Model save
Browse files- README.md +3 -7
- all_results.json +4 -4
- lora_0/adapter_config.json +4 -4
- lora_0/adapter_model.safetensors +1 -1
- train_results.json +4 -4
- trainer_state.json +959 -959
README.md
CHANGED
|
@@ -2,16 +2,12 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
-
- alignment-handbook
|
| 6 |
-
- trl
|
| 7 |
-
- sft
|
| 8 |
-
- generated_from_trainer
|
| 9 |
- trl
|
| 10 |
- sft
|
| 11 |
- alignment-handbook
|
| 12 |
- generated_from_trainer
|
| 13 |
datasets:
|
| 14 |
-
-
|
| 15 |
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 16 |
model-index:
|
| 17 |
- name: zephyr-7b-pl-qlora
|
|
@@ -23,7 +19,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 23 |
|
| 24 |
# zephyr-7b-pl-qlora
|
| 25 |
|
| 26 |
-
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
@@ -42,7 +38,7 @@ More information needed
|
|
| 42 |
### Training hyperparameters
|
| 43 |
|
| 44 |
The following hyperparameters were used during training:
|
| 45 |
-
- learning_rate:
|
| 46 |
- train_batch_size: 2
|
| 47 |
- eval_batch_size: 4
|
| 48 |
- seed: 42
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
- trl
|
| 6 |
- sft
|
| 7 |
- alignment-handbook
|
| 8 |
- generated_from_trainer
|
| 9 |
datasets:
|
| 10 |
+
- generator
|
| 11 |
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 12 |
model-index:
|
| 13 |
- name: zephyr-7b-pl-qlora
|
|
|
|
| 19 |
|
| 20 |
# zephyr-7b-pl-qlora
|
| 21 |
|
| 22 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the generator dataset.
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
|
|
|
| 38 |
### Training hyperparameters
|
| 39 |
|
| 40 |
The following hyperparameters were used during training:
|
| 41 |
+
- learning_rate: 1e-05
|
| 42 |
- train_batch_size: 2
|
| 43 |
- eval_batch_size: 4
|
| 44 |
- seed: 42
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 0.28,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 207865,
|
| 6 |
-
"train_samples_per_second": 1.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 0.28,
|
| 3 |
+
"train_loss": 0.025274057275091764,
|
| 4 |
+
"train_runtime": 26505.156,
|
| 5 |
"train_samples": 207865,
|
| 6 |
+
"train_samples_per_second": 1.449,
|
| 7 |
+
"train_steps_per_second": 0.091
|
| 8 |
}
|
lora_0/adapter_config.json
CHANGED
|
@@ -25,11 +25,11 @@
|
|
| 25 |
"target_modules": [
|
| 26 |
"o_proj",
|
| 27 |
"q_proj",
|
| 28 |
-
"k_proj",
|
| 29 |
-
"v_proj",
|
| 30 |
-
"gate_proj",
|
| 31 |
"down_proj",
|
| 32 |
-
"up_proj"
|
|
|
|
|
|
|
|
|
|
| 33 |
],
|
| 34 |
"task_type": null,
|
| 35 |
"use_dora": false,
|
|
|
|
| 25 |
"target_modules": [
|
| 26 |
"o_proj",
|
| 27 |
"q_proj",
|
|
|
|
|
|
|
|
|
|
| 28 |
"down_proj",
|
| 29 |
+
"up_proj",
|
| 30 |
+
"v_proj",
|
| 31 |
+
"k_proj",
|
| 32 |
+
"gate_proj"
|
| 33 |
],
|
| 34 |
"task_type": null,
|
| 35 |
"use_dora": false,
|
lora_0/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167832240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:608b46e495c8a7c3e5b9167b9dccb51eaba448f24ce453cb7cffe1af4b5d3067
|
| 3 |
size 167832240
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 0.28,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 207865,
|
| 6 |
-
"train_samples_per_second": 1.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 0.28,
|
| 3 |
+
"train_loss": 0.025274057275091764,
|
| 4 |
+
"train_runtime": 26505.156,
|
| 5 |
"train_samples": 207865,
|
| 6 |
+
"train_samples_per_second": 1.449,
|
| 7 |
+
"train_steps_per_second": 0.091
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -10,2898 +10,2898 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
-
"learning_rate":
|
| 14 |
"loss": 0.6931,
|
| 15 |
"step": 1
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.0,
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss": 0.
|
| 21 |
"step": 5
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.0,
|
| 25 |
-
"learning_rate":
|
| 26 |
-
"loss": 0.
|
| 27 |
"step": 10
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.0,
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 0.
|
| 33 |
"step": 15
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.0,
|
| 37 |
-
"learning_rate":
|
| 38 |
-
"loss": 0.
|
| 39 |
"step": 20
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.0,
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss": 0.
|
| 45 |
"step": 25
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.0,
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 30
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.0,
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss": 0.
|
| 57 |
"step": 35
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.0,
|
| 61 |
-
"learning_rate":
|
| 62 |
-
"loss": 0.
|
| 63 |
"step": 40
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.01,
|
| 67 |
-
"learning_rate":
|
| 68 |
-
"loss": 0.
|
| 69 |
"step": 45
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.01,
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 50
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.01,
|
| 79 |
-
"learning_rate":
|
| 80 |
-
"loss": 0.
|
| 81 |
"step": 55
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.01,
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 60
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.01,
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 65
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.01,
|
| 97 |
-
"learning_rate":
|
| 98 |
-
"loss": 0.
|
| 99 |
"step": 70
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.01,
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 75
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.01,
|
| 109 |
-
"learning_rate":
|
| 110 |
-
"loss": 0.
|
| 111 |
"step": 80
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.01,
|
| 115 |
-
"learning_rate":
|
| 116 |
-
"loss": 0.
|
| 117 |
"step": 85
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.01,
|
| 121 |
-
"learning_rate":
|
| 122 |
-
"loss": 0.
|
| 123 |
"step": 90
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.01,
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 95
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.01,
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 100
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.01,
|
| 139 |
-
"learning_rate":
|
| 140 |
-
"loss": 0.
|
| 141 |
"step": 105
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.01,
|
| 145 |
-
"learning_rate":
|
| 146 |
-
"loss": 0.
|
| 147 |
"step": 110
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.01,
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss": 0.
|
| 153 |
"step": 115
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.01,
|
| 157 |
-
"learning_rate":
|
| 158 |
-
"loss": 0.
|
| 159 |
"step": 120
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.01,
|
| 163 |
-
"learning_rate":
|
| 164 |
-
"loss": 0.
|
| 165 |
"step": 125
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.01,
|
| 169 |
-
"learning_rate":
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 130
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.02,
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 135
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.02,
|
| 181 |
-
"learning_rate":
|
| 182 |
-
"loss": 0.
|
| 183 |
"step": 140
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.02,
|
| 187 |
-
"learning_rate":
|
| 188 |
-
"loss": 0.
|
| 189 |
"step": 145
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.02,
|
| 193 |
-
"learning_rate":
|
| 194 |
-
"loss": 0.
|
| 195 |
"step": 150
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.02,
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss": 0.
|
| 201 |
"step": 155
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.02,
|
| 205 |
-
"learning_rate":
|
| 206 |
-
"loss": 0.
|
| 207 |
"step": 160
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.02,
|
| 211 |
-
"learning_rate":
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 165
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.02,
|
| 217 |
-
"learning_rate":
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 170
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.02,
|
| 223 |
-
"learning_rate":
|
| 224 |
-
"loss": 0.
|
| 225 |
"step": 175
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.02,
|
| 229 |
-
"learning_rate":
|
| 230 |
-
"loss": 0.
|
| 231 |
"step": 180
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.02,
|
| 235 |
-
"learning_rate":
|
| 236 |
-
"loss": 0.
|
| 237 |
"step": 185
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.02,
|
| 241 |
-
"learning_rate":
|
| 242 |
-
"loss": 0.
|
| 243 |
"step": 190
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.02,
|
| 247 |
-
"learning_rate":
|
| 248 |
-
"loss": 0.
|
| 249 |
"step": 195
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.02,
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 200
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.02,
|
| 259 |
-
"learning_rate":
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 205
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.02,
|
| 265 |
-
"learning_rate":
|
| 266 |
-
"loss": 0.
|
| 267 |
"step": 210
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.02,
|
| 271 |
-
"learning_rate":
|
| 272 |
-
"loss": 0.
|
| 273 |
"step": 215
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.03,
|
| 277 |
-
"learning_rate":
|
| 278 |
-
"loss": 0.
|
| 279 |
"step": 220
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.03,
|
| 283 |
-
"learning_rate":
|
| 284 |
-
"loss": 0.
|
| 285 |
"step": 225
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 0.03,
|
| 289 |
-
"learning_rate":
|
| 290 |
-
"loss": 0.
|
| 291 |
"step": 230
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 0.03,
|
| 295 |
-
"learning_rate":
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 235
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.03,
|
| 301 |
-
"learning_rate":
|
| 302 |
-
"loss": 0.
|
| 303 |
"step": 240
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 0.03,
|
| 307 |
-
"learning_rate":
|
| 308 |
-
"loss": 0.
|
| 309 |
"step": 245
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.03,
|
| 313 |
-
"learning_rate":
|
| 314 |
-
"loss": 0.
|
| 315 |
"step": 250
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 0.03,
|
| 319 |
-
"learning_rate":
|
| 320 |
-
"loss": 0.
|
| 321 |
"step": 255
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 0.03,
|
| 325 |
-
"learning_rate":
|
| 326 |
-
"loss": 0.
|
| 327 |
"step": 260
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 0.03,
|
| 331 |
-
"learning_rate":
|
| 332 |
-
"loss": 0.
|
| 333 |
"step": 265
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.03,
|
| 337 |
-
"learning_rate":
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 270
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.03,
|
| 343 |
-
"learning_rate":
|
| 344 |
-
"loss": 0.
|
| 345 |
"step": 275
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 0.03,
|
| 349 |
-
"learning_rate":
|
| 350 |
-
"loss": 0.
|
| 351 |
"step": 280
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 0.03,
|
| 355 |
-
"learning_rate":
|
| 356 |
-
"loss": 0.
|
| 357 |
"step": 285
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 0.03,
|
| 361 |
-
"learning_rate":
|
| 362 |
-
"loss": 0.
|
| 363 |
"step": 290
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 0.03,
|
| 367 |
-
"learning_rate":
|
| 368 |
-
"loss": 0.
|
| 369 |
"step": 295
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.03,
|
| 373 |
-
"learning_rate":
|
| 374 |
-
"loss": 0.
|
| 375 |
"step": 300
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.03,
|
| 379 |
-
"learning_rate":
|
| 380 |
-
"loss": 0.
|
| 381 |
"step": 305
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.04,
|
| 385 |
-
"learning_rate":
|
| 386 |
-
"loss": 0.
|
| 387 |
"step": 310
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 0.04,
|
| 391 |
-
"learning_rate":
|
| 392 |
-
"loss": 0.
|
| 393 |
"step": 315
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 0.04,
|
| 397 |
-
"learning_rate":
|
| 398 |
-
"loss": 0.
|
| 399 |
"step": 320
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.04,
|
| 403 |
-
"learning_rate":
|
| 404 |
-
"loss": 0.
|
| 405 |
"step": 325
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 0.04,
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"loss": 0.
|
| 411 |
"step": 330
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 0.04,
|
| 415 |
-
"learning_rate":
|
| 416 |
-
"loss": 0.
|
| 417 |
"step": 335
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.04,
|
| 421 |
-
"learning_rate":
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 340
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.04,
|
| 427 |
-
"learning_rate":
|
| 428 |
-
"loss": 0.
|
| 429 |
"step": 345
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.04,
|
| 433 |
-
"learning_rate":
|
| 434 |
-
"loss": 0.
|
| 435 |
"step": 350
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 0.04,
|
| 439 |
-
"learning_rate":
|
| 440 |
-
"loss": 0.
|
| 441 |
"step": 355
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 0.04,
|
| 445 |
-
"learning_rate":
|
| 446 |
-
"loss": 0.
|
| 447 |
"step": 360
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 0.04,
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss": 0.
|
| 453 |
"step": 365
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.04,
|
| 457 |
-
"learning_rate":
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 370
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.04,
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 375
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.04,
|
| 469 |
-
"learning_rate":
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 380
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 0.04,
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss": 0.
|
| 477 |
"step": 385
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 0.04,
|
| 481 |
-
"learning_rate":
|
| 482 |
-
"loss": 0.
|
| 483 |
"step": 390
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 0.05,
|
| 487 |
-
"learning_rate":
|
| 488 |
-
"loss": 0.
|
| 489 |
"step": 395
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.05,
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss": 0.
|
| 495 |
"step": 400
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 0.05,
|
| 499 |
-
"learning_rate":
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 405
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.05,
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 410
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.05,
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss": 0.
|
| 513 |
"step": 415
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 0.05,
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss": 0.
|
| 519 |
"step": 420
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.05,
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"loss": 0.
|
| 525 |
"step": 425
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 0.05,
|
| 529 |
-
"learning_rate":
|
| 530 |
-
"loss": 0.
|
| 531 |
"step": 430
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 0.05,
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss": 0.
|
| 537 |
"step": 435
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.05,
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 440
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.05,
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 445
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.05,
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 450
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.05,
|
| 559 |
-
"learning_rate":
|
| 560 |
-
"loss": 0.
|
| 561 |
"step": 455
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.05,
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"loss": 0.
|
| 567 |
"step": 460
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.05,
|
| 571 |
-
"learning_rate":
|
| 572 |
-
"loss": 0.
|
| 573 |
"step": 465
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 0.05,
|
| 577 |
-
"learning_rate":
|
| 578 |
-
"loss": 0.
|
| 579 |
"step": 470
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.05,
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss": 0.
|
| 585 |
"step": 475
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 0.06,
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 480
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.06,
|
| 595 |
-
"learning_rate":
|
| 596 |
-
"loss": 0.
|
| 597 |
"step": 485
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.06,
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"loss": 0.
|
| 603 |
"step": 490
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 0.06,
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss": 0.
|
| 609 |
"step": 495
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.06,
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 0.
|
| 615 |
"step": 500
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.06,
|
| 619 |
-
"learning_rate":
|
| 620 |
-
"loss": 0.
|
| 621 |
"step": 505
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 0.06,
|
| 625 |
-
"learning_rate":
|
| 626 |
-
"loss": 0.
|
| 627 |
"step": 510
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.06,
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 515
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.06,
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 520
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.06,
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"loss": 0.
|
| 645 |
"step": 525
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 0.06,
|
| 649 |
-
"learning_rate":
|
| 650 |
-
"loss": 0.
|
| 651 |
"step": 530
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 0.06,
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 535
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.06,
|
| 661 |
-
"learning_rate":
|
| 662 |
-
"loss": 0.
|
| 663 |
"step": 540
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 0.06,
|
| 667 |
-
"learning_rate":
|
| 668 |
-
"loss": 0.
|
| 669 |
"step": 545
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.06,
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 550
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.06,
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss": 0.
|
| 681 |
"step": 555
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 0.06,
|
| 685 |
-
"learning_rate":
|
| 686 |
-
"loss": 0.
|
| 687 |
"step": 560
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 0.06,
|
| 691 |
-
"learning_rate":
|
| 692 |
-
"loss": 0.
|
| 693 |
"step": 565
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 0.07,
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss": 0.
|
| 699 |
"step": 570
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.07,
|
| 703 |
-
"learning_rate":
|
| 704 |
-
"loss": 0.
|
| 705 |
"step": 575
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 0.07,
|
| 709 |
-
"learning_rate":
|
| 710 |
-
"loss": 0.
|
| 711 |
"step": 580
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 0.07,
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 585
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.07,
|
| 721 |
-
"learning_rate":
|
| 722 |
-
"loss": 0.
|
| 723 |
"step": 590
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 0.07,
|
| 727 |
-
"learning_rate":
|
| 728 |
-
"loss": 0.
|
| 729 |
"step": 595
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.07,
|
| 733 |
-
"learning_rate":
|
| 734 |
-
"loss": 0.
|
| 735 |
"step": 600
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 0.07,
|
| 739 |
-
"learning_rate":
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 605
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.07,
|
| 745 |
-
"learning_rate":
|
| 746 |
-
"loss": 0.
|
| 747 |
"step": 610
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 0.07,
|
| 751 |
-
"learning_rate":
|
| 752 |
-
"loss": 0.
|
| 753 |
"step": 615
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 0.07,
|
| 757 |
-
"learning_rate":
|
| 758 |
-
"loss": 0.
|
| 759 |
"step": 620
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.07,
|
| 763 |
-
"learning_rate":
|
| 764 |
-
"loss": 0.
