Training in progress, epoch 1, 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 2708740080
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b59a3c142cb36e782f862ff97df3be4b268f2b82af56ea2048c0b22de8ab70ac
|
| 3 |
size 2708740080
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 52521338
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:15964537a2492ff389ce57f04434fce8c9aa401544dbb2ca53af890dab8c7424
|
| 3 |
size 52521338
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15006
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9080aea5181d3066ab765d04bc9819f089e9674161d5e56c8bf2b7c839212160
|
| 3 |
size 15006
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a3706e3071c2c4cbd5f2e989bf126df210bdb1a4e75a8894205331c65155ec1
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,780 +1,28 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "mgh6/HTH_biCLIP_mean/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 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": 1.0,
|
| 13 |
-
"grad_norm": 0.
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss": 2.
|
| 16 |
"step": 129
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 1.0,
|
| 20 |
-
"eval_cosine_similarity": 0.
|
| 21 |
-
"eval_loss": 2.
|
| 22 |
-
"eval_runtime": 18.
|
| 23 |
-
"eval_samples_per_second": 22.
|
| 24 |
-
"eval_steps_per_second": 1.
|
| 25 |
"step": 129
|
| 26 |
-
},
|
| 27 |
-
{
|
| 28 |
-
"epoch": 2.0,
|
| 29 |
-
"grad_norm": 0.03998406603932381,
|
| 30 |
-
"learning_rate": 9.6640625e-05,
|
| 31 |
-
"loss": 2.7174,
|
| 32 |
-
"step": 258
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 2.0,
|
| 36 |
-
"eval_cosine_similarity": 0.35804858803749084,
|
| 37 |
-
"eval_loss": 2.7347757816314697,
|
| 38 |
-
"eval_runtime": 18.8518,
|
| 39 |
-
"eval_samples_per_second": 22.916,
|
| 40 |
-
"eval_steps_per_second": 1.432,
|
| 41 |
-
"step": 258
|
| 42 |
-
},
|
| 43 |
-
{
|
| 44 |
-
"epoch": 3.0,
|
| 45 |
-
"grad_norm": 0.09510844200849533,
|
| 46 |
-
"learning_rate": 9.496093750000001e-05,
|
| 47 |
-
"loss": 2.712,
|
| 48 |
-
"step": 387
|
| 49 |
-
},
|
| 50 |
-
{
|
| 51 |
-
"epoch": 3.0,
|
| 52 |
-
"eval_cosine_similarity": 0.31994006037712097,
|
| 53 |
-
"eval_loss": 2.7384214401245117,
|
| 54 |
-
"eval_runtime": 18.8684,
|
| 55 |
-
"eval_samples_per_second": 22.895,
|
| 56 |
-
"eval_steps_per_second": 1.431,
|
| 57 |
-
"step": 387
|
| 58 |
-
},
|
| 59 |
-
{
|
| 60 |
-
"epoch": 4.0,
|
| 61 |
-
"grad_norm": 0.09453984349966049,
|
| 62 |
-
"learning_rate": 9.328125000000001e-05,
|
| 63 |
-
"loss": 2.7083,
|
| 64 |
-
"step": 516
|
| 65 |
-
},
|
| 66 |
-
{
|
| 67 |
-
"epoch": 4.0,
|
| 68 |
-
"eval_cosine_similarity": 0.396742582321167,
|
| 69 |
-
"eval_loss": 2.7257983684539795,
|
| 70 |
-
"eval_runtime": 18.8397,
|
| 71 |
-
"eval_samples_per_second": 22.93,
|
| 72 |
-
"eval_steps_per_second": 1.433,
|
| 73 |
-
"step": 516
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 5.0,
|
| 77 |
-
"grad_norm": 0.1036021038889885,
|
| 78 |
-
"learning_rate": 9.160156250000001e-05,
|
| 79 |
-
"loss": 2.7041,
|
| 80 |
-
"step": 645
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 5.0,
|
| 84 |
-
"eval_cosine_similarity": 0.4167896807193756,
|
| 85 |
-
"eval_loss": 2.7295186519622803,
|
| 86 |
-
"eval_runtime": 18.8549,
|
| 87 |
-
"eval_samples_per_second": 22.912,
|
| 88 |
-
"eval_steps_per_second": 1.432,
|
| 89 |
-
"step": 645
