End of training
Browse files- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +743 -225
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second": 0.
|
| 8 |
"train_steps_per_second": 0.072
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"total_flos": 1.955920022807249e+18,
|
| 4 |
+
"train_loss": 0.8700304180128008,
|
| 5 |
+
"train_runtime": 179145.9612,
|
| 6 |
+
"train_samples": 153863,
|
| 7 |
+
"train_samples_per_second": 0.859,
|
| 8 |
"train_steps_per_second": 0.072
|
| 9 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second": 0.
|
| 8 |
"train_steps_per_second": 0.072
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"total_flos": 1.955920022807249e+18,
|
| 4 |
+
"train_loss": 0.8700304180128008,
|
| 5 |
+
"train_runtime": 179145.9612,
|
| 6 |
+
"train_samples": 153863,
|
| 7 |
+
"train_samples_per_second": 0.859,
|
| 8 |
"train_steps_per_second": 0.072
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -3,404 +3,922 @@
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 1.0,
|
| 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": 0.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate": 4.
|
| 15 |
-
"loss": 0.
|
| 16 |
"step": 100
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate": 4.
|
| 22 |
-
"loss": 0.
|
| 23 |
"step": 200
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"learning_rate": 4.
|
| 29 |
-
"loss": 0.
|
| 30 |
"step": 300
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate": 4.
|
| 36 |
-
"loss": 0.
|
| 37 |
"step": 400
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
-
"learning_rate": 4.
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 500
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate": 4.
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 600
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate": 4.
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 700
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate": 4.
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 800
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm":
|
| 70 |
-
"learning_rate": 4.
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 900
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate": 4.
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 1000
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss": 0.
|
| 86 |
"step": 1100
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 1200
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm":
|
| 98 |
-
"learning_rate":
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 1300
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm":
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 1400
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate":
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 1500
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 1600
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm":
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 1700
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 1800
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm":
|
| 140 |
-
"learning_rate":
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 1900
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 2000
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm":
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 2100
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 0.
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate":
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 2200
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 2300
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 2400
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm":
|
| 182 |
-
"learning_rate":
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 2500
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"grad_norm":
|
| 189 |
-
"learning_rate":
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 2600
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"grad_norm":
|
| 196 |
-
"learning_rate":
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 2700
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 0.
|
| 202 |
-
"grad_norm":
|
| 203 |
-
"learning_rate":
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 2800
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 0.
|
| 209 |
-
"grad_norm":
|
| 210 |
-
"learning_rate":
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 2900
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 0.
|
| 216 |
-
"grad_norm":
|
| 217 |
-
"learning_rate":
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 3000
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate":
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 3100
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm":
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 3200
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"grad_norm":
|
| 238 |
-
"learning_rate":
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 3300
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 0.
|
| 244 |
-
"grad_norm":
|
| 245 |
-
"learning_rate":
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 3400
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch": 0.
|
| 251 |
-
"grad_norm":
|
| 252 |
-
"learning_rate":
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 3500
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch": 0.
|
| 258 |
-
"grad_norm":
|
| 259 |
-
"learning_rate":
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 3600
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch": 0.
|
| 265 |
-
"grad_norm":
|
| 266 |
-
"learning_rate":
|
| 267 |
-
"loss": 0.
|
| 268 |
"step": 3700
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"epoch": 0.
|
| 272 |
-
"grad_norm":
|
| 273 |
-
"learning_rate":
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 3800
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 0.
|
| 279 |
-
"grad_norm":
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 3900
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 0.
|
| 286 |
-
"grad_norm":
|
| 287 |
-
"learning_rate":
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 4000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch": 0.
|
| 293 |
-
"grad_norm":
|
| 294 |
-
"learning_rate":
|
| 295 |
-
"loss": 0.
|
| 296 |
"step": 4100
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch": 0.
|
| 300 |
-
"grad_norm":
|
| 301 |
-
"learning_rate":
|
| 302 |
-
"loss": 0.
|
| 303 |
"step": 4200
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 0.
|
| 307 |
-
"grad_norm":
|
| 308 |
-
"learning_rate":
|
| 309 |
-
"loss": 0.
|
| 310 |
"step": 4300
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch": 0.
|
| 314 |
-
"grad_norm":
|
| 315 |
-
"learning_rate":
|
| 316 |
-
"loss": 0.
|
| 317 |
"step": 4400
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 0.
|
| 321 |
-
"grad_norm":
|
| 322 |
-
"learning_rate":
|
| 323 |
-
"loss": 0.
|
| 324 |
"step": 4500
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch": 0.
|
| 328 |
-
"grad_norm":
|
| 329 |
-
"learning_rate":
|
| 330 |
-
"loss": 0.
|
| 331 |
"step": 4600
|
| 332 |
},
|
| 333 |
{
|
| 334 |
-
"epoch": 0.
|
| 335 |
-
"grad_norm":
|
| 336 |
-
"learning_rate":
|
| 337 |
-
"loss": 0.
|
| 338 |
"step": 4700
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch": 0.
|
| 342 |
-
"grad_norm":
|
| 343 |
-
"learning_rate":
|
| 344 |
-
"loss": 0.
|
| 345 |
"step": 4800
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch": 0.
