_rate": 2.6090107392987575e-05,
|
| 1985 |
-
"loss": 0.9771,
|
| 1986 |
-
"step": 2630
|
| 1987 |
-
},
|
| 1988 |
-
{
|
| 1989 |
-
"epoch": 0.01667319279010724,
|
| 1990 |
-
"grad_norm": 11.389016151428223,
|
| 1991 |
-
"learning_rate": 2.591405557577721e-05,
|
| 1992 |
-
"loss": 0.9737,
|
| 1993 |
-
"step": 2640
|
| 1994 |
-
},
|
| 1995 |
-
{
|
| 1996 |
-
"epoch": 0.0167363488234031,
|
| 1997 |
-
"grad_norm": 9.189516067504883,
|
| 1998 |
-
"learning_rate": 2.5738068950530398e-05,
|
| 1999 |
-
"loss": 0.9855,
|
| 2000 |
-
"step": 2650
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 0.01679950485669896,
|
| 2004 |
-
"grad_norm": 8.623804092407227,
|
| 2005 |
-
"learning_rate": 2.5562154751368014e-05,
|
| 2006 |
-
"loss": 1.0468,
|
| 2007 |
-
"step": 2660
|
| 2008 |
-
},
|
| 2009 |
-
{
|
| 2010 |
-
"epoch": 0.016862660889994822,
|
| 2011 |
-
"grad_norm": 10.484329223632812,
|
| 2012 |
-
"learning_rate": 2.5386320209433798e-05,
|
| 2013 |
-
"loss": 1.0479,
|
| 2014 |
-
"step": 2670
|
| 2015 |
-
},
|
| 2016 |
-
{
|
| 2017 |
-
"epoch": 0.016925816923290683,
|
| 2018 |
-
"grad_norm": 9.45596981048584,
|
| 2019 |
-
"learning_rate": 2.5210572552597046e-05,
|
| 2020 |
-
"loss": 0.9985,
|
| 2021 |
-
"step": 2680
|
| 2022 |
-
},
|
| 2023 |
-
{
|
| 2024 |
-
"epoch": 0.016988972956586544,
|
| 2025 |
-
"grad_norm": 11.783865928649902,
|
| 2026 |
-
"learning_rate": 2.5034919005155583e-05,
|
| 2027 |
-
"loss": 1.0066,
|
| 2028 |
-
"step": 2690
|
| 2029 |
-
},
|
| 2030 |
-
{
|
| 2031 |
-
"epoch": 0.017052128989882405,
|
| 2032 |
-
"grad_norm": 11.132217407226562,
|
| 2033 |
-
"learning_rate": 2.4859366787538754e-05,
|
| 2034 |
-
"loss": 0.8558,
|
| 2035 |
-
"step": 2700
|
| 2036 |
-
},
|
| 2037 |
-
{
|
| 2038 |
-
"epoch": 0.017052128989882405,
|
| 2039 |
-
"eval_loss": 0.7622952461242676,
|
| 2040 |
-
"eval_runtime": 45.0496,
|
| 2041 |
-
"eval_samples_per_second": 11.099,
|
| 2042 |
-
"eval_steps_per_second": 11.099,
|
| 2043 |
-
"step": 2700
|
| 2044 |
}
|
| 2045 |
],
|
| 2046 |
"logging_steps": 10,
|
|
@@ -2060,8 +139,8 @@
|
|
| 2060 |
"attributes": {}
|
| 2061 |
}
|
| 2062 |
},
|
| 2063 |
-
"total_flos":
|
| 2064 |
-
"train_batch_size":
|
| 2065 |
"trial_name": null,
|
| 2066 |
"trial_params": null
|
| 2067 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8098691701889038,
|
| 3 |
+
"best_model_checkpoint": "./output/checkpoint-150",
|
| 4 |
+
"epoch": 0.0004736702497189557,
|
| 5 |
"eval_steps": 150,
|
| 6 |
+
"global_step": 150,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 3.157801664793038e-05,
|
| 13 |
+
"grad_norm": 9.345338821411133,
|
| 14 |
"learning_rate": 5.500000000000001e-06,
|
| 15 |
+
"loss": 0.9144,
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 6.315603329586075e-05,
|
| 20 |
+
"grad_norm": 15.730955123901367,
|
| 21 |
"learning_rate": 1.1000000000000001e-05,
|
| 22 |
+
"loss": 0.8917,
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 9.473404994379113e-05,
|
| 27 |
+
"grad_norm": 11.728280067443848,
|
| 28 |
"learning_rate": 1.65e-05,
|
| 29 |
+
"loss": 0.9514,