|
| 90 |
-
},
|
| 91 |
-
{
|
| 92 |
-
"epoch": 6.0,
|
| 93 |
-
"grad_norm": 0.10563068836927414,
|
| 94 |
-
"learning_rate": 8.9921875e-05,
|
| 95 |
-
"loss": 2.701,
|
| 96 |
-
"step": 774
|
| 97 |
-
},
|
| 98 |
-
{
|
| 99 |
-
"epoch": 6.0,
|
| 100 |
-
"eval_cosine_similarity": 0.4463934600353241,
|
| 101 |
-
"eval_loss": 2.722810983657837,
|
| 102 |
-
"eval_runtime": 18.8344,
|
| 103 |
-
"eval_samples_per_second": 22.937,
|
| 104 |
-
"eval_steps_per_second": 1.434,
|
| 105 |
-
"step": 774
|
| 106 |
-
},
|
| 107 |
-
{
|
| 108 |
-
"epoch": 7.0,
|
| 109 |
-
"grad_norm": 0.3786303400993347,
|
| 110 |
-
"learning_rate": 8.824218750000001e-05,
|
| 111 |
-
"loss": 2.6981,
|
| 112 |
-
"step": 903
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 7.0,
|
| 116 |
-
"eval_cosine_similarity": 0.3459477722644806,
|
| 117 |
-
"eval_loss": 2.7271432876586914,
|
| 118 |
-
"eval_runtime": 18.854,
|
| 119 |
-
"eval_samples_per_second": 22.913,
|
| 120 |
-
"eval_steps_per_second": 1.432,
|
| 121 |
-
"step": 903
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 8.0,
|
| 125 |
-
"grad_norm": 0.39843541383743286,
|
| 126 |
-
"learning_rate": 8.65625e-05,
|
| 127 |
-
"loss": 2.695,
|
| 128 |
-
"step": 1032
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 8.0,
|
| 132 |
-
"eval_cosine_similarity": 0.44728702306747437,
|
| 133 |
-
"eval_loss": 2.717395544052124,
|
| 134 |
-
"eval_runtime": 18.8516,
|
| 135 |
-
"eval_samples_per_second": 22.916,
|
| 136 |
-
"eval_steps_per_second": 1.432,
|
| 137 |
-
"step": 1032
|
| 138 |
-
},
|
| 139 |
-
{
|
| 140 |
-
"epoch": 9.0,
|
| 141 |
-
"grad_norm": 0.061124324798583984,
|
| 142 |
-
"learning_rate": 8.488281250000001e-05,
|
| 143 |
-
"loss": 2.6921,
|
| 144 |
-
"step": 1161
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 9.0,
|
| 148 |
-
"eval_cosine_similarity": 0.4983077645301819,
|
| 149 |
-
"eval_loss": 2.7137577533721924,
|
| 150 |
-
"eval_runtime": 18.8363,
|
| 151 |
-
"eval_samples_per_second": 22.934,
|
| 152 |
-
"eval_steps_per_second": 1.433,
|
| 153 |
-
"step": 1161
|
| 154 |
-
},
|
| 155 |
-
{
|
| 156 |
-
"epoch": 10.0,
|
| 157 |
-
"grad_norm": 0.12580373883247375,
|
| 158 |
-
"learning_rate": 8.3203125e-05,
|
| 159 |
-
"loss": 2.6898,
|
| 160 |
-
"step": 1290
|
| 161 |
-
},
|
| 162 |
-
{
|
| 163 |
-
"epoch": 10.0,
|
| 164 |
-
"eval_cosine_similarity": 0.5014428496360779,
|
| 165 |
-
"eval_loss": 2.7119338512420654,
|
| 166 |
-
"eval_runtime": 18.8721,
|
| 167 |
-
"eval_samples_per_second": 22.891,
|
| 168 |
-
"eval_steps_per_second": 1.431,
|
| 169 |
-
"step": 1290
|
| 170 |
-
},
|
| 171 |
-
{
|
| 172 |
-
"epoch": 11.0,
|
| 173 |
-
"grad_norm": 0.12061749398708344,
|
| 174 |
-
"learning_rate": 8.15234375e-05,
|
| 175 |
-
"loss": 2.6859,
|
| 176 |
-
"step": 1419
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 11.0,
|
| 180 |
-
"eval_cosine_similarity": 0.48711416125297546,
|
| 181 |
-
"eval_loss": 2.714085578918457,
|
| 182 |
-
"eval_runtime": 18.834,
|
| 183 |
-
"eval_samples_per_second": 22.937,
|
| 184 |
-
"eval_steps_per_second": 1.434,
|
| 185 |
-
"step": 1419
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 12.0,
|
| 189 |
-
"grad_norm": 0.18075203895568848,
|
| 190 |
-
"learning_rate": 7.984375e-05,
|
| 191 |
-
"loss": 2.6843,
|
| 192 |
-
"step": 1548
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 12.0,
|
| 196 |
-
"eval_cosine_similarity": 0.5338874459266663,
