Add 16 files
Browse files- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +819 -103
- training_args.bin +1 -1
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4967215360
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b86f5f45d6144d0fa35c2b8d730ce6f3cf67244843f31bae3f5f69924edf468c
|
| 3 |
size 4967215360
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3077766632
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88459e7f33fdabff763552eb4e14c7418f497a8132c7b14597e1c795f95ca8d2
|
| 3 |
size 3077766632
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:718a0f3db00824213036a2c0441849791319b7d9cf189065873bb26a7020738e
|
| 3 |
size 14645
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:67cd9271a188fb648597e505255f5e1ea37f0260a2bbe5ba6a99f0ce616fc6b3
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -1,371 +1,1087 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 0.
|
| 4 |
-
"best_model_checkpoint": "./
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.007226738934056007,
|
| 14 |
-
"grad_norm": 1.
|
| 15 |
"learning_rate": 1.44e-06,
|
| 16 |
-
"loss": 0.
|
| 17 |
"step": 10
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.014453477868112014,
|
| 21 |
-
"grad_norm": 1.
|
| 22 |
"learning_rate": 3.04e-06,
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 20
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.02168021680216802,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
"learning_rate": 4.6400000000000005e-06,
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 30
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.028906955736224028,
|
| 35 |
-
"grad_norm": 0.
|
| 36 |
"learning_rate": 6.24e-06,
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 40
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.036133694670280034,
|
| 42 |
-
"grad_norm": 0.
|
| 43 |
"learning_rate": 7.840000000000001e-06,
|
| 44 |
-
"loss": 0.
|
| 45 |
"step": 50
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04336043360433604,
|
| 49 |
-
"grad_norm": 0.
|
| 50 |
"learning_rate": 9.440000000000001e-06,
|
| 51 |
-
"loss": 0.
|
| 52 |
"step": 60
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05058717253839205,
|
| 56 |
-
"grad_norm": 0.
|
| 57 |
"learning_rate": 1.1040000000000001e-05,
|
| 58 |
-
"loss": 0.
|
| 59 |
"step": 70
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.057813911472448055,
|
| 63 |
-
"grad_norm": 0.
|
| 64 |
"learning_rate": 1.2640000000000001e-05,
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 80
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.06504065040650407,
|
| 70 |
-
"grad_norm": 0.
|
| 71 |
"learning_rate": 1.4240000000000001e-05,
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 90
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.07226738934056007,
|
| 77 |
-
"grad_norm": 0.
|
| 78 |
"learning_rate": 1.584e-05,
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 100
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.07949412827461608,
|
| 84 |
-
"grad_norm": 0.
|
| 85 |
"learning_rate": 1.7440000000000002e-05,
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 110
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.08672086720867209,
|
| 91 |
-
"grad_norm": 0.
|
| 92 |
"learning_rate": 1.904e-05,
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 120
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.0939476061427281,
|
| 98 |
-
"grad_norm": 0.
|
| 99 |
"learning_rate": 1.99801340948597e-05,
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 130
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.1011743450767841,
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
"learning_rate": 1.9930469332008943e-05,
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 140
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.10840108401084012,
|
| 112 |
-
"grad_norm": 0.
|
| 113 |
"learning_rate": 1.9880804569158184e-05,
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 150
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.11562782294489611,
|
| 119 |
-
"grad_norm": 0.
|
| 120 |
"learning_rate": 1.9831139806307428e-05,
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 160
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.12285456187895212,
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
"learning_rate": 1.9781475043456668e-05,
|
| 128 |
"loss": 0.0217,
|
| 129 |
"step": 170
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.13008130081300814,
|
| 133 |
-
"grad_norm": 0.
|
| 134 |
"learning_rate": 1.9731810280605912e-05,
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 180
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.13730803974706413,
|
| 140 |
-
"grad_norm": 0.
|
| 141 |
"learning_rate": 1.9682145517755153e-05,
|
| 142 |
"loss": 0.029,
|
| 143 |
"step": 190
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.14453477868112014,
|
| 147 |
-
"grad_norm": 0.
|
| 148 |
"learning_rate": 1.9632480754904396e-05,
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 200
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.15176151761517614,
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
"learning_rate": 1.9582815992053637e-05,
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 210
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.15898825654923215,
|
| 161 |
-
"grad_norm": 0.
|
| 162 |
"learning_rate": 1.953315122920288e-05,
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 220
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.16621499548328816,
|
| 168 |
-
"grad_norm": 0.
|
| 169 |
"learning_rate": 1.9483486466352125e-05,
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 230
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.17344173441734417,
|
| 175 |
-
"grad_norm": 0.
|
| 176 |
"learning_rate": 1.943382170350137e-05,
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 240
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.18066847335140018,
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
"learning_rate": 1.938415694065061e-05,
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.1878952122854562,
|
| 189 |
-
"grad_norm": 0.
|
| 190 |
"learning_rate": 1.9334492177799853e-05,
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 260
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.1951219512195122,
|
| 196 |
-
"grad_norm": 0.
|
| 197 |
"learning_rate": 1.9284827414949094e-05,
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 270
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.2023486901535682,
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
"learning_rate": 1.9235162652098338e-05,
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 280
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.20957542908762422,
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
"learning_rate": 1.9185497889247578e-05,
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 290
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.21680216802168023,
|
| 217 |
-
"grad_norm": 0.
|
| 218 |
"learning_rate": 1.9135833126396822e-05,
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 300
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.2240289069557362,
|
| 224 |
-
"grad_norm": 0.
|
| 225 |
"learning_rate": 1.9086168363546066e-05,
|
| 226 |
"loss": 0.0239,
|
| 227 |
"step": 310
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.23125564588979222,
|
| 231 |
-
"grad_norm": 0.
|
| 232 |
"learning_rate": 1.903650360069531e-05,
|
| 233 |
"loss": 0.0234,
|
| 234 |
"step": 320
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.23848238482384823,
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
"learning_rate": 1.898683883784455e-05,
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 330
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.24570912375790424,
|
| 245 |
-
"grad_norm": 0.
|
| 246 |
"learning_rate": 1.8937174074993794e-05,
|
| 247 |
"loss": 0.0191,
|
| 248 |
"step": 340
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.2529358626919603,
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
"learning_rate": 1.8887509312143035e-05,
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 350
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.2601626016260163,
|
| 259 |
-
"grad_norm": 0.
