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:a10239be6e3cd4f5e89f2ac25a3b3c925a1d419acc365d80f558a325fd438b16
|
| 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:a4558c7ef3834ff59ddefc932ce8df0fd0dd2dc9c93b1481f3635fd4818e96c8
|
| 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:d17c78731e93c053cc5bdd46a999e26416061fc6f401c1eac6682002c6f7cf23
|
| 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":
|
| 43 |
-
"eval_loss":
|
| 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": 0.25184789299964905,
|
| 52 |
-
"learning_rate": 3.9466531944960116e-06,
|
| 53 |
-
"loss": 22.0181884765625,
|
| 54 |
-
"step": 25
|
| 55 |
-
},
|
| 56 |
-
{
|
| 57 |
-
"epoch": 0.15,
|
| 58 |
-
"grad_norm": 0.268789142370224,
|
| 59 |
-
"learning_rate": 3.902113032590307e-06,
|
| 60 |
-
"loss": 22.01494598388672,
|
| 61 |
-
"step": 30
|
| 62 |
-
},
|
| 63 |
-
{
|
| 64 |
-
"epoch": 0.175,
|
| 65 |
-
"grad_norm": 0.2609691917896271,
|
| 66 |
-
"learning_rate": 3.844579509954769e-06,
|
| 67 |
-
"loss": 22.0348876953125,
|
| 68 |
-
"step": 35
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.2,
|
| 72 |
-
"grad_norm": 0.2492043823003769,
|
| 73 |
-
"learning_rate": 3.7744456388872234e-06,
|
| 74 |
-
"loss": 21.99768524169922,
|
| 75 |
-
"step": 40
|
| 76 |
-
},
|
| 77 |
-
{
|
| 78 |
-
"epoch": 0.2,
|
| 79 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 80 |
-
"eval_loss": 11.018282890319824,
|
| 81 |
-
"eval_runtime": 2.6799,
|
| 82 |
-
"eval_samples_per_second": 373.142,
|
| 83 |
-
"eval_steps_per_second": 62.315,
|
| 84 |
-
"step": 40
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.225,
|
| 88 |
-
"grad_norm": 0.2999836802482605,
|
| 89 |
-
"learning_rate": 3.6921905048418706e-06,
|
| 90 |
-
"loss": 21.986573791503908,
|
| 91 |
-
"step": 45
|
| 92 |
-
},
|
| 93 |
-
{
|
| 94 |
-
"epoch": 0.25,
|
| 95 |
-
"grad_norm": 0.2840990126132965,
|
| 96 |
-
"learning_rate": 3.598375993789849e-06,
|
| 97 |
-
"loss": 22.012301635742187,
|
| 98 |
-
"step": 50
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.275,
|
| 102 |
-
"grad_norm": 0.28335806727409363,
|
| 103 |
-
"learning_rate": 3.4936429539683075e-06,
|
| 104 |
-
"loss": 22.00952606201172,
|
| 105 |
-
"step": 55
|
| 106 |
-
},
|
| 107 |
-
{
|
| 108 |
-
"epoch": 0.3,
|
| 109 |
-
"grad_norm": 0.2717016935348511,
|
| 110 |
-
"learning_rate": 3.3787068182371314e-06,
|
| 111 |
-
"loss": 21.99034423828125,
|
| 112 |
-
"step": 60
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 0.3,
|
| 116 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 117 |
-
"eval_loss": 11.01639175415039,
|
| 118 |
-
"eval_runtime": 2.4905,
|
| 119 |
-
"eval_samples_per_second": 401.523,
|
| 120 |
-
"eval_steps_per_second": 67.054,
|
| 121 |
-
"step": 60
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.325,
|
| 125 |
-
"grad_norm": 0.26950836181640625,
|
| 126 |
-
"learning_rate": 3.254352716947074e-06,
|
| 127 |
-
"loss": 22.037962341308592,
|
| 128 |
-
"step": 65
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.35,
|
| 132 |
-