|
| 197 |
-
"eval_loss": 2.707056760787964,
|
| 198 |
-
"eval_runtime": 18.8491,
|
| 199 |
-
"eval_samples_per_second": 22.919,
|
| 200 |
-
"eval_steps_per_second": 1.432,
|
| 201 |
-
"step": 1548
|
| 202 |
-
},
|
| 203 |
-
{
|
| 204 |
-
"epoch": 13.0,
|
| 205 |
-
"grad_norm": 0.22249284386634827,
|
| 206 |
-
"learning_rate": 7.81640625e-05,
|
| 207 |
-
"loss": 2.6829,
|
| 208 |
-
"step": 1677
|
| 209 |
-
},
|
| 210 |
-
{
|
| 211 |
-
"epoch": 13.0,
|
| 212 |
-
"eval_cosine_similarity": 0.5601667761802673,
|
| 213 |
-
"eval_loss": 2.705392837524414,
|
| 214 |
-
"eval_runtime": 18.8352,
|
| 215 |
-
"eval_samples_per_second": 22.936,
|
| 216 |
-
"eval_steps_per_second": 1.433,
|
| 217 |
-
"step": 1677
|
| 218 |
-
},
|
| 219 |
-
{
|
| 220 |
-
"epoch": 14.0,
|
| 221 |
-
"grad_norm": 0.16165713965892792,
|
| 222 |
-
"learning_rate": 7.648437500000001e-05,
|
| 223 |
-
"loss": 2.6799,
|
| 224 |
-
"step": 1806
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 14.0,
|
| 228 |
-
"eval_cosine_similarity": 0.5292511582374573,
|
| 229 |
-
"eval_loss": 2.7065181732177734,
|
| 230 |
-
"eval_runtime": 18.8446,
|
| 231 |
-
"eval_samples_per_second": 22.924,
|
| 232 |
-
"eval_steps_per_second": 1.433,
|
| 233 |
-
"step": 1806
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 15.0,
|
| 237 |
-
"grad_norm": 0.1596257984638214,
|
| 238 |
-
"learning_rate": 7.48046875e-05,
|
| 239 |
-
"loss": 2.678,
|
| 240 |
-
"step": 1935
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 15.0,
|
| 244 |
-
"eval_cosine_similarity": 0.5406491756439209,
|
| 245 |
-
"eval_loss": 2.7067458629608154,
|
| 246 |
-
"eval_runtime": 18.8328,
|
| 247 |
-
"eval_samples_per_second": 22.939,
|
| 248 |
-
"eval_steps_per_second": 1.434,
|
| 249 |
-
"step": 1935
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 16.0,
|
| 253 |
-
"grad_norm": 0.17212730646133423,
|
| 254 |
-
"learning_rate": 7.3125e-05,
|
| 255 |
-
"loss": 2.6762,
|
| 256 |
-
"step": 2064
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 16.0,
|
| 260 |
-
"eval_cosine_similarity": 0.5383115410804749,
|
| 261 |
-
"eval_loss": 2.7063181400299072,
|
| 262 |
-
"eval_runtime": 18.8441,
|
| 263 |
-
"eval_samples_per_second": 22.925,
|
| 264 |
-
"eval_steps_per_second": 1.433,
|
| 265 |
-
"step": 2064
|
| 266 |
-
},
|
| 267 |
-
{
|
| 268 |
-
"epoch": 17.0,
|
| 269 |
-
"grad_norm": 0.16676370799541473,
|
| 270 |
-
"learning_rate": 7.14453125e-05,
|
| 271 |
-
"loss": 2.6748,
|
| 272 |
-
"step": 2193
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 17.0,
|
| 276 |
-
"eval_cosine_similarity": 0.5793458223342896,
|
| 277 |
-
"eval_loss": 2.7020437717437744,
|
| 278 |
-
"eval_runtime": 18.8795,
|
| 279 |
-
"eval_samples_per_second": 22.882,
|
| 280 |
-
"eval_steps_per_second": 1.43,
|
| 281 |
-
"step": 2193
|
| 282 |
-
},
|
| 283 |
-
{
|
| 284 |
-
"epoch": 18.0,
|
| 285 |
-
"grad_norm": 0.08970298618078232,
|
| 286 |
-
"learning_rate": 6.9765625e-05,
|
| 287 |
-
"loss": 2.6722,
|
| 288 |
-
"step": 2322
|
| 289 |
-
},
|
| 290 |
-
{
|
| 291 |
-
"epoch": 18.0,
|
| 292 |
-
"eval_cosine_similarity": 0.608212411403656,
|
| 293 |
-
"eval_loss": 2.6979143619537354,
|
| 294 |
-
"eval_runtime": 18.8361,
|
| 295 |
-
"eval_samples_per_second": 22.935,
|
| 296 |
-
"eval_steps_per_second": 1.433,
|
| 297 |
-
"step": 2322
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 19.0,
|
| 301 |
-
