Training in progress, step 20, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 847599616
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36d47f825f842719e9adc8ec2a96d35e27fb73a091c8034a06f5b1bd0e4f9af0
|
| 3 |
size 847599616
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 350603
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:03a3d87907b233ae85d2a09dc2082cbf577bf60b26f83265e798ffd5cffb3296
|
| 3 |
size 350603
|
last-checkpoint/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:e936ced482e7c9446cc6e6e7038e023dc86c658d8bb3b07d132fc5298cce6557
|
| 3 |
size 14645
|
last-checkpoint/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:ad4a07c066075e0a1b5d9477b0d2f4d6bab07235ad5efe05720caa179d7171b5
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,382 +2,49 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 20,
|
| 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": 0.025,
|
| 14 |
-
"grad_norm": 0.
|
| 15 |
"learning_rate": 1.6e-06,
|
| 16 |
-
"loss":
|
| 17 |
"step": 5
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.05,
|
| 21 |
-
"grad_norm": 0.
|
| 22 |
"learning_rate": 3.6e-06,
|
| 23 |
-
"loss":
|
| 24 |
"step": 10
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.075,
|
| 28 |
-
"grad_norm": 0.
|
| 29 |
"learning_rate": 3.995627254437549e-06,
|
| 30 |
-
"loss":
|
| 31 |
"step": 15
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.1,
|
| 35 |
-
"grad_norm": 0.
|
| 36 |
"learning_rate": 3.9778957412029366e-06,
|
| 37 |
-
"loss":
|
| 38 |
"step": 20
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.1,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 10.
|
| 44 |
-
"eval_runtime": 2.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 20
|
| 48 |
-
},
|
| 49 |
-
{
|
| 50 |
-
"epoch": 0.125,
|
| 51 |
-
"grad_norm": 0.34710416197776794,
|
| 52 |
-
"learning_rate": 3.9466531944960116e-06,
|
| 53 |
-
"loss": 21.448446655273436,
|
| 54 |
-
"step": 25
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.15,
|
| 58 |
-
"grad_norm": 0.41463083028793335,
|
| 59 |
-
"learning_rate": 3.902113032590307e-06,
|
| 60 |
-
"loss": 21.42731628417969,
|
| 61 |
-
"step": 30
|
| 62 |
-
},
|
| 63 |
-
{
|
| 64 |
-
"epoch": 0.175,
|
| 65 |
-
"grad_norm": 0.4001413583755493,
|
| 66 |
-
"learning_rate": 3.844579509954769e-06,
|
| 67 |
-
"loss": 21.273895263671875,
|
| 68 |
-
"step": 35
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.2,
|
| 72 |
-
"grad_norm": 0.37236735224723816,
|
| 73 |
-
"learning_rate": 3.7744456388872234e-06,
|
| 74 |
-
"loss": 21.325396728515624,
|
| 75 |
-
"step": 40
|
| 76 |
-
},
|
| 77 |
-
{
|
| 78 |
-
"epoch": 0.2,
|
| 79 |
-
"eval_accuracy": 0.03863492063492063,
|
| 80 |
-
"eval_loss": 10.63491153717041,
|
| 81 |
-
"eval_runtime": 2.5692,
|
| 82 |
-
"eval_samples_per_second": 389.224,
|
| 83 |
-
"eval_steps_per_second": 65.0,
|
| 84 |
-
"step": 40
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.225,
|
| 88 |
-
"grad_norm": 0.6012778878211975,
|
| 89 |
-
"learning_rate": 3.6921905048418706e-06,
|
| 90 |
-
"loss": 21.228472900390624,
|
| 91 |
-
"step": 45
|
| 92 |
-
},
|
| 93 |
-
{
|
| 94 |
-
"epoch": 0.25,
|
| 95 |
-
"grad_norm": 0.496539831161499,
|
| 96 |
-
"learning_rate": 3.598375993789849e-06,
|
| 97 |
-
"loss": 21.14513244628906,
|
| 98 |
-
"step": 50
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.275,
|
| 102 |
-
"grad_norm": 0.48418062925338745,
|
| 103 |
-
"learning_rate": 3.4936429539683075e-06,
|
| 104 |
-
"loss": 21.062947082519532,
|
| 105 |
-
"step": 55
|
| 106 |
-
},
|
| 107 |
-
{
|
| 108 |
-
"epoch": 0.3,