|
| 260 |
"learning_rate": 1.883784454929228e-05,
|
| 261 |
"loss": 0.018,
|
| 262 |
"step": 360
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.26738934056007224,
|
| 266 |
-
"grad_norm": 0.
|
| 267 |
"learning_rate": 1.878817978644152e-05,
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 370
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.27461607949412825,
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
"learning_rate": 1.8738515023590763e-05,
|
| 275 |
"loss": 0.0182,
|
| 276 |
"step": 380
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.28184281842818426,
|
| 280 |
-
"grad_norm":
|
| 281 |
"learning_rate": 1.8688850260740007e-05,
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 390
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.28906955736224027,
|
| 287 |
-
"grad_norm": 0.
|
| 288 |
"learning_rate": 1.8639185497889248e-05,
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 400
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.2962962962962963,
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
"learning_rate": 1.8589520735038492e-05,
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 410
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.3035230352303523,
|
| 301 |
-
"grad_norm": 0.
|
| 302 |
"learning_rate": 1.8539855972187736e-05,
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 420
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.3107497741644083,
|
| 308 |
-
"grad_norm": 0.
|
| 309 |
"learning_rate": 1.8490191209336976e-05,
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 430
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3179765130984643,
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
"learning_rate": 1.844052644648622e-05,
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 440
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3252032520325203,
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
"learning_rate": 1.839086168363546e-05,
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 450
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.3324299909665763,
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
"learning_rate": 1.8341196920784705e-05,
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 460
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.33965672990063234,
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
"learning_rate": 1.8291532157933945e-05,
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 470
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.34688346883468835,
|
| 343 |
-
"grad_norm": 1.
|
| 344 |
"learning_rate": 1.824186739508319e-05,
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 480
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.35411020776874436,
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
"learning_rate": 1.8192202632232433e-05,
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 490
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.36133694670280037,
|
| 357 |
-
"grad_norm": 0.
|
| 358 |
"learning_rate": 1.8142537869381677e-05,
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.36133694670280037,
|
| 364 |
-
"eval_loss": 0.
|
| 365 |
-
"eval_runtime": 43.
|
| 366 |
-
"eval_samples_per_second": 26.
|
| 367 |
-
"eval_steps_per_second": 6.
|
| 368 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 369 |
}
|
| 370 |
],
|
| 371 |
"logging_steps": 10,
|
|
@@ -385,7 +1101,7 @@
|
|
| 385 |
"attributes": {}
|
| 386 |
}
|
| 387 |
},
|
| 388 |
-
"total_flos":
|
| 389 |
"train_batch_size": 4,
|
| 390 |
"trial_name": null,
|
| 391 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 1500,
|
| 3 |
+
"best_metric": 0.022791102528572083,
|
| 4 |
+
"best_model_checkpoint": "./trained_model_20251223_144504/checkpoint-1500",
|
| 5 |
+
"epoch": 1.0838301716350496,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 1500,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.007226738934056007,
|
| 14 |
+
"grad_norm": 1.7734375,
|
| 15 |
"learning_rate": 1.44e-06,
|
| 16 |
+
"loss": 0.0577,
|
| 17 |
"step": 10
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.014453477868112014,
|
| 21 |
+
"grad_norm": 1.3828125,
|
| 22 |
"learning_rate": 3.04e-06,
|
| 23 |
+
"loss": 0.0498,
|
| 24 |
"step": 20
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.02168021680216802,
|
| 28 |
+
"grad_norm": 0.890625,
|
| 29 |
"learning_rate": 4.6400000000000005e-06,
|
| 30 |
+
"loss": 0.0334,
|
| 31 |
"step": 30
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.028906955736224028,
|
| 35 |
+
"grad_norm": 0.458984375,
|
| 36 |
"learning_rate": 6.24e-06,
|
| 37 |
+
"loss": 0.0249,
|
| 38 |
"step": 40
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.036133694670280034,
|
| 42 |
+
"grad_norm": 0.2578125,
|
| 43 |
"learning_rate": 7.840000000000001e-06,
|
| 44 |
+
"loss": 0.0194,
|
| 45 |
"step": 50
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.04336043360433604,
|
| 49 |
+
"grad_norm": 0.333984375,
|
| 50 |
"learning_rate": 9.440000000000001e-06,
|
| 51 |
+
"loss": 0.0193,
|
| 52 |
"step": 60
|
| 53 |
},
|
| 54 |
{
|
| 55 |
"epoch": 0.05058717253839205,
|
| 56 |
+
"grad_norm": 0.380859375,
|
| 57 |
"learning_rate": 1.1040000000000001e-05,
|
| 58 |
+
"loss": 0.0274,
|
| 59 |
"step": 70