"grad_norm": 0.46823370456695557,
|
| 302 |
-
"learning_rate": 6.80859375e-05,
|
| 303 |
-
"loss": 2.6707,
|
| 304 |
-
"step": 2451
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 19.0,
|
| 308 |
-
"eval_cosine_similarity": 0.5827838182449341,
|
| 309 |
-
"eval_loss": 2.699484348297119,
|
| 310 |
-
"eval_runtime": 18.8456,
|
| 311 |
-
"eval_samples_per_second": 22.923,
|
| 312 |
-
"eval_steps_per_second": 1.433,
|
| 313 |
-
"step": 2451
|
| 314 |
-
},
|
| 315 |
-
{
|
| 316 |
-
"epoch": 20.0,
|
| 317 |
-
"grad_norm": 0.22482483088970184,
|
| 318 |
-
"learning_rate": 6.640625e-05,
|
| 319 |
-
"loss": 2.6687,
|
| 320 |
-
"step": 2580
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 20.0,
|
| 324 |
-
"eval_cosine_similarity": 0.6316117644309998,
|
| 325 |
-
"eval_loss": 2.6942453384399414,
|
| 326 |
-
"eval_runtime": 18.8317,
|
| 327 |
-
"eval_samples_per_second": 22.94,
|
| 328 |
-
"eval_steps_per_second": 1.434,
|
| 329 |
-
"step": 2580
|
| 330 |
-
},
|
| 331 |
-
{
|
| 332 |
-
"epoch": 21.0,
|
| 333 |
-
"grad_norm": 0.45786142349243164,
|
| 334 |
-
"learning_rate": 6.472656249999999e-05,
|
| 335 |
-
"loss": 2.6661,
|
| 336 |
-
"step": 2709
|
| 337 |
-
},
|
| 338 |
-
{
|
| 339 |
-
"epoch": 21.0,
|
| 340 |
-
"eval_cosine_similarity": 0.6281163692474365,
|
| 341 |
-
"eval_loss": 2.6956567764282227,
|
| 342 |
-
"eval_runtime": 18.8424,
|
| 343 |
-
"eval_samples_per_second": 22.927,
|
| 344 |
-
"eval_steps_per_second": 1.433,
|
| 345 |
-
"step": 2709
|
| 346 |
-
},
|
| 347 |
-
{
|
| 348 |
-
"epoch": 22.0,
|
| 349 |
-
"grad_norm": 0.5357567667961121,
|
| 350 |
-
"learning_rate": 6.3046875e-05,
|
| 351 |
-
"loss": 2.6654,
|
| 352 |
-
"step": 2838
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"epoch": 22.0,
|
| 356 |
-
"eval_cosine_similarity": 0.639909029006958,
|
| 357 |
-
"eval_loss": 2.6937482357025146,
|
| 358 |
-
"eval_runtime": 18.8423,
|
| 359 |
-
"eval_samples_per_second": 22.927,
|
| 360 |
-
"eval_steps_per_second": 1.433,
|
| 361 |
-
"step": 2838
|
| 362 |
-
},
|
| 363 |
-
{
|
| 364 |
-
"epoch": 23.0,
|
| 365 |
-
"grad_norm": 0.5274462103843689,
|
| 366 |
-
"learning_rate": 6.136718750000001e-05,
|
| 367 |
-
"loss": 2.6634,
|
| 368 |
-
"step": 2967
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 23.0,
|
| 372 |
-
"eval_cosine_similarity": 0.6498256325721741,
|
| 373 |
-
"eval_loss": 2.691450834274292,
|
| 374 |
-
"eval_runtime": 18.8402,
|
| 375 |
-
"eval_samples_per_second": 22.93,
|
| 376 |
-
"eval_steps_per_second": 1.433,
|
| 377 |
-
"step": 2967
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 24.0,
|
| 381 |
-
"grad_norm": 0.7012160420417786,
|
| 382 |
-
"learning_rate": 5.968750000000001e-05,
|
| 383 |
-
"loss": 2.6617,
|
| 384 |
-
"step": 3096
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 24.0,
|
| 388 |
-
"eval_cosine_similarity": 0.6656497120857239,
|
| 389 |
-
"eval_loss": 2.691051959991455,
|
| 390 |
-
"eval_runtime": 18.8635,
|
| 391 |
-
"eval_samples_per_second": 22.901,
|
| 392 |
-
"eval_steps_per_second": 1.431,
|
| 393 |
-
"step": 3096
|
| 394 |
-
},
|
| 395 |
-
{
|
| 396 |
-
"epoch": 25.0,
|
| 397 |
-
"grad_norm": 0.07700519263744354,
|
| 398 |
-
"learning_rate": 5.80078125e-05,
|
| 399 |
-
"loss": 2.6605,
|
| 400 |
-
"step": 3225
|
| 401 |
-
},
|
| 402 |
-
{
|
| 403 |
-
"epoch": 25.0,
|
| 404 |
-
"eval_cosine_similarity": 0.6613298058509827,
|
| 405 |
-