|
| 349 |
-
"grad_norm":
|
| 350 |
-
"learning_rate":
|
| 351 |
-
"loss": 0.
|
| 352 |
"step": 4900
|
| 353 |
},
|
| 354 |
{
|
| 355 |
-
"epoch": 0.
|
| 356 |
-
"grad_norm":
|
| 357 |
-
"learning_rate": 3.
|
| 358 |
-
"loss": 0.
|
| 359 |
"step": 5000
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 0.
|
| 363 |
-
"grad_norm":
|
| 364 |
-
"learning_rate":
|
| 365 |
-
"loss": 0.
|
| 366 |
"step": 5100
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch": 0.
|
| 370 |
-
"grad_norm":
|
| 371 |
-
"learning_rate": 2.
|
| 372 |
-
"loss": 0.
|
| 373 |
"step": 5200
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch": 0.
|
| 377 |
-
"grad_norm":
|
| 378 |
-
"learning_rate":
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 5300
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch": 0.
|
| 384 |
-
"grad_norm":
|
| 385 |
-
"learning_rate":
|
| 386 |
-
"loss": 0.
|
| 387 |
"step": 5400
|
| 388 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 389 |
{
|
| 390 |
"epoch": 1.0,
|
| 391 |
-
"step":
|
| 392 |
-
"total_flos":
|
| 393 |
-
"train_loss": 0.
|
| 394 |
-
"train_runtime":
|
| 395 |
-
"train_samples_per_second": 0.
|
| 396 |
"train_steps_per_second": 0.072
|
| 397 |
}
|
| 398 |
],
|
| 399 |
"logging_steps": 100,
|
| 400 |
-
"max_steps":
|
| 401 |
"num_input_tokens_seen": 0,
|
| 402 |
"num_train_epochs": 1,
|
| 403 |
-
"save_steps":
|
| 404 |
"stateful_callbacks": {
|
| 405 |
"TrainerControl": {
|
| 406 |
"args": {
|
|
@@ -413,7 +931,7 @@
|
|
| 413 |
"attributes": {}
|
| 414 |
}
|
| 415 |
},
|
| 416 |
-
"total_flos":
|
| 417 |
"train_batch_size": 12,
|
| 418 |
"trial_name": null,
|
| 419 |
"trial_params": null
|
|
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 12822,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.007799095304944627,
|
| 13 |
+
"grad_norm": 63896.45703125,
|
| 14 |
+
"learning_rate": 4.961004523475277e-05,
|
| 15 |
+
"loss": 0.929,
|
| 16 |
"step": 100
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.015598190609889253,
|
| 20 |
+
"grad_norm": 79284.765625,
|
| 21 |
+
"learning_rate": 4.9220090469505544e-05,
|
| 22 |
+
"loss": 0.9281,
|
| 23 |
"step": 200
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.02339728591483388,
|
| 27 |
+
"grad_norm": 80169.203125,
|
| 28 |
+
"learning_rate": 4.883013570425831e-05,
|
| 29 |
+
"loss": 0.9385,
|
| 30 |
"step": 300
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.031196381219778507,
|
| 34 |
+
"grad_norm": 65649.6796875,
|
| 35 |
+
"learning_rate": 4.844018093901108e-05,
|
| 36 |
+
"loss": 0.9454,
|
| 37 |
"step": 400
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.038995476524723134,
|
| 41 |
+
"grad_norm": 82255.6953125,
|
| 42 |
+
"learning_rate": 4.805022617376384e-05,
|
| 43 |
+
"loss": 0.9368,
|
| 44 |
"step": 500
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.04679457182966776,
|
| 48 |
+
"grad_norm": 78376.4296875,
|
| 49 |
+
"learning_rate": 4.7660271408516614e-05,
|
| 50 |
+
"loss": 0.9318,
|
| 51 |
"step": 600
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.05459366713461238,
|
| 55 |
+
"grad_norm": 92200.078125,
|
| 56 |
+
"learning_rate": 4.7270316643269384e-05,
|
| 57 |
+
"loss": 0.9272,
|
| 58 |
"step": 700
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.06239276243955701,
|
| 62 |
+
"grad_norm": 88060.359375,
|
| 63 |
+
"learning_rate": 4.688036187802215e-05,
|
| 64 |
+
"loss": 0.9097,
|
| 65 |
"step": 800
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.07019185774450164,
|
| 69 |
+
"grad_norm": 76161.1796875,
|
| 70 |
+
"learning_rate": 4.649040711277492e-05,
|
| 71 |
+
"loss": 0.9475,
|
| 72 |
"step": 900
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.07799095304944627,
|
| 76 |
+
"grad_norm": 90524.1796875,
|
| 77 |
+
"learning_rate": 4.610045234752769e-05,
|
| 78 |
+
"loss": 0.9481,
|
| 79 |
"step": 1000
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.08579004835439089,
|
| 83 |
+
"grad_norm": 100124.8359375,
|
| 84 |
+
"learning_rate": 4.571049758228046e-05,
|
| 85 |
+
"loss": 0.9333,
|
| 86 |
"step": 1100
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.09358914365933552,
|
| 90 |
+
"grad_norm": 65681.7734375,
|
| 91 |
+
"learning_rate": 4.5320542817033225e-05,
|
| 92 |
+
"loss": 0.9247,
|
| 93 |
"step": 1200
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.10138823896428015,