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.057813911472448055,
|
| 63 |
+
"grad_norm": 0.69921875,
|
| 64 |
"learning_rate": 1.2640000000000001e-05,
|
| 65 |
+
"loss": 0.0188,
|
| 66 |
"step": 80
|
| 67 |
},
|
| 68 |
{
|
| 69 |
"epoch": 0.06504065040650407,
|
| 70 |
+
"grad_norm": 0.271484375,
|
| 71 |
"learning_rate": 1.4240000000000001e-05,
|
| 72 |
+
"loss": 0.0194,
|
| 73 |
"step": 90
|
| 74 |
},
|
| 75 |
{
|
| 76 |
"epoch": 0.07226738934056007,
|
| 77 |
+
"grad_norm": 0.4375,
|
| 78 |
"learning_rate": 1.584e-05,
|
| 79 |
+
"loss": 0.0188,
|
| 80 |
"step": 100
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.07949412827461608,
|
| 84 |
+
"grad_norm": 0.265625,
|
| 85 |
"learning_rate": 1.7440000000000002e-05,
|
| 86 |
+
"loss": 0.018,
|
| 87 |
"step": 110
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.08672086720867209,
|
| 91 |
+
"grad_norm": 0.294921875,
|
| 92 |
"learning_rate": 1.904e-05,
|
| 93 |
+
"loss": 0.019,
|
| 94 |
"step": 120
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"epoch": 0.0939476061427281,
|
| 98 |
+
"grad_norm": 0.34375,
|
| 99 |
"learning_rate": 1.99801340948597e-05,
|
| 100 |
+
"loss": 0.0193,
|
| 101 |
"step": 130
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.1011743450767841,
|
| 105 |
+
"grad_norm": 0.5546875,
|
| 106 |
"learning_rate": 1.9930469332008943e-05,
|
| 107 |
+
"loss": 0.0184,
|
| 108 |
"step": 140
|
| 109 |
},
|
| 110 |
{
|
| 111 |
"epoch": 0.10840108401084012,
|
| 112 |
+
"grad_norm": 0.3046875,
|
| 113 |
"learning_rate": 1.9880804569158184e-05,
|
| 114 |
+
"loss": 0.0184,
|
| 115 |
"step": 150
|
| 116 |
},
|
| 117 |
{
|
| 118 |
"epoch": 0.11562782294489611,
|
| 119 |
+
"grad_norm": 0.2216796875,
|
| 120 |
"learning_rate": 1.9831139806307428e-05,
|
| 121 |
+
"loss": 0.019,
|
| 122 |
"step": 160
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.12285456187895212,
|
| 126 |
+
"grad_norm": 0.8671875,
|
| 127 |
"learning_rate": 1.9781475043456668e-05,
|
| 128 |
"loss": 0.0217,
|
| 129 |
"step": 170
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.13008130081300814,
|
| 133 |
+
"grad_norm": 0.369140625,
|
| 134 |
"learning_rate": 1.9731810280605912e-05,
|
| 135 |
+
"loss": 0.0187,
|
| 136 |
"step": 180
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.13730803974706413,
|
| 140 |
+
"grad_norm": 0.271484375,
|
| 141 |
"learning_rate": 1.9682145517755153e-05,
|
| 142 |
"loss": 0.029,
|
| 143 |
"step": 190
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 0.14453477868112014,
|
| 147 |
+
"grad_norm": 0.40234375,
|
| 148 |
"learning_rate": 1.9632480754904396e-05,
|
| 149 |
+
"loss": 0.0268,
|
| 150 |
"step": 200
|
| 151 |
},
|
| 152 |
{
|
| 153 |
"epoch": 0.15176151761517614,
|
| 154 |
+
"grad_norm": 0.330078125,
|
| 155 |
"learning_rate": 1.9582815992053637e-05,
|
| 156 |
+
"loss": 0.019,
|
| 157 |
"step": 210
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.15898825654923215,
|
| 161 |
+
"grad_norm": 0.73828125,
|
| 162 |
"learning_rate": 1.953315122920288e-05,
|
| 163 |
+
"loss": 0.0295,
|
| 164 |
"step": 220
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.16621499548328816,
|
| 168 |
+
"grad_norm": 0.5390625,
|
| 169 |
"learning_rate": 1.9483486466352125e-05,
|
| 170 |
+
"loss": 0.0224,
|
| 171 |
"step": 230
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.17344173441734417,
|
| 175 |
+
"grad_norm": 0.419921875,
|
| 176 |
"learning_rate": 1.943382170350137e-05,
|
| 177 |
+
"loss": 0.0193,
|
| 178 |
"step": 240
|
| 179 |
},
|
| 180 |
{
|
| 181 |
"epoch": 0.18066847335140018,
|
| 182 |
+
"grad_norm": 0.396484375,
|
| 183 |
"learning_rate": 1.938415694065061e-05,
|
| 184 |
+
"loss": 0.0188,
|
| 185 |
"step": 250
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 0.1878952122854562,
|
| 189 |
+
"grad_norm": 0.494140625,
|
| 190 |
"learning_rate": 1.9334492177799853e-05,
|
| 191 |
+
"loss": 0.0211,
|
| 192 |
"step": 260
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.1951219512195122,
|
| 196 |
+
"grad_norm": 0.46875,
|
| 197 |
"learning_rate": 1.9284827414949094e-05,
|
| 198 |
+
"loss": 0.0183,
|
| 199 |
"step": 270
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.2023486901535682,
|
| 203 |
+
"grad_norm": 0.404296875,
|
| 204 |
"learning_rate": 1.9235162652098338e-05,
|
| 205 |
+
"loss": 0.0197,
|
| 206 |
"step": 280
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.20957542908762422,
|
| 210 |
+
"grad_norm": 0.384765625,
|
| 211 |
"learning_rate": 1.9185497889247578e-05,
|
| 212 |
+
"loss": 0.0174,
|
| 213 |
"step": 290
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.21680216802168023,
|
| 217 |
+
"grad_norm": 0.333984375,
|
| 218 |
"learning_rate": 1.9135833126396822e-05,
|
| 219 |
+
"loss": 0.019,
|
| 220 |
"step": 300
|
| 221 |
},
|
| 222 |
{
|
| 223 |
"epoch": 0.2240289069557362,
|
| 224 |
+
"grad_norm": 0.28125,
|
| 225 |
"learning_rate": 1.9086168363546066e-05,
|
| 226 |
"loss": 0.0239,
|
| 227 |
"step": 310
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 0.23125564588979222,
|
| 231 |
+
"grad_norm": 0.328125,
|
| 232 |
"learning_rate": 1.903650360069531e-05,
|
| 233 |
"loss": 0.0234,
|
| 234 |
"step": 320
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.23848238482384823,
|
| 238 |
+
"grad_norm": 0.419921875,
|
| 239 |
"learning_rate": 1.898683883784455e-05,