"eval_loss": 2.689277410507202,
|
| 406 |
-
"eval_runtime": 18.8365,
|
| 407 |
-
"eval_samples_per_second": 22.934,
|
| 408 |
-
"eval_steps_per_second": 1.433,
|
| 409 |
-
"step": 3225
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 26.0,
|
| 413 |
-
"grad_norm": 0.15677490830421448,
|
| 414 |
-
"learning_rate": 5.6328124999999995e-05,
|
| 415 |
-
"loss": 2.6599,
|
| 416 |
-
"step": 3354
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 26.0,
|
| 420 |
-
"eval_cosine_similarity": 0.6886489391326904,
|
| 421 |
-
"eval_loss": 2.686530590057373,
|
| 422 |
-
"eval_runtime": 18.8402,
|
| 423 |
-
"eval_samples_per_second": 22.93,
|
| 424 |
-
"eval_steps_per_second": 1.433,
|
| 425 |
-
"step": 3354
|
| 426 |
-
},
|
| 427 |
-
{
|
| 428 |
-
"epoch": 27.0,
|
| 429 |
-
"grad_norm": 0.3259325623512268,
|
| 430 |
-
"learning_rate": 5.46484375e-05,
|
| 431 |
-
"loss": 2.6573,
|
| 432 |
-
"step": 3483
|
| 433 |
-
},
|
| 434 |
-
{
|
| 435 |
-
"epoch": 27.0,
|
| 436 |
-
"eval_cosine_similarity": 0.6656365394592285,
|
| 437 |
-
"eval_loss": 2.689530849456787,
|
| 438 |
-
"eval_runtime": 18.8427,
|
| 439 |
-
"eval_samples_per_second": 22.927,
|
| 440 |
-
"eval_steps_per_second": 1.433,
|
| 441 |
-
"step": 3483
|
| 442 |
-
},
|
| 443 |
-
{
|
| 444 |
-
"epoch": 28.0,
|
| 445 |
-
"grad_norm": 0.29011014103889465,
|
| 446 |
-
"learning_rate": 5.2968750000000003e-05,
|
| 447 |
-
"loss": 2.6562,
|
| 448 |
-
"step": 3612
|
| 449 |
-
},
|
| 450 |
-
{
|
| 451 |
-
"epoch": 28.0,
|
| 452 |
-
"eval_cosine_similarity": 0.6862850189208984,
|
| 453 |
-
"eval_loss": 2.687514305114746,
|
| 454 |
-
"eval_runtime": 18.8416,
|
| 455 |
-
"eval_samples_per_second": 22.928,
|
| 456 |
-
"eval_steps_per_second": 1.433,
|
| 457 |
-
"step": 3612
|
| 458 |
-
},
|
| 459 |
-
{
|
| 460 |
-
"epoch": 29.0,
|
| 461 |
-
"grad_norm": 0.5012183785438538,
|
| 462 |
-
"learning_rate": 5.1289062500000004e-05,
|
| 463 |
-
"loss": 2.6549,
|
| 464 |
-
"step": 3741
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 29.0,
|
| 468 |
-
"eval_cosine_similarity": 0.6950795650482178,
|
| 469 |
-
"eval_loss": 2.6848597526550293,
|
| 470 |
-
"eval_runtime": 18.871,
|
| 471 |
-
"eval_samples_per_second": 22.892,
|
| 472 |
-
"eval_steps_per_second": 1.431,
|
| 473 |
-
"step": 3741
|
| 474 |
-
},
|
| 475 |
-
{
|
| 476 |
-
"epoch": 30.0,
|
| 477 |
-
"grad_norm": 0.2342006117105484,
|
| 478 |
-
"learning_rate": 4.9609375000000005e-05,
|
| 479 |
-
"loss": 2.6543,
|
| 480 |
-
"step": 3870
|
| 481 |
-
},
|
| 482 |
-
{
|
| 483 |
-
"epoch": 30.0,
|
| 484 |
-
"eval_cosine_similarity": 0.6870617270469666,
|
| 485 |
-
"eval_loss": 2.6858551502227783,
|
| 486 |
-
"eval_runtime": 18.8336,
|
| 487 |
-
"eval_samples_per_second": 22.938,
|
| 488 |
-
"eval_steps_per_second": 1.434,
|
| 489 |
-
"step": 3870
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 31.0,
|
| 493 |
-
"grad_norm": 0.3240242600440979,
|
| 494 |
-
"learning_rate": 4.7929687500000005e-05,
|
| 495 |
-
"loss": 2.653,
|
| 496 |
-
"step": 3999
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 31.0,
|
| 500 |
-
"eval_cosine_similarity": 0.6926249861717224,
|
| 501 |
-
"eval_loss": 2.6856954097747803,
|
| 502 |
-
"eval_runtime": 18.8636,
|
| 503 |
-
"eval_samples_per_second": 22.901,
|
| 504 |
-
"eval_steps_per_second": 1.431,
|
| 505 |
-
"step": 3999
|
| 506 |
-
},
|
| 507 |
-
{
|
| 508 |
-
"epoch": 32.0,
|
| 509 |
-
"grad_norm": 1.058250069618225,