"grad_norm": 0.25924476981163025,
|
| 133 |
-
"learning_rate": 3.1214301147033453e-06,
|
| 134 |
-
"loss": 22.014053344726562,
|
| 135 |
-
"step": 70
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.375,
|
| 139 |
-
"grad_norm": 0.31555065512657166,
|
| 140 |
-
"learning_rate": 2.9808470076610163e-06,
|
| 141 |
-
"loss": 21.961640930175783,
|
| 142 |
-
"step": 75
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.4,
|
| 146 |
-
"grad_norm": 0.2712528705596924,
|
| 147 |
-
"learning_rate": 2.833563720990581e-06,
|
| 148 |
-
"loss": 21.988558959960937,
|
| 149 |
-
"step": 80
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.4,
|
| 153 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 154 |
-
"eval_loss": 11.01443099975586,
|
| 155 |
-
"eval_runtime": 2.612,
|
| 156 |
-
"eval_samples_per_second": 382.842,
|
| 157 |
-
"eval_steps_per_second": 63.935,
|
| 158 |
-
"step": 80
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.425,
|
| 162 |
-
"grad_norm": 0.260957270860672,
|
| 163 |
-
"learning_rate": 2.680586348883286e-06,
|
| 164 |
-
"loss": 22.034954833984376,
|
| 165 |
-
"step": 85
|
| 166 |
-
},
|
| 167 |
-
{
|
| 168 |
-
"epoch": 0.45,
|
| 169 |
-
"grad_norm": 0.2659424841403961,
|
| 170 |
-
"learning_rate": 2.5229598819076393e-06,
|
| 171 |
-
"loss": 22.037962341308592,
|
| 172 |
-
"step": 90
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.475,
|
| 176 |
-
"grad_norm": 0.2600838541984558,
|
| 177 |
-
"learning_rate": 2.36176106866422e-06,
|
| 178 |
-
"loss": 21.996064758300783,
|
| 179 |
-
"step": 95
|
| 180 |
-
},
|
| 181 |
-
{
|
| 182 |
-
"epoch": 0.5,
|
| 183 |
-
"grad_norm": 0.2419586181640625,
|
| 184 |
-
"learning_rate": 2.198091060500926e-06,
|
| 185 |
-
"loss": 22.00383605957031,
|
| 186 |
-
"step": 100
|
| 187 |
-
},
|
| 188 |
-
{
|
| 189 |
-
"epoch": 0.5,
|
| 190 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 191 |
-
"eval_loss": 11.012611389160156,
|
| 192 |
-
"eval_runtime": 2.6096,
|
| 193 |
-
"eval_samples_per_second": 383.201,
|
| 194 |
-
"eval_steps_per_second": 63.995,
|
| 195 |
-
"step": 100
|
| 196 |
-
},
|
| 197 |
-
{
|
| 198 |
-
"epoch": 0.525,
|
| 199 |
-
"grad_norm": 0.24510444700717926,
|
| 200 |
-
"learning_rate": 2.033067889532717e-06,
|
| 201 |
-
"loss": 21.980421447753905,
|
| 202 |
-
"step": 105
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"epoch": 0.55,
|
| 206 |
-
"grad_norm": 0.24764317274093628,
|
| 207 |
-
"learning_rate": 1.8678188313486012e-06,
|
| 208 |
-
"loss": 21.973478698730467,
|
| 209 |
-
"step": 110
|
| 210 |
-
},
|
| 211 |
-
{
|
| 212 |
-
"epoch": 0.575,
|
| 213 |
-
"grad_norm": 0.28122493624687195,
|
| 214 |
-
"learning_rate": 1.7034727045763157e-06,
|
| 215 |
-
"loss": 21.949703979492188,
|
| 216 |
-
"step": 115
|
| 217 |
-
},
|
| 218 |
-
{
|
| 219 |
-
"epoch": 0.6,
|
| 220 |
-
"grad_norm": 0.250843346118927,
|
| 221 |
-
"learning_rate": 1.541152159906497e-06,
|
| 222 |
-
"loss": 21.94451141357422,
|
| 223 |
-
"step": 120
|
| 224 |
-
},
|
| 225 |
-
{
|
| 226 |
-
"epoch": 0.6,
|