|
| 109 |
-
"grad_norm": 0.542204737663269,
|
| 110 |
-
"learning_rate": 3.3787068182371314e-06,
|
| 111 |
-
"loss": 20.998651123046876,
|
| 112 |
-
"step": 60
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 0.3,
|
| 116 |
-
"eval_accuracy": 0.05380952380952381,
|
| 117 |
-
"eval_loss": 10.491544723510742,
|
| 118 |
-
"eval_runtime": 2.5694,
|
| 119 |
-
"eval_samples_per_second": 389.199,
|
| 120 |
-
"eval_steps_per_second": 64.996,
|
| 121 |
-
"step": 60
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.325,
|
| 125 |
-
"grad_norm": 0.5018299221992493,
|
| 126 |
-
"learning_rate": 3.254352716947074e-06,
|
| 127 |
-
"loss": 21.052398681640625,
|
| 128 |
-
"step": 65
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.35,
|
| 132 |
-
"grad_norm": 0.5535529255867004,
|
| 133 |
-
"learning_rate": 3.1214301147033453e-06,
|
| 134 |
-
"loss": 20.81998291015625,
|
| 135 |
-
"step": 70
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.375,
|
| 139 |
-
"grad_norm": 0.6942564249038696,
|
| 140 |
-
"learning_rate": 2.9808470076610163e-06,
|
| 141 |
-
"loss": 20.625360107421876,
|
| 142 |
-
"step": 75
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.4,
|
| 146 |
-
"grad_norm": 0.6362656950950623,
|
| 147 |
-
"learning_rate": 2.833563720990581e-06,
|
| 148 |
-
"loss": 20.59354248046875,
|
| 149 |
-
"step": 80
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.4,
|
| 153 |
-
"eval_accuracy": 0.06626984126984127,
|
| 154 |
-
"eval_loss": 10.32943344116211,
|
| 155 |
-
"eval_runtime": 2.5422,
|
| 156 |
-
"eval_samples_per_second": 393.359,
|
| 157 |
-
"eval_steps_per_second": 65.691,
|
| 158 |
-
"step": 80
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.425,
|
| 162 |
-
"grad_norm": 0.6189765334129333,
|
| 163 |
-
"learning_rate": 2.680586348883286e-06,
|
| 164 |
-
"loss": 20.522158813476562,
|
| 165 |
-
"step": 85
|
| 166 |
-
},
|
| 167 |
-
{
|
| 168 |
-
"epoch": 0.45,
|
| 169 |
-
"grad_norm": 0.5756471157073975,
|
| 170 |
-
"learning_rate": 2.5229598819076393e-06,
|
| 171 |
-
"loss": 20.492315673828124,
|
| 172 |
-
"step": 90
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.475,
|
| 176 |
-
"grad_norm": 0.5617444515228271,
|
| 177 |
-
"learning_rate": 2.36176106866422e-06,
|
| 178 |
-
"loss": 20.49109802246094,
|
| 179 |
-
"step": 95
|
| 180 |
-
},
|
| 181 |
-
{
|
| 182 |
-
"epoch": 0.5,
|
| 183 |
-
"grad_norm": 0.5637349486351013,
|
| 184 |
-
"learning_rate": 2.198091060500926e-06,
|
| 185 |
-
"loss": 20.489166259765625,
|
| 186 |
-
"step": 100
|
| 187 |
-
},
|
| 188 |
-
{
|
| 189 |
-
"epoch": 0.5,
|
| 190 |
-
"eval_accuracy": 0.0712063492063492,
|
| 191 |
-
"eval_loss": 10.189505577087402,
|
| 192 |
-
"eval_runtime": 2.8969,
|
| 193 |
-
"eval_samples_per_second": 345.191,
|
| 194 |
-
"eval_steps_per_second": 57.647,
|
| 195 |
-
"step": 100
|
| 196 |
-
},
|
| 197 |
-
{
|
| 198 |
-
"epoch": 0.525,
|
| 199 |
-
"grad_norm": 0.5826505422592163,
|
| 200 |
-
"learning_rate": 2.033067889532717e-06,
|
| 201 |
-
"loss": 20.34880828857422,
|
| 202 |
-
"step": 105
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"epoch": 0.55,
|
| 206 |
-
"grad_norm": 0.5714936256408691,
|
| 207 |
-
"learning_rate": 1.8678188313486012e-06,
|
| 208 |
-
"loss": 20.163304138183594,
|
| 209 |
-
"step": 110
|
| 210 |
-
},
|
| 211 |
-
{
|
| 212 |
-
"epoch": 0.575,
|
| 213 |
-
"grad_norm": 0.7302822470664978,
|
| 214 |
-
"learning_rate": 1.7034727045763157e-06,
|
| 215 |
-
"loss": 20.174502563476562,
|
| 216 |
-
"step": 115
|
| 217 |
-
},
|
| 218 |
-
{
|
| 219 |
-
"epoch": 0.6,
|
| 220 |
-