|
| 240 |
+
"loss": 0.0195,
|
| 241 |
"step": 330
|
| 242 |
},
|
| 243 |
{
|
| 244 |
"epoch": 0.24570912375790424,
|
| 245 |
+
"grad_norm": 0.33203125,
|
| 246 |
"learning_rate": 1.8937174074993794e-05,
|
| 247 |
"loss": 0.0191,
|
| 248 |
"step": 340
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.2529358626919603,
|
| 252 |
+
"grad_norm": 0.2734375,
|
| 253 |
"learning_rate": 1.8887509312143035e-05,
|
| 254 |
+
"loss": 0.0186,
|
| 255 |
"step": 350
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.2601626016260163,
|
| 259 |
+
"grad_norm": 0.64453125,
|
| 260 |
"learning_rate": 1.883784454929228e-05,
|
| 261 |
"loss": 0.018,
|
| 262 |
"step": 360
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.26738934056007224,
|
| 266 |
+
"grad_norm": 0.25390625,
|
| 267 |
"learning_rate": 1.878817978644152e-05,
|
| 268 |
+
"loss": 0.0217,
|
| 269 |
"step": 370
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.27461607949412825,
|
| 273 |
+
"grad_norm": 0.384765625,
|
| 274 |
"learning_rate": 1.8738515023590763e-05,
|
| 275 |
"loss": 0.0182,
|
| 276 |
"step": 380
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 0.28184281842818426,
|
| 280 |
+
"grad_norm": 1.1015625,
|
| 281 |
"learning_rate": 1.8688850260740007e-05,
|
| 282 |
+
"loss": 0.0257,
|
| 283 |
"step": 390
|
| 284 |
},
|
| 285 |
{
|
| 286 |
"epoch": 0.28906955736224027,
|
| 287 |
+
"grad_norm": 0.30078125,
|
| 288 |
"learning_rate": 1.8639185497889248e-05,
|
| 289 |
+
"loss": 0.0437,
|
| 290 |
"step": 400
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.2962962962962963,
|
| 294 |
+
"grad_norm": 0.42578125,
|
| 295 |
"learning_rate": 1.8589520735038492e-05,
|
| 296 |
+
"loss": 0.0607,
|
| 297 |
"step": 410
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.3035230352303523,
|
| 301 |
+
"grad_norm": 0.267578125,
|
| 302 |
"learning_rate": 1.8539855972187736e-05,
|
| 303 |
+
"loss": 0.0187,
|
| 304 |
"step": 420
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"epoch": 0.3107497741644083,
|
| 308 |
+
"grad_norm": 0.20703125,
|
| 309 |
"learning_rate": 1.8490191209336976e-05,
|
| 310 |
+
"loss": 0.0199,
|
| 311 |
"step": 430
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.3179765130984643,
|
| 315 |
+
"grad_norm": 0.5234375,
|
| 316 |
"learning_rate": 1.844052644648622e-05,
|
| 317 |
+
"loss": 0.0181,
|
| 318 |
"step": 440
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 0.3252032520325203,
|
| 322 |
+
"grad_norm": 0.296875,
|
| 323 |
"learning_rate": 1.839086168363546e-05,
|
| 324 |
+
"loss": 0.0199,
|
| 325 |
"step": 450
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"epoch": 0.3324299909665763,
|
| 329 |
+
"grad_norm": 0.3203125,
|
| 330 |
"learning_rate": 1.8341196920784705e-05,
|
| 331 |
+
"loss": 0.0223,
|
| 332 |
"step": 460
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.33965672990063234,
|
| 336 |
+
"grad_norm": 0.3125,
|
| 337 |
"learning_rate": 1.8291532157933945e-05,
|
| 338 |
+
"loss": 0.018,
|
| 339 |
"step": 470
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.34688346883468835,
|
| 343 |
+
"grad_norm": 1.0703125,
|
| 344 |
"learning_rate": 1.824186739508319e-05,
|
| 345 |
+
"loss": 0.0398,
|
| 346 |
"step": 480
|
| 347 |
},
|
| 348 |
{
|
| 349 |
"epoch": 0.35411020776874436,
|
| 350 |
+
"grad_norm": 0.357421875,
|
| 351 |
"learning_rate": 1.8192202632232433e-05,
|
| 352 |
+
"loss": 0.0286,
|
| 353 |
"step": 490
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 0.36133694670280037,
|
| 357 |
+
"grad_norm": 0.3125,
|
| 358 |
"learning_rate": 1.8142537869381677e-05,
|
| 359 |
+
"loss": 0.0193,
|
| 360 |
"step": 500
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 0.36133694670280037,
|
| 364 |
+
"eval_loss": 0.023125287145376205,
|
| 365 |
+
"eval_runtime": 43.3724,
|
| 366 |
+
"eval_samples_per_second": 26.883,
|
| 367 |
+
"eval_steps_per_second": 6.732,
|
| 368 |
"step": 500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.3685636856368564,
|
| 372 |
+
"grad_norm": 0.470703125,
|
| 373 |
+
"learning_rate": 1.8092873106530917e-05,
|
| 374 |
+
"loss": 0.0202,
|
| 375 |
+
"step": 510
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.3757904245709124,
|
| 379 |
+
"grad_norm": 0.328125,
|
| 380 |
+
"learning_rate": 1.804320834368016e-05,
|
| 381 |
+
"loss": 0.0179,
|
| 382 |
+
"step": 520
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.3830171635049684,
|
| 386 |
+
"grad_norm": 0.337890625,
|
| 387 |
+
"learning_rate": 1.7993543580829402e-05,
|
| 388 |
+
"loss": 0.0178,
|
| 389 |
+
"step": 530
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.3902439024390244,
|
| 393 |
+
"grad_norm": 0.2490234375,
|
| 394 |
+
"learning_rate": 1.7943878817978646e-05,
|
| 395 |
+
"loss": 0.0187,
|
| 396 |
+
"step": 540
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.3974706413730804,
|
| 400 |
+
"grad_norm": 0.39453125,
|
| 401 |
+
"learning_rate": 1.7894214055127886e-05,
|
| 402 |
+
"loss": 0.0194,
|
| 403 |
+
"step": 550
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.4046973803071364,
|
| 407 |
+
"grad_norm": 0.38671875,
|
| 408 |
+
"learning_rate": 1.784454929227713e-05,
|
| 409 |
+
"loss": 0.0288,
|
| 410 |
+
"step": 560
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.41192411924119243,
|
| 414 |
+
"grad_norm": 1.0,