|
| 765 |
"step": 625
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 0.07,
|
| 769 |
-
"learning_rate":
|
| 770 |
-
"loss": 0.
|
| 771 |
"step": 630
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 0.07,
|
| 775 |
-
"learning_rate":
|
| 776 |
-
"loss": 0.
|
| 777 |
"step": 635
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 0.07,
|
| 781 |
-
"learning_rate":
|
| 782 |
-
"loss": 0.
|
| 783 |
"step": 640
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.07,
|
| 787 |
-
"learning_rate":
|
| 788 |
-
"loss": 0.
|
| 789 |
"step": 645
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.07,
|
| 793 |
-
"learning_rate":
|
| 794 |
-
"loss": 0.
|
| 795 |
"step": 650
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.08,
|
| 799 |
-
"learning_rate":
|
| 800 |
-
"loss": 0.
|
| 801 |
"step": 655
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 0.08,
|
| 805 |
-
"learning_rate":
|
| 806 |
-
"loss": 0.
|
| 807 |
"step": 660
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 0.08,
|
| 811 |
-
"learning_rate":
|
| 812 |
-
"loss": 0.
|
| 813 |
"step": 665
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 0.08,
|
| 817 |
-
"learning_rate":
|
| 818 |
-
"loss": 0.
|
| 819 |
"step": 670
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 0.08,
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss": 0.
|
| 825 |
"step": 675
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 0.08,
|
| 829 |
-
"learning_rate":
|
| 830 |
-
"loss": 0.
|
| 831 |
"step": 680
|
| 832 |
},
|
| 833 |
{
|
| 834 |
"epoch": 0.08,
|
| 835 |
-
"learning_rate":
|
| 836 |
-
"loss": 0.
|
| 837 |
"step": 685
|
| 838 |
},
|
| 839 |
{
|
| 840 |
"epoch": 0.08,
|
| 841 |
-
"learning_rate":
|
| 842 |
-
"loss": 0.
|
| 843 |
"step": 690
|
| 844 |
},
|
| 845 |
{
|
| 846 |
"epoch": 0.08,
|
| 847 |
-
"learning_rate":
|
| 848 |
-
"loss": 0.
|
| 849 |
"step": 695
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 0.08,
|
| 853 |
-
"learning_rate":
|
| 854 |
-
"loss": 0.
|
| 855 |
"step": 700
|
| 856 |
},
|
| 857 |
{
|
| 858 |
"epoch": 0.08,
|
| 859 |
-
"learning_rate":
|
| 860 |
-
"loss": 0.
|
| 861 |
"step": 705
|
| 862 |
},
|
| 863 |
{
|
| 864 |
"epoch": 0.08,
|
| 865 |
-
"learning_rate":
|
| 866 |
-
"loss": 0.
|
| 867 |
"step": 710
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 0.08,
|
| 871 |
-
"learning_rate":
|
| 872 |
-
"loss": 0.
|
| 873 |
"step": 715
|
| 874 |
},
|
| 875 |
{
|
| 876 |
"epoch": 0.08,
|
| 877 |
-
"learning_rate":
|
| 878 |
-
"loss": 0.
|
| 879 |
"step": 720
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 0.08,
|
| 883 |
-
"learning_rate":
|
| 884 |
-
"loss": 0.
|
| 885 |
"step": 725
|
| 886 |
},
|
| 887 |
{
|
| 888 |
"epoch": 0.08,
|
| 889 |
-
"learning_rate":
|
| 890 |
-
"loss": 0.
|
| 891 |
"step": 730
|
| 892 |
},
|
| 893 |
{
|
| 894 |
"epoch": 0.08,
|
| 895 |
-
"learning_rate":
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 735
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 0.08,
|
| 901 |
-
"learning_rate":
|
| 902 |
-
"loss": 0.
|
| 903 |
"step": 740
|
| 904 |
},
|
| 905 |
{
|
| 906 |
"epoch": 0.09,
|
| 907 |
-
"learning_rate":
|
| 908 |
-
"loss": 0.
|
| 909 |
"step": 745
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 0.09,
|
| 913 |
-
"learning_rate":
|
| 914 |
-
"loss": 0.
|
| 915 |
"step": 750
|
| 916 |
},
|
| 917 |
{
|
| 918 |
"epoch": 0.09,
|
| 919 |
-
"learning_rate":
|
| 920 |
-
"loss": 0.
|
| 921 |
"step": 755
|
| 922 |
},
|
| 923 |
{
|
| 924 |
"epoch": 0.09,
|
| 925 |
-
"learning_rate":
|
| 926 |
-
"loss": 0.
|
| 927 |
"step": 760
|
| 928 |
},
|
| 929 |
{
|
| 930 |
"epoch": 0.09,
|
| 931 |
-
"learning_rate":
|
| 932 |
-
"loss": 0.
|
| 933 |
"step": 765
|
| 934 |
},
|
| 935 |
{
|
| 936 |
"epoch": 0.09,
|
| 937 |
-
"learning_rate":
|
| 938 |
-
"loss": 0.
|
| 939 |
"step": 770
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.09,
|
| 943 |
-
"learning_rate":
|
| 944 |
-
"loss": 0.
|
| 945 |
"step": 775
|
| 946 |
},
|
| 947 |
{
|
| 948 |
"epoch": 0.09,
|
| 949 |
-
"learning_rate":
|
| 950 |
-
"loss": 0.
|
| 951 |
"step": 780
|
| 952 |
},
|
| 953 |
{
|
| 954 |
"epoch": 0.09,
|
| 955 |
-
"learning_rate":
|
| 956 |
-
"loss": 0.
|
| 957 |
"step": 785
|
| 958 |
},
|
| 959 |
{
|
| 960 |
"epoch": 0.09,
|
| 961 |
-
"learning_rate":
|
| 962 |
-
"loss": 0.
|
| 963 |
"step": 790
|
| 964 |
},
|
| 965 |
{
|
| 966 |
"epoch": 0.09,
|
| 967 |
-
"learning_rate":
|
| 968 |
-
"loss": 0.
|
| 969 |
"step": 795
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 0.09,
|
| 973 |
-
"learning_rate":
|
| 974 |
-
"loss": 0.
|
| 975 |
"step": 800
|
| 976 |
},
|
| 977 |
{
|
| 978 |
"epoch": 0.09,
|
| 979 |
-
"learning_rate":
|
| 980 |
-
"loss": 0.
|
| 981 |
"step": 805
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 0.09,
|
| 985 |
-
"learning_rate":
|
| 986 |
-
"loss": 0.
|
| 987 |
"step": 810
|
| 988 |
},
|
| 989 |
{
|
| 990 |
"epoch": 0.09,
|
| 991 |
-
"learning_rate":
|
| 992 |
-
"loss": 0.
|
| 993 |
"step": 815
|
| 994 |
},
|
| 995 |
{
|
| 996 |
"epoch": 0.09,
|
| 997 |
-
"learning_rate":
|
| 998 |
-
"loss": 0.
|
| 999 |
"step": 820
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 0.09,
|
| 1003 |
-
"learning_rate":
|
| 1004 |
-
"loss": 0.
|
| 1005 |
"step": 825
|
| 1006 |
},
|
| 1007 |
{
|
| 1008 |
"epoch": 0.1,
|
| 1009 |
-
"learning_rate":
|
| 1010 |
-
"loss": 0.
|
| 1011 |
"step": 830
|
| 1012 |
},
|
| 1013 |
{
|
| 1014 |
"epoch": 0.1,
|
| 1015 |
-
"learning_rate":
|
| 1016 |
-
"loss": 0.
|
| 1017 |
"step": 835
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
"epoch": 0.1,
|
| 1021 |
-
"learning_rate":
|
| 1022 |
-
"loss": 0.
|
| 1023 |
"step": 840
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 0.1,
|
| 1027 |
-
"learning_rate":
|
| 1028 |
-
"loss": 0.
|
| 1029 |
"step": 845
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 0.1,
|
| 1033 |
-
"learning_rate":
|
| 1034 |
-
"loss": 0.
|
| 1035 |
"step": 850
|
| 1036 |
},
|
| 1037 |
{
|
| 1038 |
"epoch": 0.1,
|
| 1039 |
-
"learning_rate":
|
| 1040 |
-
"loss": 0.
|
| 1041 |
"step": 855
|
| 1042 |
},
|
| 1043 |
{
|
| 1044 |
"epoch": 0.1,
|
| 1045 |
-
"learning_rate":
|
| 1046 |
-
"loss": 0.
|
| 1047 |
"step": 860
|
| 1048 |
},
|
| 1049 |
{
|
| 1050 |
"epoch": 0.1,
|
| 1051 |
-
"learning_rate":
|
| 1052 |
-
"loss": 0.
|
| 1053 |
"step": 865
|
| 1054 |
},
|
| 1055 |
{
|
| 1056 |
"epoch": 0.1,
|
| 1057 |
-
"learning_rate":
|
| 1058 |
-
"loss": 0.
|
| 1059 |
"step": 870
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 0.1,
|
| 1063 |
-
"learning_rate":
|
| 1064 |
-
"loss": 0.
|
| 1065 |
"step": 875
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.1,
|
| 1069 |
-
"learning_rate":
|
| 1070 |
-
"loss": 0.
|
| 1071 |
"step": 880
|
| 1072 |
},
|
| 1073 |
{
|
| 1074 |
"epoch": 0.1,
|
| 1075 |
-
"learning_rate":
|
| 1076 |
-
"loss": 0.
|
| 1077 |
"step": 885
|
| 1078 |
},
|
| 1079 |
{
|
| 1080 |
"epoch": 0.1,
|
| 1081 |
-
"learning_rate":
|
| 1082 |
-
"loss": 0.
|
| 1083 |
"step": 890
|
| 1084 |
},
|
| 1085 |
{
|
| 1086 |
"epoch": 0.1,
|
| 1087 |
-
"learning_rate":
|
| 1088 |
-
"loss": 0.
|
| 1089 |
"step": 895
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 0.1,
|
| 1093 |
-
"learning_rate":
|
| 1094 |
-
"loss": 0.
|
| 1095 |
"step": 900
|
| 1096 |
},
|
| 1097 |
{
|
| 1098 |
"epoch": 0.1,
|
| 1099 |
-
"learning_rate":
|
| 1100 |
-
"loss": 0.
|
| 1101 |
"step": 905
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
"epoch": 0.1,
|
| 1105 |
-
"learning_rate":
|
| 1106 |
-
"loss": 0.
|
| 1107 |
"step": 910
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 0.1,
|
| 1111 |
-
"learning_rate":
|
| 1112 |
-
"loss": 0.
|
| 1113 |
"step": 915
|
| 1114 |
},
|
| 1115 |
{
|
| 1116 |
"epoch": 0.11,
|
| 1117 |
-
"learning_rate":
|
| 1118 |
-
"loss": 0.
|
| 1119 |
"step": 920
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 0.11,
|
| 1123 |
-
"learning_rate":
|
| 1124 |
-
"loss": 0.
|
| 1125 |
"step": 925
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
"epoch": 0.11,
|
| 1129 |
-
"learning_rate":
|
| 1130 |
-
"loss": 0.
|
| 1131 |
"step": 930
|
| 1132 |
},
|
| 1133 |
{
|
| 1134 |
"epoch": 0.11,
|
| 1135 |
-
"learning_rate":
|
| 1136 |
-
"loss": 0.
|
| 1137 |
"step": 935
|
| 1138 |
},
|
| 1139 |
{
|
| 1140 |
"epoch": 0.11,
|
| 1141 |
-
"learning_rate":
|
| 1142 |
-
"loss": 0.
|
| 1143 |
"step": 940
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
"epoch": 0.11,
|
| 1147 |
-
"learning_rate":
|
| 1148 |
-
"loss": 0.
|
| 1149 |
"step": 945
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.11,
|
| 1153 |
-
"learning_rate":
|
| 1154 |
-
"loss": 0.
|
| 1155 |
"step": 950
|
| 1156 |
},
|
| 1157 |
{
|
| 1158 |
"epoch": 0.11,
|
| 1159 |
-
"learning_rate":
|
| 1160 |
-
"loss": 0.
|
| 1161 |
"step": 955
|
| 1162 |
},
|
| 1163 |
{
|
| 1164 |
"epoch": 0.11,
|
| 1165 |
-
"learning_rate":
|
| 1166 |
-
"loss": 0.
|
| 1167 |
"step": 960
|
| 1168 |
},
|
| 1169 |
{
|
| 1170 |
"epoch": 0.11,
|
| 1171 |
-
"learning_rate":
|
| 1172 |
-
"loss": 0.
|
| 1173 |
"step": 965
|
| 1174 |
},
|
| 1175 |
{
|
| 1176 |
"epoch": 0.11,
|
| 1177 |
-
"learning_rate":
|
| 1178 |
-
"loss": 0.
|
| 1179 |
"step": 970
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.11,
|
| 1183 |
-
"learning_rate":
|
| 1184 |
-
"loss": 0.
|
| 1185 |
"step": 975
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
"epoch": 0.11,
|
| 1189 |
-
"learning_rate":
|
| 1190 |
-
"loss": 0.
|
| 1191 |
"step": 980
|
| 1192 |
},
|
| 1193 |
{
|
| 1194 |
"epoch": 0.11,
|
| 1195 |
-
"learning_rate":
|
| 1196 |
-
"loss": 0.
|
| 1197 |
"step": 985
|
| 1198 |
},
|
| 1199 |
{
|
| 1200 |
"epoch": 0.11,
|
| 1201 |
-
"learning_rate":
|
| 1202 |
-
"loss": 0.
|
| 1203 |
"step": 990
|
| 1204 |
},
|
| 1205 |
{
|
| 1206 |
"epoch": 0.11,
|
| 1207 |
-
"learning_rate":
|
| 1208 |
-
"loss": 0.
|
| 1209 |
"step": 995
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 0.11,
|
| 1213 |
-
"learning_rate":
|
| 1214 |
-
"loss": 0.
|
| 1215 |
"step": 1000
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
"epoch": 0.12,
|
| 1219 |
-
"learning_rate":
|
| 1220 |
-
"loss": 0.
|
| 1221 |
"step": 1005
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 0.12,
|
| 1225 |
-
"learning_rate":
|
| 1226 |
-
"loss": 0.
|
| 1227 |
"step": 1010
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
"epoch": 0.12,
|
| 1231 |
-
"learning_rate":
|
| 1232 |
-
"loss": 0.
|
| 1233 |
"step": 1015
|
| 1234 |
},
|
| 1235 |
{
|
| 1236 |
"epoch": 0.12,
|
| 1237 |
-
"learning_rate":
|
| 1238 |
-
"loss": 0.
|
| 1239 |
"step": 1020
|
| 1240 |
},
|
| 1241 |
{
|
| 1242 |
"epoch": 0.12,
|
| 1243 |
-
"learning_rate":
|
| 1244 |
-
"loss": 0.
|
| 1245 |
"step": 1025
|
| 1246 |
},
|
| 1247 |
{
|
| 1248 |
"epoch": 0.12,
|
| 1249 |
-
"learning_rate":
|
| 1250 |
-
"loss": 0.
|
| 1251 |
"step": 1030
|
| 1252 |
},
|
| 1253 |
{
|
| 1254 |
"epoch": 0.12,
|
| 1255 |
-
"learning_rate":
|
| 1256 |
-
"loss": 0.
|
| 1257 |
"step": 1035
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
"epoch": 0.12,
|
| 1261 |
-
"learning_rate":
|
| 1262 |
-
"loss": 0.
|
| 1263 |
"step": 1040
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 0.12,
|
| 1267 |
-
"learning_rate":
|
| 1268 |
-
"loss": 0.
|
| 1269 |
"step": 1045
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
"epoch": 0.12,
|
| 1273 |
-
"learning_rate":
|
| 1274 |
-
"loss": 0.
|
| 1275 |
"step": 1050
|
| 1276 |
},
|
| 1277 |
{
|
| 1278 |
"epoch": 0.12,
|
| 1279 |
-
"learning_rate":
|
| 1280 |
-
"loss": 0.
|
| 1281 |
"step": 1055
|
| 1282 |
},
|
| 1283 |
{
|
| 1284 |
"epoch": 0.12,
|
| 1285 |
-
"learning_rate":
|
| 1286 |
-
"loss": 0.
|
| 1287 |
"step": 1060
|
| 1288 |
},
|
| 1289 |
{
|
| 1290 |
"epoch": 0.12,
|
| 1291 |
-
"learning_rate":
|
| 1292 |
-
"loss": 0.
|
| 1293 |
"step": 1065
|
| 1294 |
},
|
| 1295 |
{
|
| 1296 |
"epoch": 0.12,
|
| 1297 |
-
"learning_rate":
|
| 1298 |
-
"loss": 0.