|
| 97 |
+
"grad_norm": 76333.234375,
|
| 98 |
+
"learning_rate": 4.4930588051785995e-05,
|
| 99 |
+
"loss": 0.9098,
|
| 100 |
"step": 1300
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.10918733426922476,
|
| 104 |
+
"grad_norm": 77002.0703125,
|
| 105 |
+
"learning_rate": 4.4540633286538766e-05,
|
| 106 |
+
"loss": 0.9034,
|
| 107 |
"step": 1400
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.1169864295741694,
|
| 111 |
+
"grad_norm": 72669.1875,
|
| 112 |
+
"learning_rate": 4.415067852129153e-05,
|
| 113 |
+
"loss": 0.9149,
|
| 114 |
"step": 1500
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.12478552487911403,
|
| 118 |
+
"grad_norm": 71785.34375,
|
| 119 |
+
"learning_rate": 4.37607237560443e-05,
|
| 120 |
+
"loss": 0.9051,
|
| 121 |
"step": 1600
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.13258462018405864,
|
| 125 |
+
"grad_norm": 72728.5859375,
|
| 126 |
+
"learning_rate": 4.3370768990797065e-05,
|
| 127 |
+
"loss": 0.9268,
|
| 128 |
"step": 1700
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.14038371548900327,
|
| 132 |
+
"grad_norm": 69223.3203125,
|
| 133 |
+
"learning_rate": 4.2980814225549836e-05,
|
| 134 |
+
"loss": 0.9294,
|
| 135 |
"step": 1800
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.1481828107939479,
|
| 139 |
+
"grad_norm": 62745.203125,
|
| 140 |
+
"learning_rate": 4.2590859460302607e-05,
|
| 141 |
+
"loss": 0.9038,
|
| 142 |
"step": 1900
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.15598190609889254,
|
| 146 |
+
"grad_norm": 79897.8671875,
|
| 147 |
+
"learning_rate": 4.220090469505538e-05,
|
| 148 |
+
"loss": 0.8994,
|
| 149 |
"step": 2000
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.16378100140383717,
|
| 153 |
+
"grad_norm": 74131.546875,
|
| 154 |
+
"learning_rate": 4.181094992980815e-05,
|
| 155 |
+
"loss": 0.9144,
|
| 156 |
"step": 2100
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 0.17158009670878177,
|
| 160 |
+
"grad_norm": 67760.7109375,
|
| 161 |
+
"learning_rate": 4.142099516456091e-05,
|
| 162 |
+
"loss": 0.9285,
|
| 163 |
"step": 2200
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.1793791920137264,
|
| 167 |
+
"grad_norm": 78868.1484375,
|
| 168 |
+
"learning_rate": 4.103104039931368e-05,
|
| 169 |
+
"loss": 0.8973,
|
| 170 |
"step": 2300
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.18717828731867103,
|
| 174 |
+
"grad_norm": 71780.2109375,
|
| 175 |
+
"learning_rate": 4.064108563406645e-05,
|
| 176 |
+
"loss": 0.8961,
|
| 177 |
"step": 2400
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.19497738262361566,
|
| 181 |
+
"grad_norm": 77858.109375,
|
| 182 |
+
"learning_rate": 4.025113086881922e-05,
|
| 183 |
+
"loss": 0.9171,
|
| 184 |
"step": 2500
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 0.2027764779285603,
|
| 188 |
+
"grad_norm": 70062.796875,
|
| 189 |
+
"learning_rate": 3.986117610357199e-05,
|
| 190 |
+
"loss": 0.8903,
|
| 191 |
"step": 2600
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.21057557323350493,
|
| 195 |
+
"grad_norm": 171091.046875,
|
| 196 |
+
"learning_rate": 3.947122133832475e-05,
|
| 197 |
+
"loss": 0.9061,
|
| 198 |
"step": 2700
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 0.21837466853844953,
|
| 202 |
+
"grad_norm": 84371.34375,
|
| 203 |
+
"learning_rate": 3.908126657307752e-05,
|
| 204 |
+
"loss": 0.9315,
|
| 205 |
"step": 2800
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.22617376384339416,
|
| 209 |
+
"grad_norm": 62273.2578125,
|
| 210 |
+
"learning_rate": 3.8691311807830294e-05,
|
| 211 |
+
"loss": 0.9115,
|
| 212 |
"step": 2900
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 0.2339728591483388,
|
| 216 |
+
"grad_norm": 60464.15625,
|
| 217 |
+
"learning_rate": 3.8301357042583065e-05,
|
| 218 |
+
"loss": 0.8888,
|
| 219 |
"step": 3000
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.24177195445328342,
|
| 223 |
+
"grad_norm": 67546.828125,
|
| 224 |
+
"learning_rate": 3.7911402277335836e-05,
|
| 225 |
+
"loss": 0.9031,
|
| 226 |
"step": 3100
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.24957104975822805,
|
| 230 |
+
"grad_norm": 75904.6328125,
|
| 231 |
+
"learning_rate": 3.75214475120886e-05,
|
| 232 |
+
"loss": 0.9146,
|
| 233 |
"step": 3200
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.2573701450631727,