|
| 415 |
+
"learning_rate": 1.7794884529426374e-05,
|
| 416 |
+
"loss": 0.0215,
|
| 417 |
+
"step": 570
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.41915085817524844,
|
| 421 |
+
"grad_norm": 0.6484375,
|
| 422 |
+
"learning_rate": 1.7745219766575618e-05,
|
| 423 |
+
"loss": 0.0185,
|
| 424 |
+
"step": 580
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.42637759710930445,
|
| 428 |
+
"grad_norm": 0.34375,
|
| 429 |
+
"learning_rate": 1.769555500372486e-05,
|
| 430 |
+
"loss": 0.0197,
|
| 431 |
+
"step": 590
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.43360433604336046,
|
| 435 |
+
"grad_norm": 0.458984375,
|
| 436 |
+
"learning_rate": 1.7645890240874102e-05,
|
| 437 |
+
"loss": 0.0191,
|
| 438 |
+
"step": 600
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.4408310749774164,
|
| 442 |
+
"grad_norm": 1.015625,
|
| 443 |
+
"learning_rate": 1.7596225478023343e-05,
|
| 444 |
+
"loss": 0.0218,
|
| 445 |
+
"step": 610
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.4480578139114724,
|
| 449 |
+
"grad_norm": 0.416015625,
|
| 450 |
+
"learning_rate": 1.7546560715172587e-05,
|
| 451 |
+
"loss": 0.0199,
|
| 452 |
+
"step": 620
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.45528455284552843,
|
| 456 |
+
"grad_norm": 0.80078125,
|
| 457 |
+
"learning_rate": 1.7496895952321827e-05,
|
| 458 |
+
"loss": 0.0256,
|
| 459 |
+
"step": 630
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.46251129177958444,
|
| 463 |
+
"grad_norm": 0.46484375,
|
| 464 |
+
"learning_rate": 1.744723118947107e-05,
|
| 465 |
+
"loss": 0.0195,
|
| 466 |
+
"step": 640
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.46973803071364045,
|
| 470 |
+
"grad_norm": 0.54296875,
|
| 471 |
+
"learning_rate": 1.7397566426620315e-05,
|
| 472 |
+
"loss": 0.0191,
|
| 473 |
+
"step": 650
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.47696476964769646,
|
| 477 |
+
"grad_norm": 0.294921875,
|
| 478 |
+
"learning_rate": 1.7347901663769556e-05,
|
| 479 |
+
"loss": 0.0194,
|
| 480 |
+
"step": 660
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.48419150858175247,
|
| 484 |
+
"grad_norm": 0.41796875,
|
| 485 |
+
"learning_rate": 1.72982369009188e-05,
|
| 486 |
+
"loss": 0.0182,
|
| 487 |
+
"step": 670
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.4914182475158085,
|
| 491 |
+
"grad_norm": 0.484375,
|
| 492 |
+
"learning_rate": 1.7248572138068044e-05,
|
| 493 |
+
"loss": 0.0188,
|
| 494 |
+
"step": 680
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.4986449864498645,
|
| 498 |
+
"grad_norm": 0.4765625,
|
| 499 |
+
"learning_rate": 1.7198907375217284e-05,
|
| 500 |
+
"loss": 0.02,
|
| 501 |
+
"step": 690
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.5058717253839206,
|
| 505 |
+
"grad_norm": 0.2578125,
|
| 506 |
+
"learning_rate": 1.7149242612366528e-05,
|
| 507 |
+
"loss": 0.0191,
|
| 508 |
+
"step": 700
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.5130984643179766,
|
| 512 |
+
"grad_norm": 0.416015625,
|
| 513 |
+
"learning_rate": 1.709957784951577e-05,
|
| 514 |
+
"loss": 0.0197,
|
| 515 |
+
"step": 710
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.5203252032520326,
|
| 519 |
+
"grad_norm": 0.388671875,
|
| 520 |
+
"learning_rate": 1.7049913086665013e-05,
|
| 521 |
+
"loss": 0.02,
|
| 522 |
+
"step": 720
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.5275519421860885,
|
| 526 |
+
"grad_norm": 1.2265625,
|
| 527 |
+
"learning_rate": 1.7000248323814253e-05,
|
| 528 |
+
"loss": 0.0233,
|
| 529 |
+
"step": 730
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.5347786811201445,
|
| 533 |
+
"grad_norm": 0.400390625,
|
| 534 |
+
"learning_rate": 1.6950583560963497e-05,
|
| 535 |
+
"loss": 0.0175,
|
| 536 |
+
"step": 740
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.5420054200542005,
|
| 540 |
+
"grad_norm": 0.416015625,
|
| 541 |
+
"learning_rate": 1.690091879811274e-05,
|
| 542 |
+
"loss": 0.019,
|
| 543 |
+
"step": 750
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.5492321589882565,
|
| 547 |
+
"grad_norm": 0.306640625,
|
| 548 |
+
"learning_rate": 1.6851254035261985e-05,
|
| 549 |
+
"loss": 0.0306,
|
| 550 |
+
"step": 760
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.5564588979223125,
|
| 554 |
+
"grad_norm": 0.2333984375,
|
| 555 |
+
"learning_rate": 1.6801589272411225e-05,
|
| 556 |
+
"loss": 0.0187,
|
| 557 |
+
"step": 770
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.5636856368563685,
|
| 561 |
+
"grad_norm": 0.2060546875,
|
| 562 |
+
"learning_rate": 1.675192450956047e-05,
|
| 563 |
+
"loss": 0.0184,
|
| 564 |
+
"step": 780
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.5709123757904245,
|
| 568 |
+
"grad_norm": 1.0390625,
|
| 569 |
+
"learning_rate": 1.670225974670971e-05,
|
| 570 |
+
"loss": 0.0242,
|
| 571 |
+
"step": 790
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.5781391147244805,
|
| 575 |
+
"grad_norm": 0.49609375,
|
| 576 |
+
"learning_rate": 1.6652594983858954e-05,
|
| 577 |
+
"loss": 0.0228,
|
| 578 |
+
"step": 800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.5853658536585366,
|
| 582 |
+
"grad_norm": 0.3828125,
|
| 583 |
+
"learning_rate": 1.6602930221008194e-05,
|
| 584 |
+
"loss": 0.0197,
|
| 585 |
+
"step": 810