|
| 1299 |
"step": 1070
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
"epoch": 0.12,
|
| 1303 |
-
"learning_rate":
|
| 1304 |
-
"loss": 0.
|
| 1305 |
"step": 1075
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 0.12,
|
| 1309 |
-
"learning_rate":
|
| 1310 |
-
"loss": 0.
|
| 1311 |
"step": 1080
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
"epoch": 0.12,
|
| 1315 |
-
"learning_rate":
|
| 1316 |
-
"loss": 0.
|
| 1317 |
"step": 1085
|
| 1318 |
},
|
| 1319 |
{
|
| 1320 |
"epoch": 0.13,
|
| 1321 |
-
"learning_rate":
|
| 1322 |
-
"loss": 0.
|
| 1323 |
"step": 1090
|
| 1324 |
},
|
| 1325 |
{
|
| 1326 |
"epoch": 0.13,
|
| 1327 |
-
"learning_rate":
|
| 1328 |
-
"loss": 0.
|
| 1329 |
"step": 1095
|
| 1330 |
},
|
| 1331 |
{
|
| 1332 |
"epoch": 0.13,
|
| 1333 |
-
"learning_rate":
|
| 1334 |
-
"loss": 0.
|
| 1335 |
"step": 1100
|
| 1336 |
},
|
| 1337 |
{
|
| 1338 |
"epoch": 0.13,
|
| 1339 |
-
"learning_rate":
|
| 1340 |
-
"loss": 0.
|
| 1341 |
"step": 1105
|
| 1342 |
},
|
| 1343 |
{
|
| 1344 |
"epoch": 0.13,
|
| 1345 |
-
"learning_rate":
|
| 1346 |
-
"loss": 0.
|
| 1347 |
"step": 1110
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 0.13,
|
| 1351 |
-
"learning_rate":
|
| 1352 |
-
"loss": 0.
|
| 1353 |
"step": 1115
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
"epoch": 0.13,
|
| 1357 |
-
"learning_rate":
|
| 1358 |
-
"loss": 0.
|
| 1359 |
"step": 1120
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
"epoch": 0.13,
|
| 1363 |
-
"learning_rate":
|
| 1364 |
-
"loss": 0.
|
| 1365 |
"step": 1125
|
| 1366 |
},
|
| 1367 |
{
|
| 1368 |
"epoch": 0.13,
|
| 1369 |
-
"learning_rate":
|
| 1370 |
-
"loss": 0.
|
| 1371 |
"step": 1130
|
| 1372 |
},
|
| 1373 |
{
|
| 1374 |
"epoch": 0.13,
|
| 1375 |
-
"learning_rate":
|
| 1376 |
-
"loss": 0.
|
| 1377 |
"step": 1135
|
| 1378 |
},
|
| 1379 |
{
|
| 1380 |
"epoch": 0.13,
|
| 1381 |
-
"learning_rate":
|
| 1382 |
-
"loss": 0.
|
| 1383 |
"step": 1140
|
| 1384 |
},
|
| 1385 |
{
|
| 1386 |
"epoch": 0.13,
|
| 1387 |
-
"learning_rate":
|
| 1388 |
-
"loss": 0.
|
| 1389 |
"step": 1145
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 0.13,
|
| 1393 |
-
"learning_rate":
|
| 1394 |
-
"loss": 0.
|
| 1395 |
"step": 1150
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
"epoch": 0.13,
|
| 1399 |
-
"learning_rate":
|
| 1400 |
-
"loss": 0.
|
| 1401 |
"step": 1155
|
| 1402 |
},
|
| 1403 |
{
|
| 1404 |
"epoch": 0.13,
|
| 1405 |
-
"learning_rate":
|
| 1406 |
-
"loss": 0.
|
| 1407 |
"step": 1160
|
| 1408 |
},
|
| 1409 |
{
|
| 1410 |
"epoch": 0.13,
|
| 1411 |
-
"learning_rate":
|
| 1412 |
-
"loss": 0.
|
| 1413 |
"step": 1165
|
| 1414 |
},
|
| 1415 |
{
|
| 1416 |
"epoch": 0.13,
|
| 1417 |
-
"learning_rate":
|
| 1418 |
-
"loss": 0.
|
| 1419 |
"step": 1170
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
"epoch": 0.13,
|
| 1423 |
-
"learning_rate":
|
| 1424 |
-
"loss": 0.
|
| 1425 |
"step": 1175
|
| 1426 |
},
|
| 1427 |
{
|
| 1428 |
"epoch": 0.14,
|
| 1429 |
-
"learning_rate":
|
| 1430 |
-
"loss": 0.
|
| 1431 |
"step": 1180
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 0.14,
|
| 1435 |
-
"learning_rate":
|
| 1436 |
-
"loss": 0.
|
| 1437 |
"step": 1185
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
"epoch": 0.14,
|
| 1441 |
-
"learning_rate":
|
| 1442 |
-
"loss": 0.
|
| 1443 |
"step": 1190
|
| 1444 |
},
|
| 1445 |
{
|
| 1446 |
"epoch": 0.14,
|
| 1447 |
-
"learning_rate":
|
| 1448 |
-
"loss": 0.
|
| 1449 |
"step": 1195
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
"epoch": 0.14,
|
| 1453 |
-
"learning_rate":
|
| 1454 |
-
"loss": 0.
|
| 1455 |
"step": 1200
|
| 1456 |
},
|
| 1457 |
{
|
| 1458 |
"epoch": 0.14,
|
| 1459 |
-
"learning_rate":
|
| 1460 |
-
"loss": 0.
|
| 1461 |
"step": 1205
|
| 1462 |
},
|
| 1463 |
{
|
| 1464 |
"epoch": 0.14,
|
| 1465 |
-
"learning_rate":
|
| 1466 |
-
"loss": 0.
|
| 1467 |
"step": 1210
|
| 1468 |
},
|
| 1469 |
{
|
| 1470 |
"epoch": 0.14,
|
| 1471 |
-
"learning_rate":
|
| 1472 |
-
"loss": 0.
|
| 1473 |
"step": 1215
|
| 1474 |
},
|
| 1475 |
{
|
| 1476 |
"epoch": 0.14,
|
| 1477 |
-
"learning_rate":
|
| 1478 |
-
"loss": 0.
|
| 1479 |
"step": 1220
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
"epoch": 0.14,
|
| 1483 |
-
"learning_rate":
|
| 1484 |
-
"loss": 0.
|
| 1485 |
"step": 1225
|
| 1486 |
},
|
| 1487 |
{
|
| 1488 |
"epoch": 0.14,
|
| 1489 |
-
"learning_rate":
|
| 1490 |
-
"loss": 0.
|
| 1491 |
"step": 1230
|
| 1492 |
},
|
| 1493 |
{
|
| 1494 |
"epoch": 0.14,
|
| 1495 |
-
"learning_rate":
|
| 1496 |
-
"loss": 0.
|
| 1497 |
"step": 1235
|
| 1498 |
},
|
| 1499 |
{
|
| 1500 |
"epoch": 0.14,
|
| 1501 |
-
"learning_rate":
|
| 1502 |
-
"loss": 0.
|
| 1503 |
"step": 1240
|
| 1504 |
},
|
| 1505 |
{
|
| 1506 |
"epoch": 0.14,
|
| 1507 |
-
"learning_rate":
|
| 1508 |
-
"loss": 0.
|
| 1509 |
"step": 1245
|
| 1510 |
},
|
| 1511 |
{
|
| 1512 |
"epoch": 0.14,
|
| 1513 |
-
"learning_rate":
|
| 1514 |
-
"loss": 0.
|
| 1515 |
"step": 1250
|
| 1516 |
},
|
| 1517 |
{
|
| 1518 |
"epoch": 0.14,
|
| 1519 |
-
"learning_rate":
|
| 1520 |
-
"loss": 0.
|
| 1521 |
"step": 1255
|
| 1522 |
},
|
| 1523 |
{
|
| 1524 |
"epoch": 0.14,
|
| 1525 |
-
"learning_rate":
|
| 1526 |
-
"loss": 0.
|
| 1527 |
"step": 1260
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
"epoch": 0.15,
|
| 1531 |
-
"learning_rate":
|
| 1532 |
-
"loss": 0.
|
| 1533 |
"step": 1265
|
| 1534 |
},
|
| 1535 |
{
|
| 1536 |
"epoch": 0.15,
|
| 1537 |
-
"learning_rate":
|
| 1538 |
-
"loss": 0.
|
| 1539 |
"step": 1270
|
| 1540 |
},
|
| 1541 |
{
|
| 1542 |
"epoch": 0.15,
|
| 1543 |
-
"learning_rate":
|
| 1544 |
-
"loss": 0.
|
| 1545 |
"step": 1275
|
| 1546 |
},
|
| 1547 |
{
|
| 1548 |
"epoch": 0.15,
|
| 1549 |
-
"learning_rate":
|
| 1550 |
-
"loss": 0.
|
| 1551 |
"step": 1280
|
| 1552 |
},
|
| 1553 |
{
|
| 1554 |
"epoch": 0.15,
|
| 1555 |
-
"learning_rate":
|
| 1556 |
-
"loss": 0.
|
| 1557 |
"step": 1285
|
| 1558 |
},
|
| 1559 |
{
|
| 1560 |
"epoch": 0.15,
|
| 1561 |
-
"learning_rate":
|
| 1562 |
-
"loss": 0.
|
| 1563 |
"step": 1290
|
| 1564 |
},
|
| 1565 |
{
|
| 1566 |
"epoch": 0.15,
|
| 1567 |
-
"learning_rate":
|
| 1568 |
-
"loss": 0.
|
| 1569 |
"step": 1295
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
"epoch": 0.15,
|
| 1573 |
-
"learning_rate":
|
| 1574 |
-
"loss": 0.
|
| 1575 |
"step": 1300
|
| 1576 |
},
|
| 1577 |
{
|
| 1578 |
"epoch": 0.15,
|
| 1579 |
-
"learning_rate":
|
| 1580 |
-
"loss": 0.
|
| 1581 |
"step": 1305
|
| 1582 |
},
|
| 1583 |
{
|
| 1584 |
"epoch": 0.15,
|
| 1585 |
-
"learning_rate":
|
| 1586 |
-
"loss": 0.
|
| 1587 |
"step": 1310
|
| 1588 |
},
|
| 1589 |
{
|
| 1590 |
"epoch": 0.15,
|
| 1591 |
-
"learning_rate":
|
| 1592 |
-
"loss": 0.
|
| 1593 |
"step": 1315
|
| 1594 |
},
|
| 1595 |
{
|
| 1596 |
"epoch": 0.15,
|
| 1597 |
-
"learning_rate":
|
| 1598 |
-
"loss": 0.
|
| 1599 |
"step": 1320
|
| 1600 |
},
|
| 1601 |
{
|
| 1602 |
"epoch": 0.15,
|
| 1603 |
-
"learning_rate":
|
| 1604 |
-
"loss": 0.
|
| 1605 |
"step": 1325
|
| 1606 |
},
|
| 1607 |
{
|
| 1608 |
"epoch": 0.15,
|
| 1609 |
-
"learning_rate":
|
| 1610 |
-
"loss": 0.
|
| 1611 |
"step": 1330
|
| 1612 |
},
|
| 1613 |
{
|
| 1614 |
"epoch": 0.15,
|
| 1615 |
-
"learning_rate":
|
| 1616 |
-
"loss": 0.
|
| 1617 |
"step": 1335
|
| 1618 |
},
|
| 1619 |
{
|
| 1620 |
"epoch": 0.15,
|
| 1621 |
-
"learning_rate":
|
| 1622 |
-
"loss": 0.
|
| 1623 |
"step": 1340
|
| 1624 |
},
|
| 1625 |
{
|
| 1626 |
"epoch": 0.15,
|
| 1627 |
-
"learning_rate":
|
| 1628 |
-
"loss": 0.
|
| 1629 |
"step": 1345
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 0.15,
|
| 1633 |
-
"learning_rate":
|
| 1634 |
-
"loss": 0.
|
| 1635 |
"step": 1350
|
| 1636 |
},
|
| 1637 |
{
|
| 1638 |
"epoch": 0.16,
|
| 1639 |
-
"learning_rate":
|
| 1640 |
-
"loss": 0.
|
| 1641 |
"step": 1355
|
| 1642 |
},
|
| 1643 |
{
|
| 1644 |
"epoch": 0.16,
|
| 1645 |
-
"learning_rate":
|
| 1646 |
-
"loss": 0.
|
| 1647 |
"step": 1360
|
| 1648 |
},
|
| 1649 |
{
|
| 1650 |
"epoch": 0.16,
|
| 1651 |
-
"learning_rate":
|
| 1652 |
-
"loss": 0.
|
| 1653 |
"step": 1365
|
| 1654 |
},
|
| 1655 |
{
|
| 1656 |
"epoch": 0.16,
|
| 1657 |
-
"learning_rate":
|
| 1658 |
-
"loss": 0.
|
| 1659 |
"step": 1370
|
| 1660 |
},
|
| 1661 |
{
|
| 1662 |
"epoch": 0.16,
|
| 1663 |
-
"learning_rate":
|
| 1664 |
-
"loss": 0.
|
| 1665 |
"step": 1375
|
| 1666 |
},
|
| 1667 |
{
|
| 1668 |
"epoch": 0.16,
|
| 1669 |
-
"learning_rate":
|
| 1670 |
-
"loss": 0.
|
| 1671 |
"step": 1380
|
| 1672 |
},
|
| 1673 |
{
|
| 1674 |
"epoch": 0.16,
|
| 1675 |
-
"learning_rate":
|
| 1676 |
-
"loss": 0.
|
| 1677 |
"step": 1385
|
| 1678 |
},
|
| 1679 |
{
|
| 1680 |
"epoch": 0.16,
|
| 1681 |
-
"learning_rate":
|
| 1682 |
-
"loss": 0.
|
| 1683 |
"step": 1390
|
| 1684 |
},
|
| 1685 |
{
|
| 1686 |
"epoch": 0.16,
|
| 1687 |
-
"learning_rate":
|
| 1688 |
-
"loss": 0.
|
| 1689 |
"step": 1395
|
| 1690 |
},
|
| 1691 |
{
|
| 1692 |
"epoch": 0.16,
|
| 1693 |
-
"learning_rate":
|
| 1694 |
-
"loss": 0.
|
| 1695 |
"step": 1400
|
| 1696 |
},
|
| 1697 |
{
|
| 1698 |
"epoch": 0.16,
|
| 1699 |
-
"learning_rate":
|
| 1700 |
-
"loss": 0.
|
| 1701 |
"step": 1405
|
| 1702 |
},
|
| 1703 |
{
|
| 1704 |
"epoch": 0.16,
|
| 1705 |
-
"learning_rate":
|
| 1706 |
-
"loss": 0.
|
| 1707 |
"step": 1410
|
| 1708 |
},
|
| 1709 |
{
|
| 1710 |
"epoch": 0.16,
|
| 1711 |
-
"learning_rate":
|
| 1712 |
-
"loss": 0.
|
| 1713 |
"step": 1415
|
| 1714 |
},
|
| 1715 |
{
|
| 1716 |
"epoch": 0.16,
|
| 1717 |
-
"learning_rate":
|
| 1718 |
-
"loss": 0.
|
| 1719 |
"step": 1420
|
| 1720 |
},
|
| 1721 |
{
|
| 1722 |
"epoch": 0.16,
|
| 1723 |
-
"learning_rate":
|
| 1724 |
-
"loss": 0.
|
| 1725 |
"step": 1425
|
| 1726 |
},
|
| 1727 |
{
|
| 1728 |
"epoch": 0.16,
|
| 1729 |
-
"learning_rate":
|
| 1730 |
-
"loss": 0.
|
| 1731 |
"step": 1430
|
| 1732 |
},
|
| 1733 |
{
|
| 1734 |
"epoch": 0.16,
|
| 1735 |
-
"learning_rate":
|
| 1736 |
-
"loss": 0.
|
| 1737 |
"step": 1435
|
| 1738 |
},
|
| 1739 |
{
|
| 1740 |
"epoch": 0.17,
|
| 1741 |
-
"learning_rate":
|
| 1742 |
-
"loss": 0.
|
| 1743 |
"step": 1440
|
| 1744 |
},
|
| 1745 |
{
|
| 1746 |
"epoch": 0.17,
|
| 1747 |
-
"learning_rate":
|
| 1748 |
-
"loss": 0.
|
| 1749 |
"step": 1445
|
| 1750 |
},
|
| 1751 |
{
|
| 1752 |
"epoch": 0.17,
|
| 1753 |
-
"learning_rate":
|
| 1754 |
-
"loss": 0.