| 227 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 228 |
-
"eval_loss": 11.010734558105469,
|
| 229 |
-
"eval_runtime": 2.7011,
|
| 230 |
-
"eval_samples_per_second": 370.215,
|
| 231 |
-
"eval_steps_per_second": 61.826,
|
| 232 |
-
"step": 120
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.625,
|
| 236 |
-
"grad_norm": 0.24475905299186707,
|
| 237 |
-
"learning_rate": 1.3819660112501052e-06,
|
| 238 |
-
"loss": 22.0078369140625,
|
| 239 |
-
"step": 125
|
| 240 |
-
},
|
| 241 |
-
{
|
| 242 |
-
"epoch": 0.65,
|
| 243 |
-
"grad_norm": 0.2379191815853119,
|
| 244 |
-
"learning_rate": 1.227001661415096e-06,
|
| 245 |
-
"loss": 22.009359741210936,
|
| 246 |
-
"step": 130
|
| 247 |
-
},
|
| 248 |
-
{
|
| 249 |
-
"epoch": 0.675,
|
| 250 |
-
"grad_norm": 0.26111480593681335,
|
| 251 |
-
"learning_rate": 1.0773176740426246e-06,
|
| 252 |
-
"loss": 21.976290893554687,
|
| 253 |
-
"step": 135
|
| 254 |
-
},
|
| 255 |
-
{
|
| 256 |
-
"epoch": 0.7,
|
| 257 |
-
"grad_norm": 0.28270959854125977,
|
| 258 |
-
"learning_rate": 9.339365425440129e-07,
|
| 259 |
-
"loss": 22.019345092773438,
|
| 260 |
-
"step": 140
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.7,
|
| 264 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 265 |
-
"eval_loss": 11.009251594543457,
|
| 266 |
-
"eval_runtime": 2.5123,
|
| 267 |
-
"eval_samples_per_second": 398.045,
|
| 268 |
-
"eval_steps_per_second": 66.473,
|
| 269 |
-
"step": 140
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.725,
|
| 273 |
-
"grad_norm": 0.2701219916343689,
|
| 274 |
-
"learning_rate": 7.978377054339498e-07,
|
| 275 |
-
"loss": 21.985748291015625,
|
| 276 |
-
"step": 145
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.75,
|
| 280 |
-
"grad_norm": 0.24651674926280975,
|
| 281 |
-
"learning_rate": 6.699508557723032e-07,
|
| 282 |
-
"loss": 22.00152130126953,
|
| 283 |
-
"step": 150
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.775,
|
| 287 |
-
"grad_norm": 0.26024770736694336,
|
| 288 |
-
"learning_rate": 5.511495904178221e-07,
|
| 289 |
-
"loss": 22.02308044433594,
|
| 290 |
-
"step": 155
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.8,
|
| 294 |
-
"grad_norm": 0.25778868794441223,
|
| 295 |
-
"learning_rate": 4.4224544247577535e-07,
|
| 296 |
-
"loss": 21.997288513183594,
|
| 297 |
-
"step": 160
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.8,
|
| 301 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 302 |
-
"eval_loss": 11.007829666137695,
|
| 303 |
-
"eval_runtime": 2.7386,
|
| 304 |
-
"eval_samples_per_second": 365.156,
|
| 305 |
-
"eval_steps_per_second": 60.981,
|
| 306 |
-
"step": 160
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 0.825,
|
| 310 |
-
"grad_norm": 0.2776707112789154,
|
| 311 |
-
"learning_rate": 3.439823377039599e-07,
|
| 312 |
-
"loss": 22.04285583496094,
|
| 313 |
-
"step": 165
|
| 314 |
-
},
|
| 315 |
-
{
|
| 316 |
-
"epoch": 0.85,
|
| 317 |
-
"grad_norm": 0.2490997463464737,
|
| 318 |
-
"learning_rate": 2.570315127454452e-07,
|
| 319 |
-
"loss": 21.98439178466797,
|
| 320 |
-