|
| 510 |
-
"learning_rate": 4.6250000000000006e-05,
|
| 511 |
-
"loss": 2.6513,
|
| 512 |
-
"step": 4128
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 32.0,
|
| 516 |
-
"eval_cosine_similarity": 0.7236000895500183,
|
| 517 |
-
"eval_loss": 2.682251214981079,
|
| 518 |
-
"eval_runtime": 18.8353,
|
| 519 |
-
"eval_samples_per_second": 22.936,
|
| 520 |
-
"eval_steps_per_second": 1.433,
|
| 521 |
-
"step": 4128
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 33.0,
|
| 525 |
-
"grad_norm": 0.1793142408132553,
|
| 526 |
-
"learning_rate": 4.45703125e-05,
|
| 527 |
-
"loss": 2.6501,
|
| 528 |
-
"step": 4257
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 33.0,
|
| 532 |
-
"eval_cosine_similarity": 0.6964766979217529,
|
| 533 |
-
"eval_loss": 2.6835083961486816,
|
| 534 |
-
"eval_runtime": 18.8683,
|
| 535 |
-
"eval_samples_per_second": 22.896,
|
| 536 |
-
"eval_steps_per_second": 1.431,
|
| 537 |
-
"step": 4257
|
| 538 |
-
},
|
| 539 |
-
{
|
| 540 |
-
"epoch": 34.0,
|
| 541 |
-
"grad_norm": 0.12666787207126617,
|
| 542 |
-
"learning_rate": 4.2890625e-05,
|
| 543 |
-
"loss": 2.6492,
|
| 544 |
-
"step": 4386
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 34.0,
|
| 548 |
-
"eval_cosine_similarity": 0.7385993003845215,
|
| 549 |
-
"eval_loss": 2.6802611351013184,
|
| 550 |
-
"eval_runtime": 18.8352,
|
| 551 |
-
"eval_samples_per_second": 22.936,
|
| 552 |
-
"eval_steps_per_second": 1.433,
|
| 553 |
-
"step": 4386
|
| 554 |
-
},
|
| 555 |
-
{
|
| 556 |
-
"epoch": 35.0,
|
| 557 |
-
"grad_norm": 0.28480854630470276,
|
| 558 |
-
"learning_rate": 4.12109375e-05,
|
| 559 |
-
"loss": 2.6477,
|
| 560 |
-
"step": 4515
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 35.0,
|
| 564 |
-
"eval_cosine_similarity": 0.7121383547782898,
|
| 565 |
-
"eval_loss": 2.6818628311157227,
|
| 566 |
-
"eval_runtime": 18.8567,
|
| 567 |
-
"eval_samples_per_second": 22.91,
|
| 568 |
-
"eval_steps_per_second": 1.432,
|
| 569 |
-
"step": 4515
|
| 570 |
-
},
|
| 571 |
-
{
|
| 572 |
-
"epoch": 36.0,
|
| 573 |
-
"grad_norm": 0.13797371089458466,
|
| 574 |
-
"learning_rate": 3.953125e-05,
|
| 575 |
-
"loss": 2.6475,
|
| 576 |
-
"step": 4644
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"epoch": 36.0,
|
| 580 |
-
"eval_cosine_similarity": 0.7440741658210754,
|
| 581 |
-
"eval_loss": 2.6793031692504883,
|
| 582 |
-
"eval_runtime": 18.8323,
|
| 583 |
-
"eval_samples_per_second": 22.939,
|
| 584 |
-
"eval_steps_per_second": 1.434,
|
| 585 |
-
"step": 4644
|
| 586 |
-
},
|
| 587 |
-
{
|
| 588 |
-
"epoch": 37.0,
|
| 589 |
-
"grad_norm": 0.3601125180721283,
|
| 590 |
-
"learning_rate": 3.78515625e-05,
|
| 591 |
-
"loss": 2.6472,
|
| 592 |
-
"step": 4773
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 37.0,
|
| 596 |
-
"eval_cosine_similarity": 0.7328048348426819,
|
| 597 |
-
"eval_loss": 2.682004690170288,
|
| 598 |
-
"eval_runtime": 18.8457,
|
| 599 |
-
"eval_samples_per_second": 22.923,
|
| 600 |
-
"eval_steps_per_second": 1.433,
|
| 601 |
-
"step": 4773
|
| 602 |
-
},
|
| 603 |
-
{
|
| 604 |
-
"epoch": 38.0,
|
| 605 |
-
"grad_norm": 0.49930471181869507,
|
| 606 |
-
"learning_rate": 3.6171875e-05,
|
| 607 |
-
"loss": 2.6452,
|
| 608 |
-
"step": 4902
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 38.0,
|
| 612 |
-
"eval_cosine_similarity": 0.7482897639274597,
|
| 613 |
-
"eval_loss": 2.678091287612915,
|