|
| 1755 |
"step": 1450
|
| 1756 |
},
|
| 1757 |
{
|
| 1758 |
"epoch": 0.17,
|
| 1759 |
-
"learning_rate":
|
| 1760 |
-
"loss": 0.
|
| 1761 |
"step": 1455
|
| 1762 |
},
|
| 1763 |
{
|
| 1764 |
"epoch": 0.17,
|
| 1765 |
-
"learning_rate":
|
| 1766 |
-
"loss": 0.
|
| 1767 |
"step": 1460
|
| 1768 |
},
|
| 1769 |
{
|
| 1770 |
"epoch": 0.17,
|
| 1771 |
-
"learning_rate":
|
| 1772 |
-
"loss": 0.
|
| 1773 |
"step": 1465
|
| 1774 |
},
|
| 1775 |
{
|
| 1776 |
"epoch": 0.17,
|
| 1777 |
-
"learning_rate":
|
| 1778 |
-
"loss": 0.
|
| 1779 |
"step": 1470
|
| 1780 |
},
|
| 1781 |
{
|
| 1782 |
"epoch": 0.17,
|
| 1783 |
-
"learning_rate":
|
| 1784 |
-
"loss": 0.
|
| 1785 |
"step": 1475
|
| 1786 |
},
|
| 1787 |
{
|
| 1788 |
"epoch": 0.17,
|
| 1789 |
-
"learning_rate":
|
| 1790 |
-
"loss": 0.
|
| 1791 |
"step": 1480
|
| 1792 |
},
|
| 1793 |
{
|
| 1794 |
"epoch": 0.17,
|
| 1795 |
-
"learning_rate":
|
| 1796 |
-
"loss": 0.
|
| 1797 |
"step": 1485
|
| 1798 |
},
|
| 1799 |
{
|
| 1800 |
"epoch": 0.17,
|
| 1801 |
-
"learning_rate":
|
| 1802 |
-
"loss": 0.
|
| 1803 |
"step": 1490
|
| 1804 |
},
|
| 1805 |
{
|
| 1806 |
"epoch": 0.17,
|
| 1807 |
-
"learning_rate":
|
| 1808 |
-
"loss": 0.
|
| 1809 |
"step": 1495
|
| 1810 |
},
|
| 1811 |
{
|
| 1812 |
"epoch": 0.17,
|
| 1813 |
-
"learning_rate":
|
| 1814 |
-
"loss": 0.
|
| 1815 |
"step": 1500
|
| 1816 |
},
|
| 1817 |
{
|
| 1818 |
"epoch": 0.17,
|
| 1819 |
-
"learning_rate":
|
| 1820 |
-
"loss": 0.
|
| 1821 |
"step": 1505
|
| 1822 |
},
|
| 1823 |
{
|
| 1824 |
"epoch": 0.17,
|
| 1825 |
-
"learning_rate":
|
| 1826 |
-
"loss": 0.
|
| 1827 |
"step": 1510
|
| 1828 |
},
|
| 1829 |
{
|
| 1830 |
"epoch": 0.17,
|
| 1831 |
-
"learning_rate":
|
| 1832 |
-
"loss": 0.
|
| 1833 |
"step": 1515
|
| 1834 |
},
|
| 1835 |
{
|
| 1836 |
"epoch": 0.17,
|
| 1837 |
-
"learning_rate":
|
| 1838 |
-
"loss": 0.
|
| 1839 |
"step": 1520
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
"epoch": 0.17,
|
| 1843 |
-
"learning_rate":
|
| 1844 |
-
"loss": 0.
|
| 1845 |
"step": 1525
|
| 1846 |
},
|
| 1847 |
{
|
| 1848 |
"epoch": 0.18,
|
| 1849 |
-
"learning_rate":
|
| 1850 |
-
"loss": 0.
|
| 1851 |
"step": 1530
|
| 1852 |
},
|
| 1853 |
{
|
| 1854 |
"epoch": 0.18,
|
| 1855 |
-
"learning_rate":
|
| 1856 |
-
"loss": 0.
|
| 1857 |
"step": 1535
|
| 1858 |
},
|
| 1859 |
{
|
| 1860 |
"epoch": 0.18,
|
| 1861 |
-
"learning_rate":
|
| 1862 |
-
"loss": 0.
|
| 1863 |
"step": 1540
|
| 1864 |
},
|
| 1865 |
{
|
| 1866 |
"epoch": 0.18,
|
| 1867 |
-
"learning_rate":
|
| 1868 |
-
"loss": 0.
|
| 1869 |
"step": 1545
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 0.18,
|
| 1873 |
-
"learning_rate":
|
| 1874 |
-
"loss": 0.
|
| 1875 |
"step": 1550
|
| 1876 |
},
|
| 1877 |
{
|
| 1878 |
"epoch": 0.18,
|
| 1879 |
-
"learning_rate":
|
| 1880 |
-
"loss": 0.
|
| 1881 |
"step": 1555
|
| 1882 |
},
|
| 1883 |
{
|
| 1884 |
"epoch": 0.18,
|
| 1885 |
-
"learning_rate":
|
| 1886 |
-
"loss": 0.
|
| 1887 |
"step": 1560
|
| 1888 |
},
|
| 1889 |
{
|
| 1890 |
"epoch": 0.18,
|
| 1891 |
-
"learning_rate":
|
| 1892 |
-
"loss": 0.
|
| 1893 |
"step": 1565
|
| 1894 |
},
|
| 1895 |
{
|
| 1896 |
"epoch": 0.18,
|
| 1897 |
-
"learning_rate":
|
| 1898 |
-
"loss": 0.
|
| 1899 |
"step": 1570
|
| 1900 |
},
|
| 1901 |
{
|
| 1902 |
"epoch": 0.18,
|
| 1903 |
-
"learning_rate":
|
| 1904 |
-
"loss": 0.
|
| 1905 |
"step": 1575
|
| 1906 |
},
|
| 1907 |
{
|
| 1908 |
"epoch": 0.18,
|
| 1909 |
-
"learning_rate":
|
| 1910 |
-
"loss": 0.
|
| 1911 |
"step": 1580
|
| 1912 |
},
|
| 1913 |
{
|
| 1914 |
"epoch": 0.18,
|
| 1915 |
-
"learning_rate":
|
| 1916 |
-
"loss": 0.
|
| 1917 |
"step": 1585
|
| 1918 |
},
|
| 1919 |
{
|
| 1920 |
"epoch": 0.18,
|
| 1921 |
-
"learning_rate":
|
| 1922 |
-
"loss": 0.
|
| 1923 |
"step": 1590
|
| 1924 |
},
|
| 1925 |
{
|
| 1926 |
"epoch": 0.18,
|
| 1927 |
-
"learning_rate":
|
| 1928 |
-
"loss": 0.
|
| 1929 |
"step": 1595
|
| 1930 |
},
|
| 1931 |
{
|
| 1932 |
"epoch": 0.18,
|
| 1933 |
-
"learning_rate":
|
| 1934 |
-
"loss": 0.
|
| 1935 |
"step": 1600
|
| 1936 |
},
|
| 1937 |
{
|
| 1938 |
"epoch": 0.18,
|
| 1939 |
-
"learning_rate":
|
| 1940 |
-
"loss": 0.
|
| 1941 |
"step": 1605
|
| 1942 |
},
|
| 1943 |
{
|
| 1944 |
"epoch": 0.18,
|
| 1945 |
-
"learning_rate":
|
| 1946 |
-
"loss": 0.
|
| 1947 |
"step": 1610
|
| 1948 |
},
|
| 1949 |
{
|
| 1950 |
"epoch": 0.19,
|
| 1951 |
-
"learning_rate":
|
| 1952 |
-
"loss": 0.
|
| 1953 |
"step": 1615
|
| 1954 |
},
|
| 1955 |
{
|
| 1956 |
"epoch": 0.19,
|
| 1957 |
-
"learning_rate":
|
| 1958 |
-
"loss": 0.
|
| 1959 |
"step": 1620
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
"epoch": 0.19,
|
| 1963 |
-
"learning_rate":
|
| 1964 |
-
"loss": 0.
|
| 1965 |
"step": 1625
|
| 1966 |
},
|
| 1967 |
{
|
| 1968 |
"epoch": 0.19,
|
| 1969 |
-
"learning_rate":
|
| 1970 |
-
"loss": 0.
|
| 1971 |
"step": 1630
|
| 1972 |
},
|
| 1973 |
{
|
| 1974 |
"epoch": 0.19,
|
| 1975 |
-
"learning_rate":
|
| 1976 |
-
"loss": 0.
|
| 1977 |
"step": 1635
|
| 1978 |
},
|
| 1979 |
{
|
| 1980 |
"epoch": 0.19,
|
| 1981 |
-
"learning_rate":
|
| 1982 |
-
"loss": 0.
|
| 1983 |
"step": 1640
|
| 1984 |
},
|
| 1985 |
{
|
| 1986 |
"epoch": 0.19,
|
| 1987 |
-
"learning_rate":
|
| 1988 |
-
"loss": 0.
|
| 1989 |
"step": 1645
|
| 1990 |
},
|
| 1991 |
{
|
| 1992 |
"epoch": 0.19,
|
| 1993 |
-
"learning_rate":
|
| 1994 |
-
"loss": 0.
|
| 1995 |
"step": 1650
|
| 1996 |
},
|
| 1997 |
{
|
| 1998 |
"epoch": 0.19,
|
| 1999 |
-
"learning_rate":
|
| 2000 |
-
"loss": 0.
|
| 2001 |
"step": 1655
|
| 2002 |
},
|
| 2003 |
{
|
| 2004 |
"epoch": 0.19,
|
| 2005 |
-
"learning_rate":
|
| 2006 |
-
"loss": 0.
|
| 2007 |
"step": 1660
|
| 2008 |
},
|
| 2009 |
{
|
| 2010 |
"epoch": 0.19,
|
| 2011 |
-
"learning_rate":
|
| 2012 |
-
"loss": 0.
|
| 2013 |
"step": 1665
|
| 2014 |
},
|
| 2015 |
{
|
| 2016 |
"epoch": 0.19,
|
| 2017 |
-
"learning_rate":
|
| 2018 |
-
"loss": 0.
|
| 2019 |
"step": 1670
|
| 2020 |
},
|
| 2021 |
{
|
| 2022 |
"epoch": 0.19,
|
| 2023 |
-
"learning_rate":
|
| 2024 |
-
"loss": 0.
|
| 2025 |
"step": 1675
|
| 2026 |
},
|
| 2027 |
{
|
| 2028 |
"epoch": 0.19,
|
| 2029 |
-
"learning_rate":
|
| 2030 |
-
"loss": 0.
|
| 2031 |
"step": 1680
|
| 2032 |
},
|
| 2033 |
{
|
| 2034 |
"epoch": 0.19,
|
| 2035 |
-
"learning_rate":
|
| 2036 |
-
"loss": 0.
|
| 2037 |
"step": 1685
|
| 2038 |
},
|
| 2039 |
{
|
| 2040 |
"epoch": 0.19,
|
| 2041 |
-
"learning_rate":
|
| 2042 |
-
"loss": 0.
|
| 2043 |
"step": 1690
|
| 2044 |
},
|
| 2045 |
{
|
| 2046 |
"epoch": 0.19,
|
| 2047 |
-
"learning_rate":
|
| 2048 |
-
"loss": 0.
|
| 2049 |
"step": 1695
|
| 2050 |
},
|
| 2051 |
{
|
| 2052 |
"epoch": 0.2,
|
| 2053 |
-
"learning_rate":
|
| 2054 |
-
"loss": 0.
|
| 2055 |
"step": 1700
|
| 2056 |
},
|
| 2057 |
{
|
| 2058 |
"epoch": 0.2,
|
| 2059 |
-
"learning_rate":
|
| 2060 |
-
"loss": 0.
|
| 2061 |
"step": 1705
|
| 2062 |
},
|
| 2063 |
{
|
| 2064 |
"epoch": 0.2,
|
| 2065 |
-
"learning_rate":
|
| 2066 |
-
"loss": 0.
|
| 2067 |
"step": 1710
|
| 2068 |
},
|
| 2069 |
{
|
| 2070 |
"epoch": 0.2,
|
| 2071 |
-
"learning_rate":
|
| 2072 |
-
"loss": 0.
|
| 2073 |
"step": 1715
|
| 2074 |
},
|
| 2075 |
{
|
| 2076 |
"epoch": 0.2,
|
| 2077 |
-
"learning_rate":
|
| 2078 |
-
"loss": 0.
|
| 2079 |
"step": 1720
|
| 2080 |
},
|
| 2081 |
{
|
| 2082 |
"epoch": 0.2,
|
| 2083 |
-
"learning_rate":
|
| 2084 |
-
"loss": 0.
|
| 2085 |
"step": 1725
|
| 2086 |
},
|
| 2087 |
{
|
| 2088 |
"epoch": 0.2,
|
| 2089 |
-
"learning_rate":
|
| 2090 |
-
"loss": 0.
|
| 2091 |
"step": 1730
|
| 2092 |
},
|
| 2093 |
{
|
| 2094 |
"epoch": 0.2,
|
| 2095 |
-
"learning_rate":
|
| 2096 |
-
"loss": 0.
|
| 2097 |
"step": 1735
|
| 2098 |
},
|
| 2099 |
{
|
| 2100 |
"epoch": 0.2,
|
| 2101 |
-
"learning_rate":
|
| 2102 |
-
"loss": 0.
|
| 2103 |
"step": 1740
|
| 2104 |
},
|
| 2105 |
{
|
| 2106 |
"epoch": 0.2,
|
| 2107 |
-
"learning_rate":
|
| 2108 |
-
"loss": 0.
|
| 2109 |
"step": 1745
|
| 2110 |
},
|
| 2111 |
{
|
| 2112 |
"epoch": 0.2,
|
| 2113 |
-
"learning_rate":
|
| 2114 |
-
"loss": 0.
|
| 2115 |
"step": 1750
|
| 2116 |
},
|
| 2117 |
{
|
| 2118 |
"epoch": 0.2,
|
| 2119 |
-
"learning_rate":
|
| 2120 |
-
"loss": 0.
|
| 2121 |
"step": 1755
|
| 2122 |
},
|
| 2123 |
{
|
| 2124 |
"epoch": 0.2,
|
| 2125 |
-
"learning_rate":
|
| 2126 |
-
"loss": 0.
|
| 2127 |
"step": 1760
|
| 2128 |
},
|
| 2129 |
{
|
| 2130 |
"epoch": 0.2,
|
| 2131 |
-
"learning_rate":
|
| 2132 |
-
"loss": 0.
|
| 2133 |
"step": 1765
|
| 2134 |
},
|
| 2135 |
{
|
| 2136 |
"epoch": 0.2,
|
| 2137 |
-
"learning_rate":
|
| 2138 |
-
"loss": 0.
|
| 2139 |
"step": 1770
|
| 2140 |
},
|
| 2141 |
{
|
| 2142 |
"epoch": 0.2,
|
| 2143 |
-
"learning_rate":
|
| 2144 |
-
"loss": 0.
|
| 2145 |
"step": 1775
|
| 2146 |
},
|
| 2147 |
{
|
| 2148 |
"epoch": 0.2,
|
| 2149 |
-
"learning_rate":
|
| 2150 |
-
"loss": 0.
|
| 2151 |
"step": 1780
|
| 2152 |
},
|
| 2153 |
{
|
| 2154 |
"epoch": 0.2,
|
| 2155 |
-
"learning_rate":
|
| 2156 |
-
"loss": 0.
|
| 2157 |
"step": 1785
|
| 2158 |
},
|
| 2159 |
{
|
| 2160 |
"epoch": 0.21,
|
| 2161 |
-
"learning_rate":
|
| 2162 |
-
"loss": 0.
|
| 2163 |
"step": 1790
|
| 2164 |
},
|
| 2165 |
{
|
| 2166 |
"epoch": 0.21,
|
| 2167 |
-
"learning_rate":
|
| 2168 |
-
"loss": 0.
|
| 2169 |
"step": 1795
|
| 2170 |
},
|
| 2171 |
{
|
| 2172 |
"epoch": 0.21,
|
| 2173 |
-
"learning_rate":
|
| 2174 |
-
"loss": 0.
|
| 2175 |
"step": 1800
|
| 2176 |
},
|
| 2177 |
{
|
| 2178 |
"epoch": 0.21,
|
| 2179 |
-
"learning_rate":
|
| 2180 |
-
"loss": 0.
|
| 2181 |
"step": 1805
|
| 2182 |
},
|
| 2183 |
{
|
| 2184 |
"epoch": 0.21,
|
| 2185 |
-
"learning_rate":
|
| 2186 |
-
"loss": 0.