|
| 237 |
+
"grad_norm": 81015.3828125,
|
| 238 |
+
"learning_rate": 3.713149274684137e-05,
|
| 239 |
+
"loss": 0.8935,
|
| 240 |
"step": 3300
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 0.2651692403681173,
|
| 244 |
+
"grad_norm": 69763.3203125,
|
| 245 |
+
"learning_rate": 3.6741537981594134e-05,
|
| 246 |
+
"loss": 0.9129,
|
| 247 |
"step": 3400
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.27296833567306195,
|
| 251 |
+
"grad_norm": 68604.5,
|
| 252 |
+
"learning_rate": 3.6351583216346905e-05,
|
| 253 |
+
"loss": 0.8834,
|
| 254 |
"step": 3500
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 0.28076743097800655,
|
| 258 |
+
"grad_norm": 66758.03125,
|
| 259 |
+
"learning_rate": 3.596162845109967e-05,
|
| 260 |
+
"loss": 0.8925,
|
| 261 |
"step": 3600
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.28856652628295115,
|
| 265 |
+
"grad_norm": 72424.8984375,
|
| 266 |
+
"learning_rate": 3.557167368585244e-05,
|
| 267 |
+
"loss": 0.915,
|
| 268 |
"step": 3700
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 0.2963656215878958,
|
| 272 |
+
"grad_norm": 65904.9453125,
|
| 273 |
+
"learning_rate": 3.518171892060521e-05,
|
| 274 |
+
"loss": 0.8833,
|
| 275 |
"step": 3800
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.3041647168928404,
|
| 279 |
+
"grad_norm": 62548.9453125,
|
| 280 |
+
"learning_rate": 3.479176415535798e-05,
|
| 281 |
+
"loss": 0.8991,
|
| 282 |
"step": 3900
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.3119638121977851,
|
| 286 |
+
"grad_norm": 85173.90625,
|
| 287 |
+
"learning_rate": 3.440180939011075e-05,
|
| 288 |
+
"loss": 0.9111,
|
| 289 |
"step": 4000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.3197629075027297,
|
| 293 |
+
"grad_norm": 65947.765625,
|
| 294 |
+
"learning_rate": 3.4011854624863516e-05,
|
| 295 |
+
"loss": 0.8774,
|
| 296 |
"step": 4100
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 0.32756200280767434,
|
| 300 |
+
"grad_norm": 69672.1875,
|
| 301 |
+
"learning_rate": 3.362189985961629e-05,
|
| 302 |
+
"loss": 0.8816,
|
| 303 |
"step": 4200
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 0.33536109811261894,
|
| 307 |
+
"grad_norm": 68125.9609375,
|
| 308 |
+
"learning_rate": 3.323194509436906e-05,
|
| 309 |
+
"loss": 0.8784,
|
| 310 |
"step": 4300
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 0.34316019341756354,
|
| 314 |
+
"grad_norm": 76767.6796875,
|
| 315 |
+
"learning_rate": 3.284199032912182e-05,
|
| 316 |
+
"loss": 0.899,
|
| 317 |
"step": 4400
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 0.3509592887225082,
|
| 321 |
+
"grad_norm": 64417.96484375,
|
| 322 |
+
"learning_rate": 3.245203556387459e-05,
|
| 323 |
+
"loss": 0.8944,
|
| 324 |
"step": 4500
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 0.3587583840274528,
|
| 328 |
+
"grad_norm": 62896.37109375,
|
| 329 |
+
"learning_rate": 3.206208079862736e-05,
|
| 330 |
+
"loss": 0.8745,
|
| 331 |
"step": 4600
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 0.36655747933239746,
|
| 335 |
+
"grad_norm": 58443.88671875,
|
| 336 |
+
"learning_rate": 3.167212603338013e-05,
|
| 337 |
+
"loss": 0.853,
|
| 338 |
"step": 4700
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 0.37435657463734207,
|
| 342 |
+
"grad_norm": 75019.8515625,
|
| 343 |
+
"learning_rate": 3.12821712681329e-05,
|
| 344 |
+
"loss": 0.8728,
|
| 345 |
"step": 4800
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 0.38215566994228667,
|
| 349 |
+
"grad_norm": 82645.0078125,
|
| 350 |
+
"learning_rate": 3.089221650288567e-05,
|
| 351 |
+
"loss": 0.8617,
|
| 352 |
"step": 4900
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 0.3899547652472313,
|
| 356 |
+
"grad_norm": 57014.08984375,
|
| 357 |
+
"learning_rate": 3.0502261737638436e-05,
|
| 358 |
+
"loss": 0.8566,
|
| 359 |
"step": 5000
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 0.39775386055217593,
|
| 363 |
+
"grad_norm": 71623.7578125,
|
| 364 |
+
"learning_rate": 3.01123069723912e-05,
|
| 365 |
+
"loss": 0.879,
|
| 366 |
"step": 5100
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 0.4055529558571206,
|
| 370 |
+
"grad_norm": 75911.140625,
|
| 371 |
+
"learning_rate": 2.972235220714397e-05,
|
| 372 |
+
"loss": 0.8889,
|
| 373 |
"step": 5200
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 0.4133520511620652,