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.5925925925925926,
|
| 589 |
+
"grad_norm": 0.4453125,
|
| 590 |
+
"learning_rate": 1.6553265458157438e-05,
|
| 591 |
+
"loss": 0.0192,
|
| 592 |
+
"step": 820
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.5998193315266486,
|
| 596 |
+
"grad_norm": 0.310546875,
|
| 597 |
+
"learning_rate": 1.6503600695306682e-05,
|
| 598 |
+
"loss": 0.0176,
|
| 599 |
+
"step": 830
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.6070460704607046,
|
| 603 |
+
"grad_norm": 0.337890625,
|
| 604 |
+
"learning_rate": 1.6453935932455926e-05,
|
| 605 |
+
"loss": 0.0228,
|
| 606 |
+
"step": 840
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 0.6142728093947606,
|
| 610 |
+
"grad_norm": 0.3984375,
|
| 611 |
+
"learning_rate": 1.6404271169605167e-05,
|
| 612 |
+
"loss": 0.0187,
|
| 613 |
+
"step": 850
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.6214995483288166,
|
| 617 |
+
"grad_norm": 0.34765625,
|
| 618 |
+
"learning_rate": 1.635460640675441e-05,
|
| 619 |
+
"loss": 0.0191,
|
| 620 |
+
"step": 860
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.6287262872628726,
|
| 624 |
+
"grad_norm": 0.421875,
|
| 625 |
+
"learning_rate": 1.630494164390365e-05,
|
| 626 |
+
"loss": 0.0182,
|
| 627 |
+
"step": 870
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.6359530261969286,
|
| 631 |
+
"grad_norm": 0.3359375,
|
| 632 |
+
"learning_rate": 1.6255276881052895e-05,
|
| 633 |
+
"loss": 0.019,
|
| 634 |
+
"step": 880
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.6431797651309846,
|
| 638 |
+
"grad_norm": 0.353515625,
|
| 639 |
+
"learning_rate": 1.6205612118202136e-05,
|
| 640 |
+
"loss": 0.0192,
|
| 641 |
+
"step": 890
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.6504065040650406,
|
| 645 |
+
"grad_norm": 0.357421875,
|
| 646 |
+
"learning_rate": 1.615594735535138e-05,
|
| 647 |
+
"loss": 0.0199,
|
| 648 |
+
"step": 900
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.6576332429990966,
|
| 652 |
+
"grad_norm": 0.357421875,
|
| 653 |
+
"learning_rate": 1.610628259250062e-05,
|
| 654 |
+
"loss": 0.0182,
|
| 655 |
+
"step": 910
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.6648599819331527,
|
| 659 |
+
"grad_norm": 0.36328125,
|
| 660 |
+
"learning_rate": 1.6056617829649864e-05,
|
| 661 |
+
"loss": 0.0176,
|
| 662 |
+
"step": 920
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.6720867208672087,
|
| 666 |
+
"grad_norm": 0.328125,
|
| 667 |
+
"learning_rate": 1.6006953066799108e-05,
|
| 668 |
+
"loss": 0.0197,
|
| 669 |
+
"step": 930
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.6793134598012647,
|
| 673 |
+
"grad_norm": 0.4296875,
|
| 674 |
+
"learning_rate": 1.5957288303948352e-05,
|
| 675 |
+
"loss": 0.0188,
|
| 676 |
+
"step": 940
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.6865401987353207,
|
| 680 |
+
"grad_norm": 0.345703125,
|
| 681 |
+
"learning_rate": 1.5907623541097592e-05,
|
| 682 |
+
"loss": 0.0196,
|
| 683 |
+
"step": 950
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.6937669376693767,
|
| 687 |
+
"grad_norm": 0.44921875,
|
| 688 |
+
"learning_rate": 1.5857958778246836e-05,
|
| 689 |
+
"loss": 0.019,
|
| 690 |
+
"step": 960
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.7009936766034327,
|
| 694 |
+
"grad_norm": 0.37109375,
|
| 695 |
+
"learning_rate": 1.5808294015396077e-05,
|
| 696 |
+
"loss": 0.0287,
|
| 697 |
+
"step": 970
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.7082204155374887,
|
| 701 |
+
"grad_norm": 0.44921875,
|
| 702 |
+
"learning_rate": 1.575862925254532e-05,
|
| 703 |
+
"loss": 0.0193,
|
| 704 |
+
"step": 980
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.7154471544715447,
|
| 708 |
+
"grad_norm": 0.28515625,
|
| 709 |
+
"learning_rate": 1.570896448969456e-05,
|
| 710 |
+
"loss": 0.0185,
|
| 711 |
+
"step": 990
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.7226738934056007,
|
| 715 |
+
"grad_norm": 0.5625,
|
| 716 |
+
"learning_rate": 1.5659299726843805e-05,
|
| 717 |
+
"loss": 0.0181,
|
| 718 |
+
"step": 1000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.7226738934056007,
|
| 722 |
+
"eval_loss": 0.02289458177983761,
|
| 723 |
+
"eval_runtime": 43.4716,
|
| 724 |
+
"eval_samples_per_second": 26.822,
|
| 725 |
+
"eval_steps_per_second": 6.717,
|
| 726 |
+
"step": 1000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.7299006323396567,
|
| 730 |
+
"grad_norm": 0.314453125,
|
| 731 |
+
"learning_rate": 1.560963496399305e-05,
|
| 732 |
+
"loss": 0.0181,
|
| 733 |
+
"step": 1010
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.7371273712737128,
|
| 737 |
+
"grad_norm": 0.333984375,
|
| 738 |
+
"learning_rate": 1.5559970201142293e-05,
|
| 739 |
+
"loss": 0.0517,
|
| 740 |
+
"step": 1020
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.7443541102077688,
|
| 744 |
+
"grad_norm": 0.43359375,
|
| 745 |
+
"learning_rate": 1.5510305438291533e-05,
|
| 746 |
+
"loss": 0.0314,
|
| 747 |
+
"step": 1030
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.7515808491418248,
|
| 751 |
+
"grad_norm": 0.60546875,
|
| 752 |
+
"learning_rate": 1.5460640675440777e-05,
|
| 753 |
+
"loss": 0.0185,
|
| 754 |
+
"step": 1040
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.7588075880758808,
|
| 758 |
+
"grad_norm": 0.54296875,
|
| 759 |