|
| 2187 |
"step": 1810
|
| 2188 |
},
|
| 2189 |
{
|
| 2190 |
"epoch": 0.21,
|
| 2191 |
-
"learning_rate":
|
| 2192 |
-
"loss": 0.
|
| 2193 |
"step": 1815
|
| 2194 |
},
|
| 2195 |
{
|
| 2196 |
"epoch": 0.21,
|
| 2197 |
-
"learning_rate":
|
| 2198 |
-
"loss": 0.
|
| 2199 |
"step": 1820
|
| 2200 |
},
|
| 2201 |
{
|
| 2202 |
"epoch": 0.21,
|
| 2203 |
-
"learning_rate":
|
| 2204 |
-
"loss": 0.
|
| 2205 |
"step": 1825
|
| 2206 |
},
|
| 2207 |
{
|
| 2208 |
"epoch": 0.21,
|
| 2209 |
-
"learning_rate":
|
| 2210 |
-
"loss": 0.
|
| 2211 |
"step": 1830
|
| 2212 |
},
|
| 2213 |
{
|
| 2214 |
"epoch": 0.21,
|
| 2215 |
-
"learning_rate":
|
| 2216 |
"loss": 0.0011,
|
| 2217 |
"step": 1835
|
| 2218 |
},
|
| 2219 |
{
|
| 2220 |
"epoch": 0.21,
|
| 2221 |
-
"learning_rate":
|
| 2222 |
-
"loss": 0.
|
| 2223 |
"step": 1840
|
| 2224 |
},
|
| 2225 |
{
|
| 2226 |
"epoch": 0.21,
|
| 2227 |
-
"learning_rate":
|
| 2228 |
-
"loss": 0.
|
| 2229 |
"step": 1845
|
| 2230 |
},
|
| 2231 |
{
|
| 2232 |
"epoch": 0.21,
|
| 2233 |
-
"learning_rate":
|
| 2234 |
-
"loss": 0.
|
| 2235 |
"step": 1850
|
| 2236 |
},
|
| 2237 |
{
|
| 2238 |
"epoch": 0.21,
|
| 2239 |
-
"learning_rate":
|
| 2240 |
-
"loss": 0.
|
| 2241 |
"step": 1855
|
| 2242 |
},
|
| 2243 |
{
|
| 2244 |
"epoch": 0.21,
|
| 2245 |
-
"learning_rate":
|
| 2246 |
-
"loss": 0.
|
| 2247 |
"step": 1860
|
| 2248 |
},
|
| 2249 |
{
|
| 2250 |
"epoch": 0.21,
|
| 2251 |
-
"learning_rate":
|
| 2252 |
-
"loss": 0.
|
| 2253 |
"step": 1865
|
| 2254 |
},
|
| 2255 |
{
|
| 2256 |
"epoch": 0.21,
|
| 2257 |
-
"learning_rate":
|
| 2258 |
-
"loss": 0.
|
| 2259 |
"step": 1870
|
| 2260 |
},
|
| 2261 |
{
|
| 2262 |
"epoch": 0.22,
|
| 2263 |
-
"learning_rate":
|
| 2264 |
-
"loss": 0.
|
| 2265 |
"step": 1875
|
| 2266 |
},
|
| 2267 |
{
|
| 2268 |
"epoch": 0.22,
|
| 2269 |
-
"learning_rate":
|
| 2270 |
-
"loss": 0.
|
| 2271 |
"step": 1880
|
| 2272 |
},
|
| 2273 |
{
|
| 2274 |
"epoch": 0.22,
|
| 2275 |
-
"learning_rate":
|
| 2276 |
-
"loss": 0.
|
| 2277 |
"step": 1885
|
| 2278 |
},
|
| 2279 |
{
|
| 2280 |
"epoch": 0.22,
|
| 2281 |
-
"learning_rate":
|
| 2282 |
-
"loss": 0.
|
| 2283 |
"step": 1890
|
| 2284 |
},
|
| 2285 |
{
|
| 2286 |
"epoch": 0.22,
|
| 2287 |
-
"learning_rate":
|
| 2288 |
-
"loss": 0.
|
| 2289 |
"step": 1895
|
| 2290 |
},
|
| 2291 |
{
|
| 2292 |
"epoch": 0.22,
|
| 2293 |
-
"learning_rate":
|
| 2294 |
-
"loss": 0.
|
| 2295 |
"step": 1900
|
| 2296 |
},
|
| 2297 |
{
|
| 2298 |
"epoch": 0.22,
|
| 2299 |
-
"learning_rate":
|
| 2300 |
-
"loss": 0.
|
| 2301 |
"step": 1905
|
| 2302 |
},
|
| 2303 |
{
|
| 2304 |
"epoch": 0.22,
|
| 2305 |
-
"learning_rate":
|
| 2306 |
-
"loss": 0.
|
| 2307 |
"step": 1910
|
| 2308 |
},
|
| 2309 |
{
|
| 2310 |
"epoch": 0.22,
|
| 2311 |
-
"learning_rate":
|
| 2312 |
-
"loss": 0.
|
| 2313 |
"step": 1915
|
| 2314 |
},
|
| 2315 |
{
|
| 2316 |
"epoch": 0.22,
|
| 2317 |
-
"learning_rate":
|
| 2318 |
-
"loss": 0.
|
| 2319 |
"step": 1920
|
| 2320 |
},
|
| 2321 |
{
|
| 2322 |
"epoch": 0.22,
|
| 2323 |
-
"learning_rate":
|
| 2324 |
-
"loss": 0.
|
| 2325 |
"step": 1925
|
| 2326 |
},
|
| 2327 |
{
|
| 2328 |
"epoch": 0.22,
|
| 2329 |
-
"learning_rate":
|
| 2330 |
-
"loss": 0.
|
| 2331 |
"step": 1930
|
| 2332 |
},
|
| 2333 |
{
|
| 2334 |
"epoch": 0.22,
|
| 2335 |
-
"learning_rate":
|
| 2336 |
-
"loss": 0.
|
| 2337 |
"step": 1935
|
| 2338 |
},
|
| 2339 |
{
|
| 2340 |
"epoch": 0.22,
|
| 2341 |
-
"learning_rate":
|
| 2342 |
-
"loss": 0.
|
| 2343 |
"step": 1940
|
| 2344 |
},
|
| 2345 |
{
|
| 2346 |
"epoch": 0.22,
|
| 2347 |
-
"learning_rate":
|
| 2348 |
-
"loss": 0.
|
| 2349 |
"step": 1945
|
| 2350 |
},
|
| 2351 |
{
|
| 2352 |
"epoch": 0.22,
|
| 2353 |
-
"learning_rate":
|
| 2354 |
-
"loss": 0.
|
| 2355 |
"step": 1950
|
| 2356 |
},
|
| 2357 |
{
|
| 2358 |
"epoch": 0.22,
|
| 2359 |
-
"learning_rate":
|
| 2360 |
-
"loss": 0.
|
| 2361 |
"step": 1955
|
| 2362 |
},
|
| 2363 |
{
|
| 2364 |
"epoch": 0.22,
|
| 2365 |
-
"learning_rate":
|
| 2366 |
-
"loss": 0.
|
| 2367 |
"step": 1960
|
| 2368 |
},
|
| 2369 |
{
|
| 2370 |
"epoch": 0.23,
|
| 2371 |
-
"learning_rate":
|
| 2372 |
-
"loss": 0.
|
| 2373 |
"step": 1965
|
| 2374 |
},
|
| 2375 |
{
|
| 2376 |
"epoch": 0.23,
|
| 2377 |
-
"learning_rate":
|
| 2378 |
-
"loss": 0.
|
| 2379 |
"step": 1970
|
| 2380 |
},
|
| 2381 |
{
|
| 2382 |
"epoch": 0.23,
|
| 2383 |
-
"learning_rate":
|
| 2384 |
-
"loss": 0.
|
| 2385 |
"step": 1975
|
| 2386 |
},
|
| 2387 |
{
|
| 2388 |
"epoch": 0.23,
|
| 2389 |
-
"learning_rate":
|
| 2390 |
-
"loss": 0.
|
| 2391 |
"step": 1980
|
| 2392 |
},
|
| 2393 |
{
|
| 2394 |
"epoch": 0.23,
|
| 2395 |
-
"learning_rate":
|
| 2396 |
-
"loss": 0.
|
| 2397 |
"step": 1985
|
| 2398 |
},
|
| 2399 |
{
|
| 2400 |
"epoch": 0.23,
|
| 2401 |
-
"learning_rate":
|
| 2402 |
-
"loss": 0.
|
| 2403 |
"step": 1990
|
| 2404 |
},
|
| 2405 |
{
|
| 2406 |
"epoch": 0.23,
|
| 2407 |
-
"learning_rate":
|
| 2408 |
-
"loss": 0.
|
| 2409 |
"step": 1995
|
| 2410 |
},
|
| 2411 |
{
|
| 2412 |
"epoch": 0.23,
|
| 2413 |
-
"learning_rate":
|
| 2414 |
-
"loss": 0.
|
| 2415 |
"step": 2000
|
| 2416 |
},
|
| 2417 |
{
|
| 2418 |
"epoch": 0.23,
|
| 2419 |
-
"learning_rate":
|
| 2420 |
-
"loss": 0.
|
| 2421 |
"step": 2005
|
| 2422 |
},
|
| 2423 |
{
|
| 2424 |
"epoch": 0.23,
|
| 2425 |
-
"learning_rate":
|
| 2426 |
-
"loss": 0.
|
| 2427 |
"step": 2010
|
| 2428 |
},
|
| 2429 |
{
|
| 2430 |
"epoch": 0.23,
|
| 2431 |
-
"learning_rate":
|
| 2432 |
-
"loss": 0.
|
| 2433 |
"step": 2015
|
| 2434 |
},
|
| 2435 |
{
|
| 2436 |
"epoch": 0.23,
|
| 2437 |
-
"learning_rate":
|
| 2438 |
-
"loss": 0.
|
| 2439 |
"step": 2020
|
| 2440 |
},
|
| 2441 |
{
|
| 2442 |
"epoch": 0.23,
|
| 2443 |
-
"learning_rate":
|
| 2444 |
-
"loss": 0.
|
| 2445 |
"step": 2025
|
| 2446 |
},
|
| 2447 |
{
|
| 2448 |
"epoch": 0.23,
|
| 2449 |
-
"learning_rate":
|
| 2450 |
-
"loss": 0.
|
| 2451 |
"step": 2030
|
| 2452 |
},
|
| 2453 |
{
|
| 2454 |
"epoch": 0.23,
|
| 2455 |
-
"learning_rate":
|
| 2456 |
-
"loss": 0.
|
| 2457 |
"step": 2035
|
| 2458 |
},
|
| 2459 |
{
|
| 2460 |
"epoch": 0.23,
|
| 2461 |
-
"learning_rate":
|
| 2462 |
-
"loss": 0.
|
| 2463 |
"step": 2040
|
| 2464 |
},
|
| 2465 |
{
|
| 2466 |
"epoch": 0.23,
|
| 2467 |
-
"learning_rate":
|
| 2468 |
-
"loss": 0.
|
| 2469 |
"step": 2045
|
| 2470 |
},
|
| 2471 |
{
|
| 2472 |
"epoch": 0.24,
|
| 2473 |
-
"learning_rate":
|
| 2474 |
-
"loss": 0.
|
| 2475 |
"step": 2050
|
| 2476 |
},
|
| 2477 |
{
|
| 2478 |
"epoch": 0.24,
|
| 2479 |
-
"learning_rate":
|
| 2480 |
-
"loss": 0.
|
| 2481 |
"step": 2055
|
| 2482 |
},
|
| 2483 |
{
|
| 2484 |
"epoch": 0.24,
|
| 2485 |
-
"learning_rate":
|
| 2486 |
-
"loss": 0.
|
| 2487 |
"step": 2060
|
| 2488 |
},
|
| 2489 |
{
|
| 2490 |
"epoch": 0.24,
|
| 2491 |
-
"learning_rate":
|
| 2492 |
-
"loss": 0.
|
| 2493 |
"step": 2065
|
| 2494 |
},
|
| 2495 |
{
|
| 2496 |
"epoch": 0.24,
|
| 2497 |
-
"learning_rate":
|
| 2498 |
-
"loss": 0.
|
| 2499 |
"step": 2070
|
| 2500 |
},
|
| 2501 |
{
|
| 2502 |
"epoch": 0.24,
|
| 2503 |
-
"learning_rate":
|
| 2504 |
-
"loss": 0.
|
| 2505 |
"step": 2075
|
| 2506 |
},
|
| 2507 |
{
|
| 2508 |
"epoch": 0.24,
|
| 2509 |
-
"learning_rate":
|
| 2510 |
-
"loss": 0.
|
| 2511 |
"step": 2080
|
| 2512 |
},
|
| 2513 |
{
|
| 2514 |
"epoch": 0.24,
|
| 2515 |
-
"learning_rate":
|
| 2516 |
-
"loss": 0.
|
| 2517 |
"step": 2085
|
| 2518 |
},
|
| 2519 |
{
|
| 2520 |
"epoch": 0.24,
|
| 2521 |
-
"learning_rate":
|
| 2522 |
-
"loss": 0.
|
| 2523 |
"step": 2090
|
| 2524 |
},
|
| 2525 |
{
|
| 2526 |
"epoch": 0.24,
|
| 2527 |
-
"learning_rate":
|
| 2528 |
-
"loss": 0.
|
| 2529 |
"step": 2095
|
| 2530 |
},
|
| 2531 |
{
|
| 2532 |
"epoch": 0.24,
|
| 2533 |
-
"learning_rate":
|
| 2534 |
-
"loss": 0.
|
| 2535 |
"step": 2100
|
| 2536 |
},
|
| 2537 |
{
|
| 2538 |
"epoch": 0.24,
|
| 2539 |
-
"learning_rate":
|
| 2540 |
-
"loss": 0.
|
| 2541 |
"step": 2105
|
| 2542 |
},
|
| 2543 |
{
|
| 2544 |
"epoch": 0.24,
|
| 2545 |
-
"learning_rate":
|
| 2546 |
-
"loss": 0.
|
| 2547 |
"step": 2110
|
| 2548 |
},
|
| 2549 |
{
|
| 2550 |
"epoch": 0.24,
|
| 2551 |
-
"learning_rate":
|
| 2552 |
-
"loss": 0.
|
| 2553 |
"step": 2115
|
| 2554 |
},
|
| 2555 |
{
|
| 2556 |
"epoch": 0.24,
|
| 2557 |
-
"learning_rate":
|
| 2558 |
-
"loss": 0.
|
| 2559 |
"step": 2120
|
| 2560 |
},
|
| 2561 |
{
|
| 2562 |
"epoch": 0.24,
|
| 2563 |
-
"learning_rate":
|
| 2564 |
-
"loss": 0.
|
| 2565 |
"step": 2125
|
| 2566 |
},
|
| 2567 |
{
|
| 2568 |
"epoch": 0.24,
|
| 2569 |
-
"learning_rate":
|
| 2570 |
-
"loss": 0.
|
| 2571 |
"step": 2130
|
| 2572 |
},
|
| 2573 |
{
|
| 2574 |
"epoch": 0.24,
|
| 2575 |
-
"learning_rate":
|
| 2576 |
"loss": 0.0013,
|
| 2577 |
"step": 2135
|
| 2578 |
},
|
| 2579 |
{
|
| 2580 |
"epoch": 0.25,
|
| 2581 |
-
"learning_rate":
|
| 2582 |
-
"loss": 0.
|
| 2583 |
"step": 2140
|
| 2584 |
},
|
| 2585 |
{
|
| 2586 |
"epoch": 0.25,
|
| 2587 |
-
"learning_rate":
|
| 2588 |
-
"loss": 0.
|
| 2589 |
"step": 2145
|
| 2590 |
},
|
| 2591 |
{
|
| 2592 |
"epoch": 0.25,
|
| 2593 |
-
"learning_rate":
|
| 2594 |
-
"loss": 0.
|
| 2595 |
"step": 2150
|
| 2596 |
},
|
| 2597 |
{
|
| 2598 |
"epoch": 0.25,
|
| 2599 |
-
"learning_rate":
|
| 2600 |
-
"loss": 0.
|
| 2601 |
"step": 2155
|
| 2602 |
},
|
| 2603 |
{
|
| 2604 |
"epoch": 0.25,
|
| 2605 |
-
"learning_rate":
|
| 2606 |
-
"loss": 0.
|
| 2607 |
"step": 2160
|
| 2608 |
},
|
| 2609 |
{
|
| 2610 |
"epoch": 0.25,
|
| 2611 |
-
"learning_rate":
|
| 2612 |
-
"loss": 0.
|
| 2613 |
"step": 2165
|
| 2614 |
},
|
| 2615 |
{
|
| 2616 |
"epoch": 0.25,
|
| 2617 |
-
"learning_rate":
|
| 2618 |
-
"loss": 0.