"grad_norm": 0.6190714836120605,
|
| 221 |
-
"learning_rate": 1.541152159906497e-06,
|
| 222 |
-
"loss": 20.209898376464842,
|
| 223 |
-
"step": 120
|
| 224 |
-
},
|
| 225 |
-
{
|
| 226 |
-
"epoch": 0.6,
|
| 227 |
-
"eval_accuracy": 0.07126984126984127,
|
| 228 |
-
"eval_loss": 10.068880081176758,
|
| 229 |
-
"eval_runtime": 2.8029,
|
| 230 |
-
"eval_samples_per_second": 356.77,
|
| 231 |
-
"eval_steps_per_second": 59.581,
|
| 232 |
-
"step": 120
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.625,
|
| 236 |
-
"grad_norm": 0.5889272093772888,
|
| 237 |
-
"learning_rate": 1.3819660112501052e-06,
|
| 238 |
-
"loss": 20.15807647705078,
|
| 239 |
-
"step": 125
|
| 240 |
-
},
|
| 241 |
-
{
|
| 242 |
-
"epoch": 0.65,
|
| 243 |
-
"grad_norm": 0.5750179886817932,
|
| 244 |
-
"learning_rate": 1.227001661415096e-06,
|
| 245 |
-
"loss": 20.040406799316408,
|
| 246 |
-
"step": 130
|
| 247 |
-
},
|
| 248 |
-
{
|
| 249 |
-
"epoch": 0.675,
|
| 250 |
-
"grad_norm": 0.719745934009552,
|
| 251 |
-
"learning_rate": 1.0773176740426246e-06,
|
| 252 |
-
"loss": 20.01085662841797,
|
| 253 |
-
"step": 135
|
| 254 |
-
},
|
| 255 |
-
{
|
| 256 |
-
"epoch": 0.7,
|
| 257 |
-
"grad_norm": 0.7709663510322571,
|
| 258 |
-
"learning_rate": 9.339365425440129e-07,
|
| 259 |
-
"loss": 19.724040222167968,
|
| 260 |
-
"step": 140
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.7,
|
| 264 |
-
"eval_accuracy": 0.07247619047619047,
|
| 265 |
-
"eval_loss": 9.954747200012207,
|
| 266 |
-
"eval_runtime": 2.6401,
|
| 267 |
-
"eval_samples_per_second": 378.767,
|
| 268 |
-
"eval_steps_per_second": 63.254,
|
| 269 |
-
"step": 140
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.725,
|
| 273 |
-
"grad_norm": 0.7675061821937561,
|
| 274 |
-
"learning_rate": 7.978377054339498e-07,
|
| 275 |
-
"loss": 19.789096069335937,
|
| 276 |
-
"step": 145
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.75,
|
| 280 |
-
"grad_norm": 0.6245450973510742,
|
| 281 |
-
"learning_rate": 6.699508557723032e-07,
|
| 282 |
-
"loss": 19.786654663085937,
|
| 283 |
-
"step": 150
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.775,
|
| 287 |
-
"grad_norm": 0.6990048289299011,
|
| 288 |
-
"learning_rate": 5.511495904178221e-07,
|
| 289 |
-
"loss": 19.784368896484374,
|
| 290 |
-
"step": 155
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.8,
|
| 294 |
-
"grad_norm": 0.6921213269233704,
|
| 295 |
-
"learning_rate": 4.4224544247577535e-07,
|
| 296 |
-
"loss": 19.664630126953124,
|
| 297 |
-
"step": 160
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.8,
|
| 301 |
-
"eval_accuracy": 0.07252380952380952,
|
| 302 |
-
"eval_loss": 9.85792350769043,
|
| 303 |
-
"eval_runtime": 2.5752,
|
| 304 |
-
"eval_samples_per_second": 388.323,
|
| 305 |
-
"eval_steps_per_second": 64.85,
|
| 306 |
-
"step": 160
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 0.825,
|
| 310 |
-
"grad_norm": 0.784702479839325,
|
| 311 |
-
"learning_rate": 3.439823377039599e-07,
|
| 312 |
-
"loss": 19.52422180175781,
|
| 313 |
-
"step": 165
|
| 314 |
-
},
|
| 315 |
-
{
|
| 316 |
-
"epoch": 0.85,
|
| 317 |
-
"grad_norm": 0.6510916948318481,
|
| 318 |
-
"learning_rate": 2.570315127454452e-07,
|
| 319 |
-
"loss": 19.38793029785156,
|
| 320 |
-
"step": 170
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.875,
|
| 324 |
-
"grad_norm": 0.7473388910293579,
|
| 325 |
-
"learning_rate": 1.8198692990167497e-07,
|
| 326 |
-
"loss": 19.620166015625,
|
| 327 |
-
"step": 175
|
| 328 |
-
},
|
| 329 |
-
{
|
| 330 |
-
"epoch": 0.9,
|
| 331 |
-
"grad_norm": 0.9092379212379456,
|