+
"learning_rate": 1.5410975912590018e-05,
|
| 760 |
+
"loss": 0.0183,
|
| 761 |
+
"step": 1050
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.7660343270099368,
|
| 765 |
+
"grad_norm": 0.30078125,
|
| 766 |
+
"learning_rate": 1.5361311149739262e-05,
|
| 767 |
+
"loss": 0.0193,
|
| 768 |
+
"step": 1060
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.7732610659439928,
|
| 772 |
+
"grad_norm": 0.322265625,
|
| 773 |
+
"learning_rate": 1.5311646386888502e-05,
|
| 774 |
+
"loss": 0.0185,
|
| 775 |
+
"step": 1070
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.7804878048780488,
|
| 779 |
+
"grad_norm": 0.375,
|
| 780 |
+
"learning_rate": 1.5261981624037746e-05,
|
| 781 |
+
"loss": 0.0187,
|
| 782 |
+
"step": 1080
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.7877145438121048,
|
| 786 |
+
"grad_norm": 0.373046875,
|
| 787 |
+
"learning_rate": 1.5212316861186989e-05,
|
| 788 |
+
"loss": 0.0176,
|
| 789 |
+
"step": 1090
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.7949412827461608,
|
| 793 |
+
"grad_norm": 0.328125,
|
| 794 |
+
"learning_rate": 1.5162652098336232e-05,
|
| 795 |
+
"loss": 0.0189,
|
| 796 |
+
"step": 1100
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.8021680216802168,
|
| 800 |
+
"grad_norm": 0.333984375,
|
| 801 |
+
"learning_rate": 1.5112987335485475e-05,
|
| 802 |
+
"loss": 0.0189,
|
| 803 |
+
"step": 1110
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.8093947606142728,
|
| 807 |
+
"grad_norm": 0.248046875,
|
| 808 |
+
"learning_rate": 1.5063322572634717e-05,
|
| 809 |
+
"loss": 0.0174,
|
| 810 |
+
"step": 1120
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.8166214995483289,
|
| 814 |
+
"grad_norm": 0.2216796875,
|
| 815 |
+
"learning_rate": 1.5013657809783959e-05,
|
| 816 |
+
"loss": 0.018,
|
| 817 |
+
"step": 1130
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.8238482384823849,
|
| 821 |
+
"grad_norm": 0.37109375,
|
| 822 |
+
"learning_rate": 1.4963993046933203e-05,
|
| 823 |
+
"loss": 0.0178,
|
| 824 |
+
"step": 1140
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.8310749774164409,
|
| 828 |
+
"grad_norm": 0.28125,
|
| 829 |
+
"learning_rate": 1.4914328284082444e-05,
|
| 830 |
+
"loss": 0.0165,
|
| 831 |
+
"step": 1150
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.8383017163504969,
|
| 835 |
+
"grad_norm": 0.35546875,
|
| 836 |
+
"learning_rate": 1.4864663521231688e-05,
|
| 837 |
+
"loss": 0.0238,
|
| 838 |
+
"step": 1160
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.8455284552845529,
|
| 842 |
+
"grad_norm": 0.265625,
|
| 843 |
+
"learning_rate": 1.481499875838093e-05,
|
| 844 |
+
"loss": 0.0195,
|
| 845 |
+
"step": 1170
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.8527551942186089,
|
| 849 |
+
"grad_norm": 2.46875,
|
| 850 |
+
"learning_rate": 1.4765333995530174e-05,
|
| 851 |
+
"loss": 0.0424,
|
| 852 |
+
"step": 1180
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.8599819331526649,
|
| 856 |
+
"grad_norm": 0.51953125,
|
| 857 |
+
"learning_rate": 1.4715669232679414e-05,
|
| 858 |
+
"loss": 0.0203,
|
| 859 |
+
"step": 1190
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.8672086720867209,
|
| 863 |
+
"grad_norm": 0.47265625,
|
| 864 |
+
"learning_rate": 1.4666004469828658e-05,
|
| 865 |
+
"loss": 0.0181,
|
| 866 |
+
"step": 1200
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.8744354110207768,
|
| 870 |
+
"grad_norm": 0.451171875,
|
| 871 |
+
"learning_rate": 1.46163397069779e-05,
|
| 872 |
+
"loss": 0.0189,
|
| 873 |
+
"step": 1210
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.8816621499548328,
|
| 877 |
+
"grad_norm": 0.380859375,
|
| 878 |
+
"learning_rate": 1.4566674944127144e-05,
|
| 879 |
+
"loss": 0.0322,
|
| 880 |
+
"step": 1220
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 0.8888888888888888,
|
| 884 |
+
"grad_norm": 0.298828125,
|
| 885 |
+
"learning_rate": 1.4517010181276385e-05,
|
| 886 |
+
"loss": 0.019,
|
| 887 |
+
"step": 1230
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.8961156278229448,
|
| 891 |
+
"grad_norm": 0.248046875,
|
| 892 |
+
"learning_rate": 1.4467345418425629e-05,
|
| 893 |
+
"loss": 0.0253,
|
| 894 |
+
"step": 1240
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.9033423667570009,
|
| 898 |
+
"grad_norm": 0.39453125,
|
| 899 |
+
"learning_rate": 1.4417680655574871e-05,
|
| 900 |
+
"loss": 0.018,
|
| 901 |
+
"step": 1250
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.9105691056910569,
|
| 905 |
+
"grad_norm": 0.345703125,
|
| 906 |
+
"learning_rate": 1.4368015892724115e-05,
|
| 907 |
+
"loss": 0.0217,
|
| 908 |
+
"step": 1260
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.9177958446251129,
|
| 912 |
+
"grad_norm": 0.330078125,
|
| 913 |
+
"learning_rate": 1.4318351129873355e-05,
|
| 914 |
+
"loss": 0.0336,
|
| 915 |
+
"step": 1270
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.9250225835591689,
|
| 919 |
+
"grad_norm": 0.267578125,
|
| 920 |
+
"learning_rate": 1.42686863670226e-05,
|
| 921 |
+
"loss": 0.0219,
|
| 922 |
+
"step": 1280
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.9322493224932249,
|
| 926 |
+
"grad_norm": 0.65234375,
|
| 927 |
+
"learning_rate": 1.4219021604171842e-05,
|
| 928 |
+
"loss": 0.0212,
|
| 929 |
+
"step": 1290