| 614 |
-
"eval_runtime": 18.8365,
|
| 615 |
-
"eval_samples_per_second": 22.934,
|
| 616 |
-
"eval_steps_per_second": 1.433,
|
| 617 |
-
"step": 4902
|
| 618 |
-
},
|
| 619 |
-
{
|
| 620 |
-
"epoch": 39.0,
|
| 621 |
-
"grad_norm": 0.34699180722236633,
|
| 622 |
-
"learning_rate": 3.44921875e-05,
|
| 623 |
-
"loss": 2.6442,
|
| 624 |
-
"step": 5031
|
| 625 |
-
},
|
| 626 |
-
{
|
| 627 |
-
"epoch": 39.0,
|
| 628 |
-
"eval_cosine_similarity": 0.7534282803535461,
|
| 629 |
-
"eval_loss": 2.678204298019409,
|
| 630 |
-
"eval_runtime": 18.8443,
|
| 631 |
-
"eval_samples_per_second": 22.925,
|
| 632 |
-
"eval_steps_per_second": 1.433,
|
| 633 |
-
"step": 5031
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 40.0,
|
| 637 |
-
"grad_norm": 0.36323243379592896,
|
| 638 |
-
"learning_rate": 3.2812500000000005e-05,
|
| 639 |
-
"loss": 2.6437,
|
| 640 |
-
"step": 5160
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 40.0,
|
| 644 |
-
"eval_cosine_similarity": 0.7704055905342102,
|
| 645 |
-
"eval_loss": 2.676304817199707,
|
| 646 |
-
"eval_runtime": 18.8549,
|
| 647 |
-
"eval_samples_per_second": 22.912,
|
| 648 |
-
"eval_steps_per_second": 1.432,
|
| 649 |
-
"step": 5160
|
| 650 |
-
},
|
| 651 |
-
{
|
| 652 |
-
"epoch": 41.0,
|
| 653 |
-
"grad_norm": 0.4448683261871338,
|
| 654 |
-
"learning_rate": 3.1132812500000005e-05,
|
| 655 |
-
"loss": 2.6429,
|
| 656 |
-
"step": 5289
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 41.0,
|
| 660 |
-
"eval_cosine_similarity": 0.7627379894256592,
|
| 661 |
-
"eval_loss": 2.67691707611084,
|
| 662 |
-
"eval_runtime": 18.836,
|
| 663 |
-
"eval_samples_per_second": 22.935,
|
| 664 |
-
"eval_steps_per_second": 1.433,
|
| 665 |
-
"step": 5289
|
| 666 |
-
},
|
| 667 |
-
{
|
| 668 |
-
"epoch": 42.0,
|
| 669 |
-
"grad_norm": 0.3076535165309906,
|
| 670 |
-
"learning_rate": 2.9453125000000003e-05,
|
| 671 |
-
"loss": 2.6428,
|
| 672 |
-
"step": 5418
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 42.0,
|
| 676 |
-
"eval_cosine_similarity": 0.7730197310447693,
|
| 677 |
-
"eval_loss": 2.675767660140991,
|
| 678 |
-
"eval_runtime": 18.8744,
|
| 679 |
-
"eval_samples_per_second": 22.888,
|
| 680 |
-
"eval_steps_per_second": 1.431,
|
| 681 |
-
"step": 5418
|
| 682 |
-
},
|
| 683 |
-
{
|
| 684 |
-
"epoch": 43.0,
|
| 685 |
-
"grad_norm": 0.3055357336997986,
|
| 686 |
-
"learning_rate": 2.77734375e-05,
|
| 687 |
-
"loss": 2.6415,
|
| 688 |
-
"step": 5547
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 43.0,
|
| 692 |
-
"eval_cosine_similarity": 0.7752830982208252,
|
| 693 |
-
"eval_loss": 2.6751718521118164,
|
| 694 |
-
"eval_runtime": 18.8283,
|
| 695 |
-
"eval_samples_per_second": 22.944,
|
| 696 |
-
"eval_steps_per_second": 1.434,
|
| 697 |
-
"step": 5547
|
| 698 |
-
},
|
| 699 |
-
{
|
| 700 |
-
"epoch": 44.0,
|
| 701 |
-
"grad_norm": 0.4585816562175751,
|
| 702 |
-
"learning_rate": 2.609375e-05,
|
| 703 |
-
"loss": 2.6408,
|
| 704 |
-
"step": 5676
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 44.0,
|
| 708 |
-
"eval_cosine_similarity": 0.7713160514831543,
|
| 709 |
-
"eval_loss": 2.675260305404663,
|
| 710 |
-
"eval_runtime": 18.863,
|
| 711 |
-
"eval_samples_per_second": 22.902,
|
| 712 |
-
"eval_steps_per_second": 1.431,
|
| 713 |
-
"step": 5676
|
| 714 |
-
},
|
| 715 |
-
{
|
| 716 |
-
"epoch": 45.0,
|
| 717 |
-
"grad_norm": 0.1793605536222458,
|
| 718 |
-