|
| 2619 |
"step": 2170
|
| 2620 |
},
|
| 2621 |
{
|
| 2622 |
"epoch": 0.25,
|
| 2623 |
-
"learning_rate":
|
| 2624 |
"loss": 0.001,
|
| 2625 |
"step": 2175
|
| 2626 |
},
|
| 2627 |
{
|
| 2628 |
"epoch": 0.25,
|
| 2629 |
-
"learning_rate":
|
| 2630 |
-
"loss": 0.
|
| 2631 |
"step": 2180
|
| 2632 |
},
|
| 2633 |
{
|
| 2634 |
"epoch": 0.25,
|
| 2635 |
-
"learning_rate":
|
| 2636 |
"loss": 0.0011,
|
| 2637 |
"step": 2185
|
| 2638 |
},
|
| 2639 |
{
|
| 2640 |
"epoch": 0.25,
|
| 2641 |
-
"learning_rate":
|
| 2642 |
-
"loss": 0.
|
| 2643 |
"step": 2190
|
| 2644 |
},
|
| 2645 |
{
|
| 2646 |
"epoch": 0.25,
|
| 2647 |
-
"learning_rate":
|
| 2648 |
-
"loss": 0.
|
| 2649 |
"step": 2195
|
| 2650 |
},
|
| 2651 |
{
|
| 2652 |
"epoch": 0.25,
|
| 2653 |
-
"learning_rate":
|
| 2654 |
-
"loss": 0.
|
| 2655 |
"step": 2200
|
| 2656 |
},
|
| 2657 |
{
|
| 2658 |
"epoch": 0.25,
|
| 2659 |
-
"learning_rate":
|
| 2660 |
-
"loss": 0.
|
| 2661 |
"step": 2205
|
| 2662 |
},
|
| 2663 |
{
|
| 2664 |
"epoch": 0.25,
|
| 2665 |
-
"learning_rate":
|
| 2666 |
-
"loss": 0.
|
| 2667 |
"step": 2210
|
| 2668 |
},
|
| 2669 |
{
|
| 2670 |
"epoch": 0.25,
|
| 2671 |
-
"learning_rate":
|
| 2672 |
-
"loss": 0.
|
| 2673 |
"step": 2215
|
| 2674 |
},
|
| 2675 |
{
|
| 2676 |
"epoch": 0.25,
|
| 2677 |
-
"learning_rate":
|
| 2678 |
-
"loss": 0.
|
| 2679 |
"step": 2220
|
| 2680 |
},
|
| 2681 |
{
|
| 2682 |
"epoch": 0.26,
|
| 2683 |
-
"learning_rate":
|
| 2684 |
-
"loss": 0.
|
| 2685 |
"step": 2225
|
| 2686 |
},
|
| 2687 |
{
|
| 2688 |
"epoch": 0.26,
|
| 2689 |
-
"learning_rate":
|
| 2690 |
-
"loss": 0.
|
| 2691 |
"step": 2230
|
| 2692 |
},
|
| 2693 |
{
|
| 2694 |
"epoch": 0.26,
|
| 2695 |
-
"learning_rate":
|
| 2696 |
-
"loss": 0.
|
| 2697 |
"step": 2235
|
| 2698 |
},
|
| 2699 |
{
|
| 2700 |
"epoch": 0.26,
|
| 2701 |
-
"learning_rate":
|
| 2702 |
-
"loss": 0.
|
| 2703 |
"step": 2240
|
| 2704 |
},
|
| 2705 |
{
|
| 2706 |
"epoch": 0.26,
|
| 2707 |
-
"learning_rate":
|
| 2708 |
-
"loss": 0.
|
| 2709 |
"step": 2245
|
| 2710 |
},
|
| 2711 |
{
|
| 2712 |
"epoch": 0.26,
|
| 2713 |
-
"learning_rate":
|
| 2714 |
-
"loss": 0.
|
| 2715 |
"step": 2250
|
| 2716 |
},
|
| 2717 |
{
|
| 2718 |
"epoch": 0.26,
|
| 2719 |
-
"learning_rate":
|
| 2720 |
"loss": 0.001,
|
| 2721 |
"step": 2255
|
| 2722 |
},
|
| 2723 |
{
|
| 2724 |
"epoch": 0.26,
|
| 2725 |
-
"learning_rate":
|
| 2726 |
-
"loss": 0.
|
| 2727 |
"step": 2260
|
| 2728 |
},
|
| 2729 |
{
|
| 2730 |
"epoch": 0.26,
|
| 2731 |
-
"learning_rate":
|
| 2732 |
-
"loss": 0.
|
| 2733 |
"step": 2265
|
| 2734 |
},
|
| 2735 |
{
|
| 2736 |
"epoch": 0.26,
|
| 2737 |
-
"learning_rate":
|
| 2738 |
-
"loss": 0.
|
| 2739 |
"step": 2270
|
| 2740 |
},
|
| 2741 |
{
|
| 2742 |
"epoch": 0.26,
|
| 2743 |
-
"learning_rate":
|
| 2744 |
-
"loss": 0.
|
| 2745 |
"step": 2275
|
| 2746 |
},
|
| 2747 |
{
|
| 2748 |
"epoch": 0.26,
|
| 2749 |
-
"learning_rate":
|
| 2750 |
-
"loss": 0.
|
| 2751 |
"step": 2280
|
| 2752 |
},
|
| 2753 |
{
|
| 2754 |
"epoch": 0.26,
|
| 2755 |
-
"learning_rate":
|
| 2756 |
-
"loss": 0.
|
| 2757 |
"step": 2285
|
| 2758 |
},
|
| 2759 |
{
|
| 2760 |
"epoch": 0.26,
|
| 2761 |
-
"learning_rate":
|
| 2762 |
-
"loss": 0.
|
| 2763 |
"step": 2290
|
| 2764 |
},
|
| 2765 |
{
|
| 2766 |
"epoch": 0.26,
|
| 2767 |
-
"learning_rate":
|
| 2768 |
-
"loss": 0.
|
| 2769 |
"step": 2295
|
| 2770 |
},
|
| 2771 |
{
|
| 2772 |
"epoch": 0.26,
|
| 2773 |
-
"learning_rate":
|
| 2774 |
-
"loss": 0.
|
| 2775 |
"step": 2300
|
| 2776 |
},
|
| 2777 |
{
|
| 2778 |
"epoch": 0.26,
|
| 2779 |
-
"learning_rate":
|
| 2780 |
-
"loss": 0.
|
| 2781 |
"step": 2305
|
| 2782 |
},
|
| 2783 |
{
|
| 2784 |
"epoch": 0.27,
|
| 2785 |
-
"learning_rate":
|
| 2786 |
-
"loss": 0.
|
| 2787 |
"step": 2310
|
| 2788 |
},
|
| 2789 |
{
|
| 2790 |
"epoch": 0.27,
|
| 2791 |
-
"learning_rate":
|
| 2792 |
-
"loss": 0.
|
| 2793 |
"step": 2315
|
| 2794 |
},
|
| 2795 |
{
|
| 2796 |
"epoch": 0.27,
|
| 2797 |
-
"learning_rate":
|
| 2798 |
-
"loss": 0.
|
| 2799 |
"step": 2320
|
| 2800 |
},
|
| 2801 |
{
|
| 2802 |
"epoch": 0.27,
|
| 2803 |
-
"learning_rate":
|
| 2804 |
-
"loss": 0.
|
| 2805 |
"step": 2325
|
| 2806 |
},
|
| 2807 |
{
|
| 2808 |
"epoch": 0.27,
|
| 2809 |
-
"learning_rate":
|
| 2810 |
-
"loss": 0.
|
| 2811 |
"step": 2330
|
| 2812 |
},
|
| 2813 |
{
|
| 2814 |
"epoch": 0.27,
|
| 2815 |
-
"learning_rate":
|
| 2816 |
-
"loss": 0.
|
| 2817 |
"step": 2335
|
| 2818 |
},
|
| 2819 |
{
|
| 2820 |
"epoch": 0.27,
|
| 2821 |
-
"learning_rate":
|
| 2822 |
-
"loss": 0.
|
| 2823 |
"step": 2340
|
| 2824 |
},
|
| 2825 |
{
|
| 2826 |
"epoch": 0.27,
|
| 2827 |
-
"learning_rate":
|
| 2828 |
"loss": 0.0015,
|
| 2829 |
"step": 2345
|
| 2830 |
},
|
| 2831 |
{
|
| 2832 |
"epoch": 0.27,
|
| 2833 |
-
"learning_rate":
|
| 2834 |
-
"loss": 0.
|
| 2835 |
"step": 2350
|
| 2836 |
},
|
| 2837 |
{
|
| 2838 |
"epoch": 0.27,
|
| 2839 |
-
"learning_rate":
|
| 2840 |
-
"loss": 0.
|
| 2841 |
"step": 2355
|
| 2842 |
},
|
| 2843 |
{
|
| 2844 |
"epoch": 0.27,
|
| 2845 |
-
"learning_rate":
|
| 2846 |
-
"loss": 0.
|
| 2847 |
"step": 2360
|
| 2848 |
},
|
| 2849 |
{
|
| 2850 |
"epoch": 0.27,
|
| 2851 |
-
"learning_rate":
|
| 2852 |
-
"loss": 0.
|
| 2853 |
"step": 2365
|
| 2854 |
},
|
| 2855 |
{
|
| 2856 |
"epoch": 0.27,
|
| 2857 |
-
"learning_rate":
|
| 2858 |
-
"loss": 0.
|
| 2859 |
"step": 2370
|
| 2860 |
},
|
| 2861 |
{
|
| 2862 |
"epoch": 0.27,
|
| 2863 |
-
"learning_rate":
|
| 2864 |
-
"loss": 0.
|
| 2865 |
"step": 2375
|
| 2866 |
},
|
| 2867 |
{
|
| 2868 |
"epoch": 0.27,
|
| 2869 |
-
"learning_rate":
|
| 2870 |
-
"loss": 0.
|
| 2871 |
"step": 2380
|
| 2872 |
},
|
| 2873 |
{
|
| 2874 |
"epoch": 0.27,
|
| 2875 |
-
"learning_rate":
|
| 2876 |
-
"loss": 0.
|
| 2877 |
"step": 2385
|
| 2878 |
},
|
| 2879 |
{
|
| 2880 |
"epoch": 0.27,
|
| 2881 |
-
"learning_rate":
|
| 2882 |
-
"loss": 0.
|
| 2883 |
"step": 2390
|
| 2884 |
},
|
| 2885 |
{
|
| 2886 |
"epoch": 0.27,
|
| 2887 |
-
"learning_rate":
|
| 2888 |
-
"loss": 0.
|
| 2889 |
"step": 2395
|
| 2890 |
},
|
| 2891 |
{
|
| 2892 |
"epoch": 0.28,
|
| 2893 |
-
"learning_rate":
|
| 2894 |
-
"loss": 0.
|
| 2895 |
"step": 2400
|
| 2896 |
},
|
| 2897 |
{
|
| 2898 |
"epoch": 0.28,
|
| 2899 |
"step": 2400,
|
| 2900 |
"total_flos": 3.3948129315510026e+18,
|
| 2901 |
-
"train_loss": 0.
|
| 2902 |
-
"train_runtime":
|
| 2903 |
-
"train_samples_per_second": 1.
|
| 2904 |
-
"train_steps_per_second": 0.
|
| 2905 |
}
|
| 2906 |
],
|
| 2907 |
"logging_steps": 5,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 4.166666666666667e-08,
|
| 14 |
"loss": 0.6931,
|
| 15 |
"step": 1
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.0,
|
| 19 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 20 |
+
"loss": 0.6932,
|
| 21 |
"step": 5
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.0,
|
| 25 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 26 |
+
"loss": 0.6926,
|
| 27 |
"step": 10
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.0,
|
| 31 |
+
"learning_rate": 6.25e-07,
|
| 32 |
+
"loss": 0.6907,
|
| 33 |
"step": 15
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.0,
|
| 37 |
+
"learning_rate": 8.333333333333333e-07,
|
| 38 |
+
"loss": 0.6867,
|
| 39 |
"step": 20
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.0,
|
| 43 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 44 |
+
"loss": 0.6798,
|
| 45 |
"step": 25
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.0,
|
| 49 |
+
"learning_rate": 1.25e-06,
|
| 50 |
+
"loss": 0.6683,
|
| 51 |
"step": 30
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.0,
|
| 55 |
+
"learning_rate": 1.4583333333333335e-06,
|
| 56 |
+
"loss": 0.6532,
|
| 57 |
"step": 35
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.0,
|
| 61 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 62 |
+
"loss": 0.635,
|
| 63 |
"step": 40
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.01,
|
| 67 |
+
"learning_rate": 1.8750000000000003e-06,
|
| 68 |
+
"loss": 0.6108,
|
| 69 |
"step": 45
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.01,
|
| 73 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 74 |
+
"loss": 0.5844,
|
| 75 |
"step": 50
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.01,
|
| 79 |
+
"learning_rate": 2.2916666666666666e-06,
|
| 80 |
+
"loss": 0.5544,
|
| 81 |
"step": 55
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.01,
|
| 85 |
+
"learning_rate": 2.5e-06,
|
| 86 |
+
"loss": 0.5227,
|
| 87 |
"step": 60
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.01,
|
| 91 |
+
"learning_rate": 2.7083333333333334e-06,
|
| 92 |
+
"loss": 0.4867,
|
| 93 |
"step": 65
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.01,
|
| 97 |
+
"learning_rate": 2.916666666666667e-06,
|
| 98 |
+
"loss": 0.4471,
|
| 99 |
"step": 70
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.01,
|
| 103 |
+
"learning_rate": 3.125e-06,
|
| 104 |
+
"loss": 0.4093,
|
| 105 |
"step": 75
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.01,
|
| 109 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 110 |
+
"loss": 0.3671,
|
| 111 |
"step": 80
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.01,
|
| 115 |
+
"learning_rate": 3.5416666666666673e-06,
|
| 116 |
+
"loss": 0.3131,
|
| 117 |
"step": 85
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.01,
|
| 121 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 122 |
+
"loss": 0.267,
|
| 123 |
"step": 90
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.01,
|
| 127 |
+
"learning_rate": 3.958333333333333e-06,
|
| 128 |
+
"loss": 0.2313,
|
| 129 |
"step": 95
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.01,
|
| 133 |
+
"learning_rate": 4.166666666666667e-06,
|
| 134 |
+
"loss": 0.1819,
|
| 135 |
"step": 100
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.01,
|
| 139 |
+
"learning_rate": 4.3750000000000005e-06,
|
| 140 |
+
"loss": 0.142,
|
| 141 |
"step": 105
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.01,
|
| 145 |
+
"learning_rate": 4.583333333333333e-06,
|
| 146 |
+
"loss": 0.1105,
|
| 147 |
"step": 110
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.01,
|
| 151 |
+
"learning_rate": 4.791666666666668e-06,
|
| 152 |
+
"loss": 0.0886,
|
| 153 |
"step": 115
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.01,
|
| 157 |
+
"learning_rate": 5e-06,
|
| 158 |
+
"loss": 0.0729,
|
| 159 |
"step": 120
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.01,
|
| 163 |
+
"learning_rate": 5.208333333333334e-06,
|
| 164 |
+
"loss": 0.0656,
|
| 165 |
"step": 125
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.01,
|
| 169 |
+
"learning_rate": 5.416666666666667e-06,
|
| 170 |
+
"loss": 0.0502,
|
| 171 |
"step": 130
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.02,
|
| 175 |
+
"learning_rate": 5.625e-06,
|
| 176 |
+
"loss": 0.0511,
|
| 177 |
"step": 135
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.02,
|
| 181 |
+
"learning_rate": 5.833333333333334e-06,
|
| 182 |
+
"loss": 0.0404,
|
| 183 |
"step": 140
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.02,
|
| 187 |
+
"learning_rate": 6.041666666666667e-06,
|
| 188 |
+
"loss": 0.0422,
|
| 189 |
"step": 145
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.02,
|
| 193 |
+
"learning_rate": 6.25e-06,
|
| 194 |
+
"loss": 0.0345,
|
| 195 |
"step": 150
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.02,
|
| 199 |
+
"learning_rate": 6.458333333333334e-06,
|
| 200 |
+
"loss": 0.0282,
|
| 201 |
"step": 155
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.02,
|
| 205 |
+
"learning_rate": 6.666666666666667e-06,
|
| 206 |
+
"loss": 0.0292,
|
| 207 |
"step": 160
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.02,
|
| 211 |
+
"learning_rate": 6.875e-06,
|
| 212 |
+
"loss": 0.0279,
|
| 213 |
"step": 165
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.02,
|
| 217 |
+
"learning_rate": 7.083333333333335e-06,
|
| 218 |
+
"loss": 0.0236,
|
| 219 |
"step": 170
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.02,
|
| 223 |
+
"learning_rate": 7.291666666666667e-06,
|
| 224 |
+
"loss": 0.0219,
|
| 225 |
"step": 175
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.02,
|
| 229 |
+
"learning_rate": 7.500000000000001e-06,
|
| 230 |
+
"loss": 0.0223,
|
| 231 |
"step": 180
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.02,
|
| 235 |
+
"learning_rate": 7.708333333333334e-06,
|
| 236 |
+
"loss": 0.0225,
|
| 237 |
"step": 185
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.02,
|
| 241 |
+
"learning_rate": 7.916666666666667e-06,
|
| 242 |
+
"loss": 0.0175,
|
| 243 |
"step": 190
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.02,
|
| 247 |
+
"learning_rate": 8.125000000000001e-06,
|
| 248 |
+
"loss": 0.0152,
|
| 249 |
"step": 195
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.02,
|
| 253 |
+
"learning_rate": 8.333333333333334e-06,
|
| 254 |
+
"loss": 0.0153,
|
| 255 |
"step": 200
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.02,
|
| 259 |
+
"learning_rate": 8.541666666666666e-06,
|
| 260 |
+
"loss": 0.0132,
|
| 261 |
"step": 205