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.9394760614272809,
|
| 933 |
+
"grad_norm": 0.259765625,
|
| 934 |
+
"learning_rate": 1.4169356841321085e-05,
|
| 935 |
+
"loss": 0.0172,
|
| 936 |
+
"step": 1300
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 0.9467028003613369,
|
| 940 |
+
"grad_norm": 0.61328125,
|
| 941 |
+
"learning_rate": 1.4119692078470326e-05,
|
| 942 |
+
"loss": 0.0187,
|
| 943 |
+
"step": 1310
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.9539295392953929,
|
| 947 |
+
"grad_norm": 0.3671875,
|
| 948 |
+
"learning_rate": 1.407002731561957e-05,
|
| 949 |
+
"loss": 0.0179,
|
| 950 |
+
"step": 1320
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.9611562782294489,
|
| 954 |
+
"grad_norm": 0.3046875,
|
| 955 |
+
"learning_rate": 1.4020362552768812e-05,
|
| 956 |
+
"loss": 0.0179,
|
| 957 |
+
"step": 1330
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.9683830171635049,
|
| 961 |
+
"grad_norm": 0.56640625,
|
| 962 |
+
"learning_rate": 1.3970697789918054e-05,
|
| 963 |
+
"loss": 0.0187,
|
| 964 |
+
"step": 1340
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.975609756097561,
|
| 968 |
+
"grad_norm": 0.380859375,
|
| 969 |
+
"learning_rate": 1.3921033027067297e-05,
|
| 970 |
+
"loss": 0.0193,
|
| 971 |
+
"step": 1350
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 0.982836495031617,
|
| 975 |
+
"grad_norm": 0.38671875,
|
| 976 |
+
"learning_rate": 1.387136826421654e-05,
|
| 977 |
+
"loss": 0.02,
|
| 978 |
+
"step": 1360
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 0.990063233965673,
|
| 982 |
+
"grad_norm": 0.34375,
|
| 983 |
+
"learning_rate": 1.3821703501365781e-05,
|
| 984 |
+
"loss": 0.0171,
|
| 985 |
+
"step": 1370
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 0.997289972899729,
|
| 989 |
+
"grad_norm": 0.3203125,
|
| 990 |
+
"learning_rate": 1.3772038738515025e-05,
|
| 991 |
+
"loss": 0.0197,
|
| 992 |
+
"step": 1380
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.0043360433604336,
|
| 996 |
+
"grad_norm": 0.2333984375,
|
| 997 |
+
"learning_rate": 1.3722373975664267e-05,
|
| 998 |
+
"loss": 0.0167,
|
| 999 |
+
"step": 1390
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 1.0115627822944897,
|
| 1003 |
+
"grad_norm": 0.1865234375,
|
| 1004 |
+
"learning_rate": 1.3672709212813511e-05,
|
| 1005 |
+
"loss": 0.0184,
|
| 1006 |
+
"step": 1400
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 1.0187895212285456,
|
| 1010 |
+
"grad_norm": 0.2099609375,
|
| 1011 |
+
"learning_rate": 1.3623044449962752e-05,
|
| 1012 |
+
"loss": 0.015,
|
| 1013 |
+
"step": 1410
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 1.0260162601626017,
|
| 1017 |
+
"grad_norm": 0.169921875,
|
| 1018 |
+
"learning_rate": 1.3573379687111996e-05,
|
| 1019 |
+
"loss": 0.0145,
|
| 1020 |
+
"step": 1420
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.0332429990966576,
|
| 1024 |
+
"grad_norm": 0.19921875,
|
| 1025 |
+
"learning_rate": 1.3523714924261238e-05,
|
| 1026 |
+
"loss": 0.0151,
|
| 1027 |
+
"step": 1430
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 1.0404697380307137,
|
| 1031 |
+
"grad_norm": 0.1337890625,
|
| 1032 |
+
"learning_rate": 1.3474050161410482e-05,
|
| 1033 |
+
"loss": 0.016,
|
| 1034 |
+
"step": 1440
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.0476964769647696,
|
| 1038 |
+
"grad_norm": 0.173828125,
|
| 1039 |
+
"learning_rate": 1.3424385398559722e-05,
|
| 1040 |
+
"loss": 0.0153,
|
| 1041 |
+
"step": 1450
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 1.0549232158988258,
|
| 1045 |
+
"grad_norm": 0.2275390625,
|
| 1046 |
+
"learning_rate": 1.3374720635708966e-05,
|
| 1047 |
+
"loss": 0.02,
|
| 1048 |
+
"step": 1460
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 1.0621499548328817,
|
| 1052 |
+
"grad_norm": 0.2451171875,
|
| 1053 |
+
"learning_rate": 1.3325055872858208e-05,
|
| 1054 |
+
"loss": 0.0176,
|
| 1055 |
+
"step": 1470
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.0693766937669378,
|
| 1059 |
+
"grad_norm": 0.287109375,
|
| 1060 |
+
"learning_rate": 1.3275391110007452e-05,
|
| 1061 |
+
"loss": 0.0175,
|
| 1062 |
+
"step": 1480
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.0766034327009937,
|
| 1066 |
+
"grad_norm": 0.15234375,
|
| 1067 |
+
"learning_rate": 1.3225726347156693e-05,
|
| 1068 |
+
"loss": 0.015,
|
| 1069 |
+
"step": 1490
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 1.0838301716350496,
|
| 1073 |
+
"grad_norm": 0.15234375,
|
| 1074 |
+
"learning_rate": 1.3176061584305937e-05,
|
| 1075 |
+
"loss": 0.0226,
|
| 1076 |
+
"step": 1500
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.0838301716350496,
|
| 1080 |
+
"eval_loss": 0.022791102528572083,
|
| 1081 |
+
"eval_runtime": 43.397,
|
| 1082 |
+
"eval_samples_per_second": 26.868,
|
| 1083 |
+
"eval_steps_per_second": 6.729,
|
| 1084 |
+
"step": 1500
|
| 1085 |
}
|
| 1086 |
],
|
| 1087 |
"logging_steps": 10,
|
|
|
|
| 1101 |
"attributes": {}
|
| 1102 |
}
|
| 1103 |
},
|
| 1104 |
+
"total_flos": 6.261646670739886e+17,
|
| 1105 |
"train_batch_size": 4,
|
| 1106 |
"trial_name": null,
|
| 1107 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5841
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b079f0fe6e2318616cd3018ea7ee3f0504303f3997e29be89e5fcb15bb84cb2c
|
| 3 |
size 5841
|