| 332 |
-
"learning_rate": 1.1936121976767767e-07,
|
| 333 |
-
"loss": 19.587705993652342,
|
| 334 |
-
"step": 180
|
| 335 |
-
},
|
| 336 |
-
{
|
| 337 |
-
"epoch": 0.9,
|
| 338 |
-
"eval_accuracy": 0.07433333333333333,
|
| 339 |
-
"eval_loss": 9.763090133666992,
|
| 340 |
-
"eval_runtime": 2.562,
|
| 341 |
-
"eval_samples_per_second": 390.315,
|
| 342 |
-
"eval_steps_per_second": 65.183,
|
| 343 |
-
"step": 180
|
| 344 |
-
},
|
| 345 |
-
{
|
| 346 |
-
"epoch": 0.925,
|
| 347 |
-
"grad_norm": 0.7244570255279541,
|
| 348 |
-
"learning_rate": 6.958217944530287e-08,
|
| 349 |
-
"loss": 19.646340942382814,
|
| 350 |
-
"step": 185
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.95,
|
| 354 |
-
"grad_norm": 0.8131192922592163,
|
| 355 |
-
"learning_rate": 3.2989850255235265e-08,
|
| 356 |
-
"loss": 19.582081604003907,
|
| 357 |
-
"step": 190
|
| 358 |
-
},
|
| 359 |
-
{
|
| 360 |
-
"epoch": 0.975,
|
| 361 |
-
"grad_norm": 0.8881648182868958,
|
| 362 |
-
"learning_rate": 9.834194909977168e-09,
|
| 363 |
-
"loss": 19.577865600585938,
|
| 364 |
-
"step": 195
|
| 365 |
-
},
|
| 366 |
-
{
|
| 367 |
-
"epoch": 1.0,
|
| 368 |
-
"grad_norm": 0.9109742641448975,
|
| 369 |
-
"learning_rate": 2.7339001506199164e-10,
|
| 370 |
-
"loss": 19.3023681640625,
|
| 371 |
-
"step": 200
|
| 372 |
-
},
|
| 373 |
-
{
|
| 374 |
-
"epoch": 1.0,
|
| 375 |
-
"eval_accuracy": 0.07755555555555556,
|
| 376 |
-
"eval_loss": 9.678764343261719,
|
| 377 |
-
"eval_runtime": 2.5817,
|
| 378 |
-
"eval_samples_per_second": 387.336,
|
| 379 |
-
"eval_steps_per_second": 64.685,
|
| 380 |
-
"step": 200
|
| 381 |
}
|
| 382 |
],
|
| 383 |
"logging_steps": 5,
|
|
@@ -392,7 +59,7 @@
|
|
| 392 |
"should_evaluate": false,
|
| 393 |
"should_log": false,
|
| 394 |
"should_save": true,
|
| 395 |
-
"should_training_stop":
|
| 396 |
},
|
| 397 |
"attributes": {}
|
| 398 |
}
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.1,
|
| 6 |
"eval_steps": 20,
|
| 7 |
+
"global_step": 20,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.025,
|
| 14 |
+
"grad_norm": 0.15162423253059387,
|
| 15 |
"learning_rate": 1.6e-06,
|
| 16 |
+
"loss": 10.962069702148437,
|
| 17 |
"step": 5
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.05,
|
| 21 |
+
"grad_norm": 0.16617199778556824,
|
| 22 |
"learning_rate": 3.6e-06,
|
| 23 |
+
"loss": 10.898477935791016,
|
| 24 |
"step": 10
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.075,
|
| 28 |
+
"grad_norm": 0.17574748396873474,
|
| 29 |
"learning_rate": 3.995627254437549e-06,
|
| 30 |
+
"loss": 10.802017211914062,
|
| 31 |
"step": 15
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.1,
|
| 35 |
+
"grad_norm": 0.2199895679950714,
|
| 36 |
"learning_rate": 3.9778957412029366e-06,
|
| 37 |
+
"loss": 10.78037338256836,
|
| 38 |
"step": 20
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.1,
|
| 42 |
+
"eval_accuracy": 0.010476190476190476,
|
| 43 |
+
"eval_loss": 10.812881469726562,
|
| 44 |
+
"eval_runtime": 2.4156,
|
| 45 |
+
"eval_samples_per_second": 413.978,
|
| 46 |
+
"eval_steps_per_second": 69.134,
|
| 47 |
"step": 20
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
}
|
| 49 |
],
|
| 50 |
"logging_steps": 5,
|
|
|
|
| 59 |
"should_evaluate": false,
|
| 60 |
"should_log": false,
|
| 61 |
"should_save": true,
|
| 62 |
+
"should_training_stop": false
|
| 63 |
},
|
| 64 |
"attributes": {}
|
| 65 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5265
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b0b6058a46bce4b387df0ce8c61d4b0b1b316ccf92487321cd86eb8514ad352
|
| 3 |
size 5265
|