|
| 377 |
+
"grad_norm": 69251.0703125,
|
| 378 |
+
"learning_rate": 2.9332397441896742e-05,
|
| 379 |
+
"loss": 0.8685,
|
| 380 |
"step": 5300
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"epoch": 0.42115114646700985,
|
| 384 |
+
"grad_norm": 72434.6484375,
|
| 385 |
+
"learning_rate": 2.8942442676649513e-05,
|
| 386 |
+
"loss": 0.8895,
|
| 387 |
"step": 5400
|
| 388 |
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.42895024177195445,
|
| 391 |
+
"grad_norm": 69133.78125,
|
| 392 |
+
"learning_rate": 2.855248791140228e-05,
|
| 393 |
+
"loss": 0.8845,
|
| 394 |
+
"step": 5500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.43674933707689906,
|
| 398 |
+
"grad_norm": 67216.6875,
|
| 399 |
+
"learning_rate": 2.8162533146155044e-05,
|
| 400 |
+
"loss": 0.8661,
|
| 401 |
+
"step": 5600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.4445484323818437,
|
| 405 |
+
"grad_norm": 61408.3515625,
|
| 406 |
+
"learning_rate": 2.7772578380907815e-05,
|
| 407 |
+
"loss": 0.8743,
|
| 408 |
+
"step": 5700
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.4523475276867883,
|
| 412 |
+
"grad_norm": 66496.1640625,
|
| 413 |
+
"learning_rate": 2.7382623615660586e-05,
|
| 414 |
+
"loss": 0.8805,
|
| 415 |
+
"step": 5800
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.460146622991733,
|
| 419 |
+
"grad_norm": 62693.109375,
|
| 420 |
+
"learning_rate": 2.6992668850413357e-05,
|
| 421 |
+
"loss": 0.8634,
|
| 422 |
+
"step": 5900
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.4679457182966776,
|
| 426 |
+
"grad_norm": 77037.4140625,
|
| 427 |
+
"learning_rate": 2.660271408516612e-05,
|
| 428 |
+
"loss": 0.8834,
|
| 429 |
+
"step": 6000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.4757448136016222,
|
| 433 |
+
"grad_norm": 67157.15625,
|
| 434 |
+
"learning_rate": 2.6212759319918888e-05,
|
| 435 |
+
"loss": 0.8993,
|
| 436 |
+
"step": 6100
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.48354390890656684,
|
| 440 |
+
"grad_norm": 77155.8125,
|
| 441 |
+
"learning_rate": 2.582280455467166e-05,
|
| 442 |
+
"loss": 0.854,
|
| 443 |
+
"step": 6200
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.49134300421151145,
|
| 447 |
+
"grad_norm": 65097.22265625,
|
| 448 |
+
"learning_rate": 2.543284978942443e-05,
|
| 449 |
+
"loss": 0.8716,
|
| 450 |
+
"step": 6300
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.4991420995164561,
|
| 454 |
+
"grad_norm": 89373.515625,
|
| 455 |
+
"learning_rate": 2.50428950241772e-05,
|
| 456 |
+
"loss": 0.873,
|
| 457 |
+
"step": 6400
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.5069411948214008,
|
| 461 |
+
"grad_norm": 65231.74609375,
|
| 462 |
+
"learning_rate": 2.4652940258929964e-05,
|
| 463 |
+
"loss": 0.8698,
|
| 464 |
+
"step": 6500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.5147402901263454,
|
| 468 |
+
"grad_norm": 60228.23828125,
|
| 469 |
+
"learning_rate": 2.426298549368273e-05,
|
| 470 |
+
"loss": 0.8374,
|
| 471 |
+
"step": 6600
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.52253938543129,
|
| 475 |
+
"grad_norm": 73940.3125,
|
| 476 |
+
"learning_rate": 2.3873030728435502e-05,
|
| 477 |
+
"loss": 0.8669,
|
| 478 |
+
"step": 6700
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.5303384807362346,
|
| 482 |
+
"grad_norm": 80768.6953125,
|
| 483 |
+
"learning_rate": 2.3483075963188273e-05,
|
| 484 |
+
"loss": 0.8708,
|
| 485 |
+
"step": 6800
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.5381375760411792,
|
| 489 |
+
"grad_norm": 67096.984375,
|
| 490 |
+
"learning_rate": 2.309312119794104e-05,
|
| 491 |
+
"loss": 0.8797,
|
| 492 |
+
"step": 6900
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.5459366713461239,
|
| 496 |
+
"grad_norm": 56897.71875,
|
| 497 |
+
"learning_rate": 2.2703166432693808e-05,
|
| 498 |
+
"loss": 0.8728,
|
| 499 |
+
"step": 7000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.5537357666510685,
|
| 503 |
+
"grad_norm": 64982.1015625,
|
| 504 |
+
"learning_rate": 2.2313211667446575e-05,
|
| 505 |
+
"loss": 0.8735,
|
| 506 |
+
"step": 7100
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.5615348619560131,
|
| 510 |
+
"grad_norm": 62320.35546875,
|
| 511 |
+
"learning_rate": 2.1923256902199346e-05,
|
| 512 |
+
"loss": 0.8519,
|
| 513 |
+