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.02,
|
| 265 |
+
"learning_rate": 8.750000000000001e-06,
|
| 266 |
+
"loss": 0.0134,
|
| 267 |
"step": 210
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.02,
|
| 271 |
+
"learning_rate": 8.958333333333334e-06,
|
| 272 |
+
"loss": 0.0144,
|
| 273 |
"step": 215
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.03,
|
| 277 |
+
"learning_rate": 9.166666666666666e-06,
|
| 278 |
+
"loss": 0.0123,
|
| 279 |
"step": 220
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.03,
|
| 283 |
+
"learning_rate": 9.375000000000001e-06,
|
| 284 |
+
"loss": 0.0115,
|
| 285 |
"step": 225
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 0.03,
|
| 289 |
+
"learning_rate": 9.583333333333335e-06,
|
| 290 |
+
"loss": 0.0123,
|
| 291 |
"step": 230
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 0.03,
|
| 295 |
+
"learning_rate": 9.791666666666666e-06,
|
| 296 |
+
"loss": 0.0091,
|
| 297 |
"step": 235
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.03,
|
| 301 |
+
"learning_rate": 1e-05,
|
| 302 |
+
"loss": 0.0087,
|
| 303 |
"step": 240
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 0.03,
|
| 307 |
+
"learning_rate": 9.999867788160888e-06,
|
| 308 |
+
"loss": 0.0095,
|
| 309 |
"step": 245
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.03,
|
| 313 |
+
"learning_rate": 9.999471159635538e-06,
|
| 314 |
+
"loss": 0.0089,
|
| 315 |
"step": 250
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 0.03,
|
| 319 |
+
"learning_rate": 9.998810135399545e-06,
|
| 320 |
+
"loss": 0.0075,
|
| 321 |
"step": 255
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 0.03,
|
| 325 |
+
"learning_rate": 9.997884750411004e-06,
|
| 326 |
+
"loss": 0.0082,
|
| 327 |
"step": 260
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 0.03,
|
| 331 |
+
"learning_rate": 9.996695053608651e-06,
|
| 332 |
+
"loss": 0.007,
|
| 333 |
"step": 265
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.03,
|
| 337 |
+
"learning_rate": 9.99524110790929e-06,
|
| 338 |
+
"loss": 0.0067,
|
| 339 |
"step": 270
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.03,
|
| 343 |
+
"learning_rate": 9.993522990204453e-06,
|
| 344 |
+
"loss": 0.0075,
|
| 345 |
"step": 275
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 0.03,
|
| 349 |
+
"learning_rate": 9.991540791356342e-06,
|
| 350 |
+
"loss": 0.0058,
|
| 351 |
"step": 280
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 0.03,
|
| 355 |
+
"learning_rate": 9.989294616193018e-06,
|
| 356 |
+
"loss": 0.0066,
|
| 357 |
"step": 285
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 0.03,
|
| 361 |
+
"learning_rate": 9.986784583502863e-06,
|
| 362 |
+
"loss": 0.0049,
|
| 363 |
"step": 290
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 0.03,
|
| 367 |
+
"learning_rate": 9.984010826028289e-06,
|
| 368 |
+
"loss": 0.0067,
|
| 369 |
"step": 295
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.03,
|
| 373 |
+
"learning_rate": 9.980973490458728e-06,
|
| 374 |
+
"loss": 0.0055,
|
| 375 |
"step": 300
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.03,
|
| 379 |
+
"learning_rate": 9.97767273742287e-06,
|
| 380 |
+
"loss": 0.0069,
|
| 381 |
"step": 305
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.04,
|
| 385 |
+
"learning_rate": 9.974108741480167e-06,
|
| 386 |
+
"loss": 0.0067,
|
| 387 |
"step": 310
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 0.04,
|
| 391 |
+
"learning_rate": 9.970281691111598e-06,
|
| 392 |
+
"loss": 0.0051,
|
| 393 |
"step": 315
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 0.04,
|
| 397 |
+
"learning_rate": 9.966191788709716e-06,
|
| 398 |
+
"loss": 0.0053,
|
| 399 |
"step": 320
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.04,
|
| 403 |
+
"learning_rate": 9.961839250567925e-06,
|
| 404 |
+
"loss": 0.0051,
|
| 405 |
"step": 325
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 0.04,
|
| 409 |
+
"learning_rate": 9.957224306869053e-06,
|
| 410 |
+
"loss": 0.0047,
|
| 411 |
"step": 330
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 0.04,
|
| 415 |
+
"learning_rate": 9.952347201673181e-06,
|
| 416 |
+
"loss": 0.0053,
|
| 417 |
"step": 335
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.04,
|
| 421 |
+
"learning_rate": 9.947208192904722e-06,
|
| 422 |
+
"loss": 0.0052,
|
| 423 |
"step": 340
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.04,
|
| 427 |
+
"learning_rate": 9.941807552338805e-06,
|
| 428 |
+
"loss": 0.0052,
|
| 429 |
"step": 345
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.04,
|
| 433 |
+
"learning_rate": 9.936145565586871e-06,
|
| 434 |
+
"loss": 0.0046,
|
| 435 |
"step": 350
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 0.04,
|
| 439 |
+
"learning_rate": 9.930222532081597e-06,
|
| 440 |
+
"loss": 0.0048,
|
| 441 |
"step": 355
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 0.04,
|
| 445 |
+
"learning_rate": 9.924038765061042e-06,
|
| 446 |
+
"loss": 0.005,
|
| 447 |
"step": 360
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 0.04,
|
| 451 |
+
"learning_rate": 9.91759459155209e-06,
|
| 452 |
+
"loss": 0.0034,
|
| 453 |
"step": 365
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.04,
|
| 457 |
+
"learning_rate": 9.910890352353154e-06,
|
| 458 |
+
"loss": 0.0042,
|
| 459 |
"step": 370
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.04,
|
| 463 |
+
"learning_rate": 9.903926402016153e-06,
|
| 464 |
+
"loss": 0.0045,
|
| 465 |
"step": 375
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.04,
|
| 469 |
+
"learning_rate": 9.896703108827758e-06,
|
| 470 |
+
"loss": 0.0037,
|
| 471 |
"step": 380
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 0.04,
|
| 475 |
+
"learning_rate": 9.88922085478992e-06,
|
| 476 |
+
"loss": 0.0043,
|
| 477 |
"step": 385
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 0.04,
|
| 481 |
+
"learning_rate": 9.881480035599667e-06,
|
| 482 |
+
"loss": 0.0036,
|
| 483 |
"step": 390
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 0.05,
|
| 487 |
+
"learning_rate": 9.873481060628175e-06,
|
| 488 |
+
"loss": 0.0034,
|
| 489 |
"step": 395
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.05,
|
| 493 |
+
"learning_rate": 9.86522435289912e-06,
|
| 494 |
+
"loss": 0.004,
|
| 495 |
"step": 400
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 0.05,
|
| 499 |
+
"learning_rate": 9.856710349066307e-06,
|
| 500 |
+
"loss": 0.005,
|
| 501 |
"step": 405
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.05,
|
| 505 |
+
"learning_rate": 9.847939499390581e-06,
|
| 506 |
+
"loss": 0.0042,
|
| 507 |
"step": 410
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.05,
|
| 511 |
+
"learning_rate": 9.838912267716006e-06,
|
| 512 |
+
"loss": 0.0033,
|
| 513 |
"step": 415
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 0.05,
|
| 517 |
+
"learning_rate": 9.829629131445342e-06,
|
| 518 |
+
"loss": 0.0032,
|
| 519 |
"step": 420
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.05,
|
| 523 |
+
"learning_rate": 9.820090581514799e-06,
|
| 524 |
+
"loss": 0.0034,
|
| 525 |
"step": 425
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 0.05,
|
| 529 |
+
"learning_rate": 9.810297122368067e-06,
|
| 530 |
+
"loss": 0.0043,
|
| 531 |
"step": 430
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 0.05,
|
| 535 |
+
"learning_rate": 9.800249271929645e-06,
|
| 536 |
+
"loss": 0.004,
|
| 537 |
"step": 435
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.05,
|
| 541 |
+
"learning_rate": 9.789947561577445e-06,
|
| 542 |
+
"loss": 0.0036,
|
| 543 |
"step": 440
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.05,
|
| 547 |
+
"learning_rate": 9.779392536114698e-06,
|
| 548 |
+
"loss": 0.0029,
|
| 549 |
"step": 445
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.05,
|
| 553 |
+
"learning_rate": 9.768584753741134e-06,
|
| 554 |
+
"loss": 0.004,
|
| 555 |
"step": 450
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.05,
|
| 559 |
+
"learning_rate": 9.75752478602347e-06,
|
| 560 |
+
"loss": 0.0035,
|
| 561 |
"step": 455
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.05,
|
| 565 |
+
"learning_rate": 9.74621321786517e-06,
|
| 566 |
+
"loss": 0.0026,
|
| 567 |
"step": 460
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.05,
|
| 571 |
+
"learning_rate": 9.73465064747553e-06,
|
| 572 |
+
"loss": 0.0024,
|
| 573 |
"step": 465
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 0.05,
|
| 577 |
+
"learning_rate": 9.722837686338025e-06,
|
| 578 |
+
"loss": 0.0027,
|
| 579 |
"step": 470
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.05,
|
| 583 |
+
"learning_rate": 9.710774959177983e-06,
|
| 584 |
+
"loss": 0.0031,
|
| 585 |
"step": 475
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 0.06,
|
| 589 |
+
"learning_rate": 9.698463103929542e-06,
|
| 590 |
+
"loss": 0.0034,
|
| 591 |
"step": 480
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.06,
|
| 595 |
+
"learning_rate": 9.685902771701913e-06,
|
| 596 |
+
"loss": 0.0044,
|
| 597 |
"step": 485
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.06,
|
| 601 |
+
"learning_rate": 9.673094626744944e-06,
|
| 602 |
+
"loss": 0.0028,
|
| 603 |
"step": 490
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 0.06,
|
| 607 |
+
"learning_rate": 9.660039346413994e-06,
|
| 608 |
+
"loss": 0.003,
|
| 609 |
"step": 495
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.06,
|
| 613 |
+
"learning_rate": 9.646737621134112e-06,
|
| 614 |
+
"loss": 0.0027,
|
| 615 |
"step": 500
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.06,
|
| 619 |
+
"learning_rate": 9.633190154363527e-06,
|
| 620 |
+
"loss": 0.0042,
|
| 621 |
"step": 505
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 0.06,
|
| 625 |
+
"learning_rate": 9.619397662556434e-06,
|
| 626 |
+
"loss": 0.0024,
|
| 627 |
"step": 510
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.06,
|
| 631 |
+
"learning_rate": 9.605360875125119e-06,
|
| 632 |
+
"loss": 0.0023,
|
| 633 |
"step": 515
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.06,
|
| 637 |
+
"learning_rate": 9.591080534401371e-06,
|
| 638 |
+
"loss": 0.0034,
|
| 639 |
"step": 520
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.06,
|
| 643 |
+
"learning_rate": 9.576557395597237e-06,
|
| 644 |
+
"loss": 0.002,
|
| 645 |
"step": 525
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 0.06,
|
| 649 |
+
"learning_rate": 9.561792226765072e-06,
|
| 650 |
+
"loss": 0.0034,
|
| 651 |
"step": 530
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 0.06,
|
| 655 |
+
"learning_rate": 9.546785808756926e-06,
|
| 656 |
+
"loss": 0.0028,
|
| 657 |
"step": 535
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.06,
|
| 661 |
+
"learning_rate": 9.531538935183252e-06,
|
| 662 |
+
"loss": 0.0033,
|
| 663 |
"step": 540
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 0.06,
|
| 667 |
+
"learning_rate": 9.516052412370922e-06,
|
| 668 |
+
"loss": 0.0028,
|
| 669 |
"step": 545
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.06,
|
| 673 |
+
"learning_rate": 9.500327059320606e-06,
|
| 674 |
+
"loss": 0.0032,
|
| 675 |
"step": 550
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.06,
|
| 679 |
+
"learning_rate": 9.484363707663443e-06,
|
| 680 |
+
"loss": 0.0035,
|
| 681 |
"step": 555
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 0.06,
|
| 685 |
+
"learning_rate": 9.468163201617063e-06,
|
| 686 |
+
"loss": 0.0029,
|
| 687 |
"step": 560
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 0.06,
|
| 691 |
+
"learning_rate": 9.451726397940946e-06,
|
| 692 |
+
"loss": 0.0032,
|
| 693 |
"step": 565
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 0.07,
|
| 697 |
+
"learning_rate": 9.43505416589111e-06,
|
| 698 |
+
"loss": 0.0028,
|
| 699 |
"step": 570
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.07,
|
| 703 |
+
"learning_rate": 9.41814738717414e-06,
|
| 704 |
+
"loss": 0.0021,
|
| 705 |
"step": 575
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 0.07,
|
| 709 |
+
"learning_rate": 9.401006955900555e-06,
|
| 710 |
+
"loss": 0.002,
|
| 711 |
"step": 580
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 0.07,
|
| 715 |
+
"learning_rate": 9.38363377853754e-06,
|
| 716 |
+
"loss": 0.0023,
|
| 717 |
"step": 585
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.07,
|
| 721 |
+
"learning_rate": 9.366028773860981e-06,
|
| 722 |
+
"loss": 0.0024,
|
| 723 |
"step": 590
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 0.07,
|
| 727 |
+
"learning_rate": 9.348192872906896e-06,
|
| 728 |
+
"loss": 0.0029,
|
| 729 |
"step": 595
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.07,
|
| 733 |
+
"learning_rate": 9.330127018922195e-06,
|
| 734 |
+
"loss": 0.0014,
|
| 735 |
"step": 600
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 0.07,
|
| 739 |
+
"learning_rate": 9.311832167314788e-06,
|
| 740 |
+
"loss": 0.0021,
|
| 741 |
"step": 605
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.07,
|
| 745 |
+
"learning_rate": 9.293309285603066e-06,
|
| 746 |
+
"loss": 0.0016,