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.0001263120665917215,
|
| 34 |
+
"grad_norm": 10.77907943725586,
|
| 35 |
"learning_rate": 2.2000000000000003e-05,
|
| 36 |
+
"loss": 0.9063,
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.00015789008323965187,
|
| 41 |
+
"grad_norm": 11.698442459106445,
|
| 42 |
"learning_rate": 2.75e-05,
|
| 43 |
+
"loss": 0.8968,
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.00018946809988758226,
|
| 48 |
+
"grad_norm": 14.10632038116455,
|
| 49 |
"learning_rate": 3.3e-05,
|
| 50 |
+
"loss": 0.8452,
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.00022104611653551263,
|
| 55 |
+
"grad_norm": 16.12811851501465,
|
| 56 |
"learning_rate": 3.85e-05,
|
| 57 |
+
"loss": 0.8933,
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.000252624133183443,
|
| 62 |
+
"grad_norm": 15.834749221801758,
|
| 63 |
"learning_rate": 4.4000000000000006e-05,
|
| 64 |
+
"loss": 0.9842,
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.0002842021498313734,
|
| 69 |
+
"grad_norm": 9.127525329589844,
|
| 70 |
"learning_rate": 4.9500000000000004e-05,
|
| 71 |
+
"loss": 0.8358,
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.00031578016647930375,
|
| 76 |
+
"grad_norm": 16.81833839416504,
|
| 77 |
"learning_rate": 5.5e-05,
|
| 78 |
+
"loss": 0.8891,
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.00034735818312723417,
|
| 83 |
+
"grad_norm": 15.804556846618652,
|
| 84 |
"learning_rate": 5.4999434791355066e-05,
|
| 85 |
+
"loss": 0.9371,
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.00037893619977516453,
|
| 90 |
+
"grad_norm": 18.529338836669922,
|
| 91 |
"learning_rate": 5.4997739188653784e-05,
|
| 92 |
+
"loss": 1.0804,
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.0004105142164230949,
|
| 97 |
+
"grad_norm": 13.449525833129883,
|
| 98 |
"learning_rate": 5.4994913261595724e-05,
|
| 99 |
+
"loss": 0.9468,
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.00044209223307102526,
|
| 104 |
+
"grad_norm": 20.342548370361328,
|
| 105 |
"learning_rate": 5.49909571263437e-05,
|
| 106 |
+
"loss": 0.9881,
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.0004736702497189557,
|
| 111 |
+
"grad_norm": 11.419866561889648,
|
| 112 |
"learning_rate": 5.498587094551892e-05,
|
| 113 |
+
"loss": 1.0642,
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.0004736702497189557,
|
| 118 |
+
"eval_loss": 0.8098691701889038,
|
| 119 |
+
"eval_runtime": 45.4214,
|
| 120 |
+
"eval_samples_per_second": 11.008,
|
| 121 |
+
"eval_steps_per_second": 11.008,
|
| 122 |
"step": 150
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 123 |
}
|
| 124 |
],
|
| 125 |
"logging_steps": 10,
|
|
|
|
| 139 |
"attributes": {}
|
| 140 |
}
|
| 141 |
},
|
| 142 |
+
"total_flos": 4816346273759232.0,
|
| 143 |
+
"train_batch_size": 8,
|
| 144 |
"trial_name": null,
|
| 145 |
"trial_params": null
|
| 146 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5496
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f2876111fb354dec6b4e9d27b5fc744ee66c6a7ed1060873d2769d459ef601a
|
| 3 |
size 5496
|