"step": 7200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.5693339572609577,
|
| 517 |
+
"grad_norm": 74915.2578125,
|
| 518 |
+
"learning_rate": 2.1533302136952117e-05,
|
| 519 |
+
"loss": 0.849,
|
| 520 |
+
"step": 7300
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.5771330525659023,
|
| 524 |
+
"grad_norm": 69091.40625,
|
| 525 |
+
"learning_rate": 2.1143347371704884e-05,
|
| 526 |
+
"loss": 0.8451,
|
| 527 |
+
"step": 7400
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.584932147870847,
|
| 531 |
+
"grad_norm": 75015.2109375,
|
| 532 |
+
"learning_rate": 2.0753392606457652e-05,
|
| 533 |
+
"loss": 0.8757,
|
| 534 |
+
"step": 7500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5927312431757916,
|
| 538 |
+
"grad_norm": 59275.14453125,
|
| 539 |
+
"learning_rate": 2.036343784121042e-05,
|
| 540 |
+
"loss": 0.8657,
|
| 541 |
+
"step": 7600
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.6005303384807362,
|
| 545 |
+
"grad_norm": 64822.92578125,
|
| 546 |
+
"learning_rate": 1.997348307596319e-05,
|
| 547 |
+
"loss": 0.8555,
|
| 548 |
+
"step": 7700
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.6083294337856808,
|
| 552 |
+
"grad_norm": 63732.34765625,
|
| 553 |
+
"learning_rate": 1.9583528310715957e-05,
|
| 554 |
+
"loss": 0.8415,
|
| 555 |
+
"step": 7800
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.6161285290906254,
|
| 559 |
+
"grad_norm": 66452.859375,
|
| 560 |
+
"learning_rate": 1.9193573545468728e-05,
|
| 561 |
+
"loss": 0.8475,
|
| 562 |
+
"step": 7900
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.6239276243955701,
|
| 566 |
+
"grad_norm": 69227.2890625,
|
| 567 |
+
"learning_rate": 1.8803618780221496e-05,
|
| 568 |
+
"loss": 0.8333,
|
| 569 |
+
"step": 8000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.6317267197005147,
|
| 573 |
+
"grad_norm": 67040.7578125,
|
| 574 |
+
"learning_rate": 1.8413664014974263e-05,
|
| 575 |
+
"loss": 0.8317,
|
| 576 |
+
"step": 8100
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.6395258150054594,
|
| 580 |
+
"grad_norm": 74801.2734375,
|
| 581 |
+
"learning_rate": 1.8023709249727034e-05,
|
| 582 |
+
"loss": 0.8726,
|
| 583 |
+
"step": 8200
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.647324910310404,
|
| 587 |
+
"grad_norm": 63112.703125,
|
| 588 |
+
"learning_rate": 1.76337544844798e-05,
|
| 589 |
+
"loss": 0.8442,
|
| 590 |
+
"step": 8300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.6551240056153487,
|
| 594 |
+
"grad_norm": 81685.2265625,
|
| 595 |
+
"learning_rate": 1.724379971923257e-05,
|
| 596 |
+
"loss": 0.8316,
|
| 597 |
+
"step": 8400
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.6629231009202933,
|
| 601 |
+
"grad_norm": 69446.4921875,
|
| 602 |
+
"learning_rate": 1.6853844953985336e-05,
|
| 603 |
+
"loss": 0.8391,
|
| 604 |
+
"step": 8500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.6707221962252379,
|
| 608 |
+
"grad_norm": 66903.671875,
|
| 609 |
+
"learning_rate": 1.6463890188738107e-05,
|
| 610 |
+
"loss": 0.8144,
|
| 611 |
+
"step": 8600
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.6785212915301825,
|
| 615 |
+
"grad_norm": 76092.6328125,
|
| 616 |
+
"learning_rate": 1.6073935423490877e-05,
|
| 617 |
+
"loss": 0.8441,
|
| 618 |
+
"step": 8700
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.6863203868351271,
|
| 622 |
+
"grad_norm": 71659.8515625,
|
| 623 |
+
"learning_rate": 1.5683980658243645e-05,
|
| 624 |
+
"loss": 0.8388,
|
| 625 |
+
"step": 8800
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.6941194821400718,
|
| 629 |
+
"grad_norm": 61097.1171875,
|
| 630 |
+
"learning_rate": 1.5294025892996412e-05,
|
| 631 |
+
"loss": 0.8383,
|
| 632 |
+
"step": 8900
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.7019185774450164,
|
| 636 |
+
"grad_norm": 60165.2734375,
|
| 637 |
+
"learning_rate": 1.4904071127749181e-05,
|
| 638 |
+
"loss": 0.8442,
|
| 639 |
+
"step": 9000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.709717672749961,
|
| 643 |
+
"grad_norm": 71296.1171875,
|
| 644 |
+
"learning_rate": 1.451411636250195e-05,
|
| 645 |
+
"loss": 0.8537,
|
| 646 |
+
"step": 9100
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.7175167680549056,
|
| 650 |
+
"grad_norm": 67209.6328125,
|
| 651 |
+
"learning_rate": 1.4124161597254721e-05,
|
| 652 |
+
"loss": 0.8575,