"step": 170
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.875,
|
| 324 |
-
"grad_norm": 0.2551729679107666,
|
| 325 |
-
"learning_rate": 1.8198692990167497e-07,
|
| 326 |
-
"loss": 21.973941040039062,
|
| 327 |
-
"step": 175
|
| 328 |
-
},
|
| 329 |
-
{
|
| 330 |
-
"epoch": 0.9,
|
| 331 |
-
"grad_norm": 0.32327187061309814,
|
| 332 |
-
"learning_rate": 1.1936121976767767e-07,
|
| 333 |
-
"loss": 21.965573120117188,
|
| 334 |
-
"step": 180
|
| 335 |
-
},
|
| 336 |
-
{
|
| 337 |
-
"epoch": 0.9,
|
| 338 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 339 |
-
"eval_loss": 11.006654739379883,
|
| 340 |
-
"eval_runtime": 2.4962,
|
| 341 |
-
"eval_samples_per_second": 400.614,
|
| 342 |
-
"eval_steps_per_second": 66.903,
|
| 343 |
-
"step": 180
|
| 344 |
-
},
|
| 345 |
-
{
|
| 346 |
-
"epoch": 0.925,
|
| 347 |
-
"grad_norm": 0.24408257007598877,
|
| 348 |
-
"learning_rate": 6.958217944530287e-08,
|
| 349 |
-
"loss": 21.973313903808595,
|
| 350 |
-
"step": 185
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.95,
|
| 354 |
-
"grad_norm": 0.2715650796890259,
|
| 355 |
-
"learning_rate": 3.2989850255235265e-08,
|
| 356 |
-
"loss": 21.963623046875,
|
| 357 |
-
"step": 190
|
| 358 |
-
},
|
| 359 |
-
{
|
| 360 |
-
"epoch": 0.975,
|
| 361 |
-
"grad_norm": 0.2907275855541229,
|
| 362 |
-
"learning_rate": 9.834194909977168e-09,
|
| 363 |
-
"loss": 21.96564178466797,
|
| 364 |
-
"step": 195
|
| 365 |
-
},
|
| 366 |
-
{
|
| 367 |
-
"epoch": 1.0,
|
| 368 |
-
"grad_norm": 0.29036688804626465,
|
| 369 |
-
"learning_rate": 2.7339001506199164e-10,
|
| 370 |
-
"loss": 21.94970245361328,
|
| 371 |
-
"step": 200
|
| 372 |
-
},
|
| 373 |
-
{
|
| 374 |
-
"epoch": 1.0,
|
| 375 |
-
"eval_accuracy": 3.1746031746031745e-05,
|
| 376 |
-
"eval_loss": 11.00558090209961,
|
| 377 |
-
"eval_runtime": 2.4684,
|
| 378 |
-
"eval_samples_per_second": 405.122,
|
| 379 |
-
"eval_steps_per_second": 67.655,
|
| 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": 0.2450859099626541,
|
| 15 |
"learning_rate": 1.6e-06,
|
| 16 |
+
"loss": 21.895669555664064,
|
| 17 |
"step": 5
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.05,
|
| 21 |
+
"grad_norm": 0.2795044779777527,
|
| 22 |
"learning_rate": 3.6e-06,
|
| 23 |
+
"loss": 21.775099182128905,
|
| 24 |
"step": 10
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.075,
|
| 28 |
+
"grad_norm": 0.25545090436935425,
|
| 29 |
"learning_rate": 3.995627254437549e-06,
|
| 30 |
+
"loss": 21.680242919921874,
|
| 31 |
"step": 15
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.1,
|
| 35 |
+
"grad_norm": 0.3522493839263916,
|
| 36 |
"learning_rate": 3.9778957412029366e-06,
|
| 37 |
+
"loss": 21.56825408935547,
|
| 38 |
"step": 20
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.1,
|
| 42 |
+
"eval_accuracy": 0.012365079365079365,
|
| 43 |
+
"eval_loss": 10.798470497131348,
|
| 44 |
+
"eval_runtime": 2.4457,
|
| 45 |
+
"eval_samples_per_second": 408.879,
|
| 46 |
+
"eval_steps_per_second": 68.283,
|
| 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 |
}
|