"learning_rate": 2.44140625e-05,
|
| 719 |
-
"loss": 2.6401,
|
| 720 |
-
"step": 5805
|
| 721 |
-
},
|
| 722 |
-
{
|
| 723 |
-
"epoch": 45.0,
|
| 724 |
-
"eval_cosine_similarity": 0.7710201144218445,
|
| 725 |
-
"eval_loss": 2.675137519836426,
|
| 726 |
-
"eval_runtime": 18.8392,
|
| 727 |
-
"eval_samples_per_second": 22.931,
|
| 728 |
-
"eval_steps_per_second": 1.433,
|
| 729 |
-
"step": 5805
|
| 730 |
-
},
|
| 731 |
-
{
|
| 732 |
-
"epoch": 46.0,
|
| 733 |
-
"grad_norm": 0.2033383995294571,
|
| 734 |
-
"learning_rate": 2.2734375000000002e-05,
|
| 735 |
-
"loss": 2.6396,
|
| 736 |
-
"step": 5934
|
| 737 |
-
},
|
| 738 |
-
{
|
| 739 |
-
"epoch": 46.0,
|
| 740 |
-
"eval_cosine_similarity": 0.7817992568016052,
|
| 741 |
-
"eval_loss": 2.6738462448120117,
|
| 742 |
-
"eval_runtime": 18.8424,
|
| 743 |
-
"eval_samples_per_second": 22.927,
|
| 744 |
-
"eval_steps_per_second": 1.433,
|
| 745 |
-
"step": 5934
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 47.0,
|
| 749 |
-
"grad_norm": 0.3535211384296417,
|
| 750 |
-
"learning_rate": 2.10546875e-05,
|
| 751 |
-
"loss": 2.6388,
|
| 752 |
-
"step": 6063
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 47.0,
|
| 756 |
-
"eval_cosine_similarity": 0.7922196984291077,
|
| 757 |
-
"eval_loss": 2.673060894012451,
|
| 758 |
-
"eval_runtime": 18.8542,
|
| 759 |
-
"eval_samples_per_second": 22.913,
|
| 760 |
-
"eval_steps_per_second": 1.432,
|
| 761 |
-
"step": 6063
|
| 762 |
-
},
|
| 763 |
-
{
|
| 764 |
-
"epoch": 48.0,
|
| 765 |
-
"grad_norm": 0.29340553283691406,
|
| 766 |
-
"learning_rate": 1.9375e-05,
|
| 767 |
-
"loss": 2.6383,
|
| 768 |
-
"step": 6192
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 48.0,
|
| 772 |
-
"eval_cosine_similarity": 0.7896454334259033,
|
| 773 |
-
"eval_loss": 2.6733312606811523,
|
| 774 |
-
"eval_runtime": 18.8372,
|
| 775 |
-
"eval_samples_per_second": 22.933,
|
| 776 |
-
"eval_steps_per_second": 1.433,
|
| 777 |
-
"step": 6192
|
| 778 |
}
|
| 779 |
],
|
| 780 |
"logging_steps": 500,
|
|
@@ -789,7 +37,7 @@
|
|
| 789 |
"early_stopping_threshold": 0.0
|
| 790 |
},
|
| 791 |
"attributes": {
|
| 792 |
-
"early_stopping_patience_counter":
|
| 793 |
}
|
| 794 |
},
|
| 795 |
"TrainerControl": {
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.754194974899292,
|
| 3 |
+
"best_model_checkpoint": "mgh6/HTH_biCLIP_mean/checkpoint-129",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 129,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 1.0,
|
| 13 |
+
"grad_norm": 0.009723234921693802,
|
| 14 |
+
"learning_rate": 0.000983203125,
|
| 15 |
+
"loss": 2.7359,
|
| 16 |
"step": 129
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 1.0,
|
| 20 |
+
"eval_cosine_similarity": 0.15642696619033813,
|
| 21 |
+
"eval_loss": 2.754194974899292,
|
| 22 |
+
"eval_runtime": 18.9326,
|
| 23 |
+
"eval_samples_per_second": 22.818,
|
| 24 |
+
"eval_steps_per_second": 1.426,
|
| 25 |
"step": 129
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
}
|
| 27 |
],
|
| 28 |
"logging_steps": 500,
|
|
|
|
| 37 |
"early_stopping_threshold": 0.0
|
| 38 |
},
|
| 39 |
"attributes": {
|
| 40 |
+
"early_stopping_patience_counter": 0
|
| 41 |
}
|
| 42 |
},
|
| 43 |
"TrainerControl": {
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5368
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:456d4f7d71f44316c3344e7b1dd2abd8cad0993cbcf9e11348a97ca882a75f92
|
| 3 |
size 5368
|