|
| 653 |
+
"step": 9200
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.7253158633598502,
|
| 657 |
+
"grad_norm": 63280.3203125,
|
| 658 |
+
"learning_rate": 1.3734206832007487e-05,
|
| 659 |
+
"loss": 0.8358,
|
| 660 |
+
"step": 9300
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.7331149586647949,
|
| 664 |
+
"grad_norm": 69636.6875,
|
| 665 |
+
"learning_rate": 1.3344252066760258e-05,
|
| 666 |
+
"loss": 0.8475,
|
| 667 |
+
"step": 9400
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.7409140539697395,
|
| 671 |
+
"grad_norm": 68302.3828125,
|
| 672 |
+
"learning_rate": 1.2954297301513025e-05,
|
| 673 |
+
"loss": 0.8305,
|
| 674 |
+
"step": 9500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.7487131492746841,
|
| 678 |
+
"grad_norm": 68986.78125,
|
| 679 |
+
"learning_rate": 1.2564342536265794e-05,
|
| 680 |
+
"loss": 0.8667,
|
| 681 |
+
"step": 9600
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.7565122445796287,
|
| 685 |
+
"grad_norm": 69115.0546875,
|
| 686 |
+
"learning_rate": 1.2174387771018563e-05,
|
| 687 |
+
"loss": 0.8386,
|
| 688 |
+
"step": 9700
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.7643113398845733,
|
| 692 |
+
"grad_norm": 58786.73046875,
|
| 693 |
+
"learning_rate": 1.178443300577133e-05,
|
| 694 |
+
"loss": 0.8306,
|
| 695 |
+
"step": 9800
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.772110435189518,
|
| 699 |
+
"grad_norm": 75233.265625,
|
| 700 |
+
"learning_rate": 1.13944782405241e-05,
|
| 701 |
+
"loss": 0.8611,
|
| 702 |
+
"step": 9900
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.7799095304944627,
|
| 706 |
+
"grad_norm": 65838.609375,
|
| 707 |
+
"learning_rate": 1.1004523475276867e-05,
|
| 708 |
+
"loss": 0.8191,
|
| 709 |
+
"step": 10000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.7877086257994073,
|
| 713 |
+
"grad_norm": 55231.33984375,
|
| 714 |
+
"learning_rate": 1.0614568710029638e-05,
|
| 715 |
+
"loss": 0.8551,
|
| 716 |
+
"step": 10100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.7955077211043519,
|
| 720 |
+
"grad_norm": 63516.97265625,
|
| 721 |
+
"learning_rate": 1.0224613944782405e-05,
|
| 722 |
+
"loss": 0.8685,
|
| 723 |
+
"step": 10200
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.8033068164092966,
|
| 727 |
+
"grad_norm": 63661.3671875,
|
| 728 |
+
"learning_rate": 9.834659179535174e-06,
|
| 729 |
+
"loss": 0.8009,
|
| 730 |
+
"step": 10300
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.8111059117142412,
|
| 734 |
+
"grad_norm": 71024.375,
|
| 735 |
+
"learning_rate": 9.444704414287943e-06,
|
| 736 |
+
"loss": 0.8331,
|
| 737 |
+
"step": 10400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.8189050070191858,
|
| 741 |
+
"grad_norm": 80821.6640625,
|
| 742 |
+
"learning_rate": 9.05474964904071e-06,
|
| 743 |
+
"loss": 0.8344,
|
| 744 |
+
"step": 10500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.8267041023241304,
|
| 748 |
+
"grad_norm": 61162.19140625,
|
| 749 |
+
"learning_rate": 8.66479488379348e-06,
|
| 750 |
+
"loss": 0.8454,
|
| 751 |
+
"step": 10600
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.834503197629075,
|
| 755 |
+
"grad_norm": 64971.5390625,
|
| 756 |
+
"learning_rate": 8.274840118546249e-06,
|
| 757 |
+
"loss": 0.8247,
|
| 758 |
+
"step": 10700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.8423022929340197,
|
| 762 |
+
"grad_norm": 64816.95703125,
|
| 763 |
+
"learning_rate": 7.884885353299018e-06,
|
| 764 |
+
"loss": 0.8142,
|
| 765 |
+
"step": 10800
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.8501013882389643,
|
| 769 |
+
"grad_norm": 60450.6875,
|
| 770 |
+
"learning_rate": 7.494930588051786e-06,
|
| 771 |
+
"loss": 0.8128,
|
| 772 |
+
"step": 10900
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.8579004835439089,
|
| 776 |
+
"grad_norm": 71442.4375,
|
| 777 |
+
"learning_rate": 7.104975822804555e-06,
|
| 778 |
+
"loss": 0.8166,
|
| 779 |
+
"step": 11000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.8656995788488535,
|
| 783 |
+
"grad_norm": 76537.1875,
|
| 784 |
+
"learning_rate": 6.715021057557324e-06,
|
| 785 |
+
"loss": 0.852,
|
| 786 |
+
"step": 11100
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.8734986741537981,
|
| 790 |
+
"grad_norm": 65081.37890625,
|
| 791 |
+
"learning_rate": 6.325066292310092e-06,
|
| 792 |
+
"loss": 0.8394,
|
| 793 |
+
"step": 11200
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.8812977694587428,
|
| 797 |
+
"grad_norm": 73682.296875,
|
| 798 |
+
"learning_rate": 5.935111527062861e-06,
|
| 799 |
+
"loss": 0.8165,
|
| 800 |
+
"step": 11300
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.8890968647636874,
|
| 804 |
+
"grad_norm": 73644.78125,
|
| 805 |
+
"learning_rate": 5.54515676181563e-06,
|
| 806 |
+
"loss": 0.8312,
|
| 807 |
+
"step": 11400
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.896895960068632,
|
| 811 |
+
"grad_norm": 79621.890625,
|
| 812 |
+
"learning_rate": 5.155201996568398e-06,
|
| 813 |
+
"loss": 0.8404,
|
| 814 |
+
"step": 11500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.9046950553735766,
|
| 818 |
+
"grad_norm": 56045.7734375,
|
| 819 |
+
"learning_rate": 4.7652472313211666e-06,
|
| 820 |
+
"loss": 0.8295,
|
| 821 |
+
"step": 11600
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.9124941506785212,
|
| 825 |
+
"grad_norm": 66463.71875,
|
| 826 |
+
"learning_rate": 4.375292466073936e-06,
|
| 827 |
+
"loss": 0.8256,
|
| 828 |
+
"step": 11700
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.920293245983466,
|
| 832 |
+
"grad_norm": 58425.23046875,
|
| 833 |
+
"learning_rate": 3.985337700826705e-06,
|
| 834 |
+
"loss": 0.8253,
|
| 835 |
+
"step": 11800
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.9280923412884106,
|
| 839 |
+
"grad_norm": 68089.9609375,
|
| 840 |
+
"learning_rate": 3.595382935579473e-06,
|
| 841 |
+
"loss": 0.8161,
|
| 842 |
+
"step": 11900
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.9358914365933552,
|
| 846 |
+
"grad_norm": 63161.6796875,
|
| 847 |
+
"learning_rate": 3.2054281703322412e-06,
|
| 848 |
+
"loss": 0.8322,
|
| 849 |
+
"step": 12000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.9436905318982998,
|
| 853 |
+
"grad_norm": 56655.69921875,
|
| 854 |
+
"learning_rate": 2.8154734050850103e-06,
|
| 855 |
+
"loss": 0.8117,
|
| 856 |
+
"step": 12100
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.9514896272032444,
|
| 860 |
+
"grad_norm": 72667.96875,
|
| 861 |
+
"learning_rate": 2.425518639837779e-06,
|
| 862 |
+
"loss": 0.8181,
|
| 863 |
+
"step": 12200
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.9592887225081891,
|
| 867 |
+
"grad_norm": 64209.17578125,
|
| 868 |
+
"learning_rate": 2.0355638745905476e-06,
|
| 869 |
+
"loss": 0.813,
|
| 870 |
+
"step": 12300
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.9670878178131337,
|
| 874 |
+
"grad_norm": 68720.7109375,
|
| 875 |
+
"learning_rate": 1.6456091093433163e-06,
|
| 876 |
+
"loss": 0.8343,
|
| 877 |
+
"step": 12400
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.9748869131180783,
|
| 881 |
+
"grad_norm": 62702.68359375,
|
| 882 |
+
"learning_rate": 1.255654344096085e-06,
|
| 883 |
+
"loss": 0.8156,
|
| 884 |
+
"step": 12500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.9826860084230229,
|
| 888 |
+
"grad_norm": 62949.0625,
|
| 889 |
+
"learning_rate": 8.656995788488536e-07,
|
| 890 |
+
"loss": 0.8343,
|
| 891 |
+
"step": 12600
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.9904851037279676,
|
| 895 |
+
"grad_norm": 64140.0703125,
|
| 896 |
+
"learning_rate": 4.757448136016222e-07,
|
| 897 |
+
"loss": 0.8189,
|
| 898 |
+
"step": 12700
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.9982841990329122,
|
| 902 |
+
"grad_norm": 72402.578125,
|
| 903 |
+
"learning_rate": 8.579004835439089e-08,
|
| 904 |
+
"loss": 0.823,
|
| 905 |
+
"step": 12800
|
| 906 |
+
},
|
| 907 |
{
|
| 908 |
"epoch": 1.0,
|
| 909 |
+
"step": 12822,
|
| 910 |
+
"total_flos": 1.955920022807249e+18,
|
| 911 |
+
"train_loss": 0.8700304180128008,
|
| 912 |
+
"train_runtime": 179145.9612,
|
| 913 |
+
"train_samples_per_second": 0.859,
|
| 914 |
"train_steps_per_second": 0.072
|
| 915 |
}
|
| 916 |
],
|
| 917 |
"logging_steps": 100,
|
| 918 |
+
"max_steps": 12822,
|
| 919 |
"num_input_tokens_seen": 0,
|
| 920 |
"num_train_epochs": 1,
|
| 921 |
+
"save_steps": 12822,
|
| 922 |
"stateful_callbacks": {
|
| 923 |
"TrainerControl": {
|
| 924 |
"args": {
|
|
|
|
| 931 |
"attributes": {}
|
| 932 |
}
|
| 933 |
},
|
| 934 |
+
"total_flos": 1.955920022807249e+18,
|
| 935 |
"train_batch_size": 12,
|
| 936 |
"trial_name": null,
|
| 937 |
"trial_params": null
|