|
| 747 |
"step": 610
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 0.07,
|
| 751 |
+
"learning_rate": 9.274559353364734e-06,
|
| 752 |
+
"loss": 0.0021,
|
| 753 |
"step": 615
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 0.07,
|
| 757 |
+
"learning_rate": 9.255583362184998e-06,
|
| 758 |
+
"loss": 0.0025,
|
| 759 |
"step": 620
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.07,
|
| 763 |
+
"learning_rate": 9.23638231560414e-06,
|
| 764 |
+
"loss": 0.0021,
|
| 765 |
"step": 625
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 0.07,
|
| 769 |
+
"learning_rate": 9.21695722906443e-06,
|
| 770 |
+
"loss": 0.0026,
|
| 771 |
"step": 630
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 0.07,
|
| 775 |
+
"learning_rate": 9.197309129856433e-06,
|
| 776 |
+
"loss": 0.0026,
|
| 777 |
"step": 635
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 0.07,
|
| 781 |
+
"learning_rate": 9.177439057064684e-06,
|
| 782 |
+
"loss": 0.0026,
|
| 783 |
"step": 640
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.07,
|
| 787 |
+
"learning_rate": 9.157348061512728e-06,
|
| 788 |
+
"loss": 0.0027,
|
| 789 |
"step": 645
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.07,
|
| 793 |
+
"learning_rate": 9.137037205707552e-06,
|
| 794 |
+
"loss": 0.0015,
|
| 795 |
"step": 650
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.08,
|
| 799 |
+
"learning_rate": 9.116507563783402e-06,
|
| 800 |
+
"loss": 0.0015,
|
| 801 |
"step": 655
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 0.08,
|
| 805 |
+
"learning_rate": 9.09576022144496e-06,
|
| 806 |
+
"loss": 0.0019,
|
| 807 |
"step": 660
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 0.08,
|
| 811 |
+
"learning_rate": 9.074796275909941e-06,
|
| 812 |
+
"loss": 0.0024,
|
| 813 |
"step": 665
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 0.08,
|
| 817 |
+
"learning_rate": 9.053616835851062e-06,
|
| 818 |
+
"loss": 0.0025,
|
| 819 |
"step": 670
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 0.08,
|
| 823 |
+
"learning_rate": 9.032223021337415e-06,
|
| 824 |
+
"loss": 0.0021,
|
| 825 |
"step": 675
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 0.08,
|
| 829 |
+
"learning_rate": 9.01061596377522e-06,
|
| 830 |
+
"loss": 0.0021,
|
| 831 |
"step": 680
|
| 832 |
},
|
| 833 |
{
|
| 834 |
"epoch": 0.08,
|
| 835 |
+
"learning_rate": 8.988796805848008e-06,
|
| 836 |
+
"loss": 0.0022,
|
| 837 |
"step": 685
|
| 838 |
},
|
| 839 |
{
|
| 840 |
"epoch": 0.08,
|
| 841 |
+
"learning_rate": 8.966766701456177e-06,
|
| 842 |
+
"loss": 0.0025,
|
| 843 |
"step": 690
|
| 844 |
},
|
| 845 |
{
|
| 846 |
"epoch": 0.08,
|
| 847 |
+
"learning_rate": 8.944526815655974e-06,
|
| 848 |
+
"loss": 0.0019,
|
| 849 |
"step": 695
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 0.08,
|
| 853 |
+
"learning_rate": 8.92207832459788e-06,
|
| 854 |
+
"loss": 0.0026,
|
| 855 |
"step": 700
|
| 856 |
},
|
| 857 |
{
|
| 858 |
"epoch": 0.08,
|
| 859 |
+
"learning_rate": 8.899422415464409e-06,
|
| 860 |
+
"loss": 0.0019,
|
| 861 |
"step": 705
|
| 862 |
},
|
| 863 |
{
|
| 864 |
"epoch": 0.08,
|
| 865 |
+
"learning_rate": 8.87656028640733e-06,
|
| 866 |
+
"loss": 0.0019,
|
| 867 |
"step": 710
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 0.08,
|
| 871 |
+
"learning_rate": 8.85349314648429e-06,
|
| 872 |
+
"loss": 0.0015,
|
| 873 |
"step": 715
|
| 874 |
},
|
| 875 |
{
|
| 876 |
"epoch": 0.08,
|
| 877 |
+
"learning_rate": 8.83022221559489e-06,
|
| 878 |
+
"loss": 0.0017,
|
| 879 |
"step": 720
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 0.08,
|
| 883 |
+
"learning_rate": 8.806748724416156e-06,
|
| 884 |
+
"loss": 0.002,
|
| 885 |
"step": 725
|
| 886 |
},
|
| 887 |
{
|
| 888 |
"epoch": 0.08,
|
| 889 |
+
"learning_rate": 8.783073914337465e-06,
|
| 890 |
+
"loss": 0.0023,
|
| 891 |
"step": 730
|
| 892 |
},
|
| 893 |
{
|
| 894 |
"epoch": 0.08,
|
| 895 |
+
"learning_rate": 8.759199037394888e-06,
|
| 896 |
+
"loss": 0.0019,
|
| 897 |
"step": 735
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 0.08,
|
| 901 |
+
"learning_rate": 8.735125356204982e-06,
|
| 902 |
+
"loss": 0.0018,
|
| 903 |
"step": 740
|
| 904 |
},
|
| 905 |
{
|
| 906 |
"epoch": 0.09,
|
| 907 |
+
"learning_rate": 8.710854143898008e-06,
|
| 908 |
+
"loss": 0.0018,
|
| 909 |
"step": 745
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 0.09,
|
| 913 |
+
"learning_rate": 8.68638668405062e-06,
|
| 914 |
+
"loss": 0.0017,
|
| 915 |
"step": 750
|
| 916 |
},
|
| 917 |
{
|
| 918 |
"epoch": 0.09,
|
| 919 |
+
"learning_rate": 8.661724270617961e-06,
|
| 920 |
+
"loss": 0.002,
|
| 921 |
"step": 755
|
| 922 |
},
|
| 923 |
{
|
| 924 |
"epoch": 0.09,
|
| 925 |
+
"learning_rate": 8.636868207865244e-06,
|
| 926 |
+
"loss": 0.0023,
|
| 927 |
"step": 760
|
| 928 |
},
|
| 929 |
{
|
| 930 |
"epoch": 0.09,
|
| 931 |
+
"learning_rate": 8.611819810298778e-06,
|
| 932 |
+
"loss": 0.0019,
|
| 933 |
"step": 765
|
| 934 |
},
|
| 935 |
{
|
| 936 |
"epoch": 0.09,
|
| 937 |
+
"learning_rate": 8.586580402596448e-06,
|
| 938 |
+
"loss": 0.0017,
|
| 939 |
"step": 770
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.09,
|
| 943 |
+
"learning_rate": 8.561151319537656e-06,
|
| 944 |
+
"loss": 0.0018,
|
| 945 |
"step": 775
|
| 946 |
},
|
| 947 |
{
|
| 948 |
"epoch": 0.09,
|
| 949 |
+
"learning_rate": 8.535533905932739e-06,
|
| 950 |
+
"loss": 0.0019,
|
| 951 |
"step": 780
|
| 952 |
},
|
| 953 |
{
|
| 954 |
"epoch": 0.09,
|
| 955 |
+
"learning_rate": 8.509729516551842e-06,
|
| 956 |
+
"loss": 0.0022,
|
| 957 |
"step": 785
|
| 958 |
},
|
| 959 |
{
|
| 960 |
"epoch": 0.09,
|
| 961 |
+
"learning_rate": 8.483739516053276e-06,
|
| 962 |
+
"loss": 0.0015,
|
| 963 |
"step": 790
|
| 964 |
},
|
| 965 |
{
|
| 966 |
"epoch": 0.09,
|
| 967 |
+
"learning_rate": 8.457565278911349e-06,
|
| 968 |
+
"loss": 0.0017,
|
| 969 |
"step": 795
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 0.09,
|
| 973 |
+
"learning_rate": 8.43120818934367e-06,
|
| 974 |
+
"loss": 0.0024,
|
| 975 |
"step": 800
|
| 976 |
},
|
| 977 |
{
|
| 978 |
"epoch": 0.09,
|
| 979 |
+
"learning_rate": 8.404669641237952e-06,
|
| 980 |
+
"loss": 0.0013,
|
| 981 |
"step": 805
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 0.09,
|
| 985 |
+
"learning_rate": 8.377951038078303e-06,
|
| 986 |
+
"loss": 0.0017,
|
| 987 |
"step": 810
|
| 988 |
},
|
| 989 |
{
|
| 990 |
"epoch": 0.09,
|
| 991 |
+
"learning_rate": 8.35105379287098e-06,
|
| 992 |
+
"loss": 0.002,
|
| 993 |
"step": 815
|
| 994 |
},
|
| 995 |
{
|
| 996 |
"epoch": 0.09,
|
| 997 |
+
"learning_rate": 8.323979328069689e-06,
|
| 998 |
+
"loss": 0.0015,
|
| 999 |
"step": 820
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 0.09,
|
| 1003 |
+
"learning_rate": 8.296729075500345e-06,
|
| 1004 |
+
"loss": 0.0026,
|
| 1005 |
"step": 825
|
| 1006 |
},
|
| 1007 |
{
|
| 1008 |
"epoch": 0.1,
|
| 1009 |
+
"learning_rate": 8.269304476285349e-06,
|
| 1010 |
+
"loss": 0.0017,
|
| 1011 |
"step": 830
|
| 1012 |
},
|
| 1013 |
{
|
| 1014 |
"epoch": 0.1,
|
| 1015 |
+
"learning_rate": 8.241706980767382e-06,
|
| 1016 |
+
"loss": 0.0017,
|
| 1017 |
"step": 835
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
"epoch": 0.1,
|
| 1021 |
+
"learning_rate": 8.213938048432697e-06,
|
| 1022 |
+
"loss": 0.0011,
|
| 1023 |
"step": 840
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 0.1,
|
| 1027 |
+
"learning_rate": 8.185999147833943e-06,
|
| 1028 |
+
"loss": 0.0013,
|
| 1029 |
"step": 845
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 0.1,
|
| 1033 |
+
"learning_rate": 8.157891756512488e-06,
|
| 1034 |
+
"loss": 0.0016,
|
| 1035 |
"step": 850
|
| 1036 |
},
|
| 1037 |
{
|
| 1038 |
"epoch": 0.1,
|
| 1039 |
+
"learning_rate": 8.129617360920297e-06,
|
| 1040 |
+
"loss": 0.0017,
|
| 1041 |
"step": 855
|
| 1042 |
},
|
| 1043 |
{
|
| 1044 |
"epoch": 0.1,
|
| 1045 |
+
"learning_rate": 8.101177456341301e-06,
|
| 1046 |
+
"loss": 0.0018,
|
| 1047 |
"step": 860
|
| 1048 |
},
|
| 1049 |
{
|
| 1050 |
"epoch": 0.1,
|
| 1051 |
+
"learning_rate": 8.072573546812338e-06,
|
| 1052 |
+
"loss": 0.0012,
|
| 1053 |
"step": 865
|
| 1054 |
},
|
| 1055 |
{
|
| 1056 |
"epoch": 0.1,
|
| 1057 |
+
"learning_rate": 8.043807145043604e-06,
|
| 1058 |
+
"loss": 0.002,
|
| 1059 |
"step": 870
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 0.1,
|
| 1063 |
+
"learning_rate": 8.014879772338649e-06,
|
| 1064 |
+
"loss": 0.0012,
|
| 1065 |
"step": 875
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.1,
|
| 1069 |
+
"learning_rate": 7.985792958513932e-06,
|
| 1070 |
+
"loss": 0.0017,
|
| 1071 |
"step": 880
|
| 1072 |
},
|
| 1073 |
{
|
| 1074 |
"epoch": 0.1,
|
| 1075 |
+
"learning_rate": 7.956548241817914e-06,
|
| 1076 |
+
"loss": 0.0015,
|
| 1077 |
"step": 885
|
| 1078 |
},
|
| 1079 |
{
|
| 1080 |
"epoch": 0.1,
|
| 1081 |
+
"learning_rate": 7.927147168849704e-06,
|
| 1082 |
+
"loss": 0.0013,
|
| 1083 |
"step": 890
|
| 1084 |
},
|
| 1085 |
{
|
| 1086 |
"epoch": 0.1,
|
| 1087 |
+
"learning_rate": 7.897591294477276e-06,
|
| 1088 |
+
"loss": 0.0016,
|
| 1089 |
"step": 895
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 0.1,
|
| 1093 |
+
"learning_rate": 7.86788218175523e-06,
|
| 1094 |
+
"loss": 0.0014,
|
| 1095 |
"step": 900
|
| 1096 |
},
|
| 1097 |
{
|
| 1098 |
"epoch": 0.1,
|
| 1099 |
+
"learning_rate": 7.838021401842145e-06,
|
| 1100 |
+
"loss": 0.0013,
|
| 1101 |
"step": 905
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
"epoch": 0.1,
|
| 1105 |
+
"learning_rate": 7.808010533917465e-06,
|
| 1106 |
+
"loss": 0.0014,
|
| 1107 |
"step": 910
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 0.1,
|
| 1111 |
+
"learning_rate": 7.777851165098012e-06,
|
| 1112 |
+
"loss": 0.0017,
|
| 1113 |
"step": 915
|
| 1114 |
},
|
| 1115 |
{
|
| 1116 |
"epoch": 0.11,
|
| 1117 |
+
"learning_rate": 7.747544890354031e-06,
|
| 1118 |
+
"loss": 0.0017,
|
| 1119 |
"step": 920
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 0.11,
|
| 1123 |
+
"learning_rate": 7.71709331242485e-06,
|
| 1124 |
+
"loss": 0.0014,
|
| 1125 |
"step": 925
|
| 1126 |
},
|
| 1127 |
{
|
| 1128 |
"epoch": 0.11,
|
| 1129 |
+
"learning_rate": 7.686498041734121e-06,
|
| 1130 |
+
"loss": 0.0019,
|
| 1131 |
"step": 930
|
| 1132 |
},
|
| 1133 |
{
|
| 1134 |
"epoch": 0.11,
|
| 1135 |
+
"learning_rate": 7.655760696304642e-06,
|
| 1136 |
+
"loss": 0.0012,
|
| 1137 |
"step": 935
|
| 1138 |
},
|
| 1139 |
{
|
| 1140 |
"epoch": 0.11,
|
| 1141 |
+
"learning_rate": 7.624882901672801e-06,
|
| 1142 |
+
"loss": 0.0017,
|
| 1143 |
"step": 940
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
"epoch": 0.11,
|
| 1147 |
+
"learning_rate": 7.593866290802608e-06,
|
| 1148 |
+
"loss": 0.0012,
|
| 1149 |
"step": 945
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.11,
|
| 1153 |
+
"learning_rate": 7.562712503999327e-06,
|
| 1154 |
+
"loss": 0.0009,
|
| 1155 |
"step": 950
|
| 1156 |
},
|
| 1157 |
{
|
| 1158 |
"epoch": 0.11,
|
| 1159 |
+
"learning_rate": 7.531423188822738e-06,
|
| 1160 |
+
"loss": 0.0014,
|
| 1161 |
"step": 955
|
| 1162 |
},
|
| 1163 |
{
|
| 1164 |
"epoch": 0.11,
|
| 1165 |
+
"learning_rate": 7.500000000000001e-06,
|
| 1166 |
+
"loss": 0.0014,
|
| 1167 |
"step": 960
|
| 1168 |
},
|
| 1169 |
{
|
| 1170 |
"epoch": 0.11,
|
| 1171 |
+
"learning_rate": 7.468444599338152e-06,
|
| 1172 |
+
"loss": 0.0012,
|
| 1173 |
"step": 965
|
| 1174 |
},
|
| 1175 |
{
|
| 1176 |
"epoch": 0.11,
|
| 1177 |
+
"learning_rate": 7.4367586556362125e-06,
|
| 1178 |
+
"loss": 0.0011,
|
| 1179 |
"step": 970
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.11,
|
| 1183 |
+
"learning_rate": 7.404943844596939e-06,
|
| 1184 |
+
"loss": 0.0011,
|
| 1185 |
"step": 975
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
"epoch": 0.11,
|
| 1189 |
+
"learning_rate": 7.373001848738203e-06,
|
| 1190 |
+
"loss": 0.0015,
|
| 1191 |
"step": 980
|
| 1192 |
},
|
| 1193 |
{
|
| 1194 |
"epoch": 0.11,
|
| 1195 |
+
"learning_rate": 7.340934357304011e-06,
|
| 1196 |
+
"loss": 0.002,
|
| 1197 |
"step": 985
|
| 1198 |
},
|
| 1199 |
{
|
| 1200 |
"epoch": 0.11,
|
| 1201 |
+
"learning_rate": 7.308743066175172e-06,
|
| 1202 |
+
"loss": 0.0012,
|
| 1203 |
"step": 990
|
| 1204 |
},
|
| 1205 |
{
|
| 1206 |
"epoch": 0.11,
|
| 1207 |
+
"learning_rate": 7.276429677779603e-06,
|
| 1208 |
+
"loss": 0.0016,
|
| 1209 |
"step": 995
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 0.11,
|
| 1213 |
+
"learning_rate": 7.243995901002312e-06,
|
| 1214 |
+
"loss": 0.0019,
|
| 1215 |
"step": 1000
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
"epoch": 0.12,
|
| 1219 |
+
"learning_rate": 7.211443451095007e-06,
|
| 1220 |
+
"loss": 0.0013,
|
| 1221 |
"step": 1005
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 0.12,
|
| 1225 |
+
"learning_rate": 7.178774049585397e-06,
|
| 1226 |
+
"loss": 0.0014,
|
| 1227 |
"step": 1010
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
"epoch": 0.12,
|
| 1231 |
+
"learning_rate": 7.145989424186146e-06,
|
| 1232 |
+
"loss": 0.0021,
|
| 1233 |
"step": 1015
|
| 1234 |
},
|
| 1235 |
{
|
| 1236 |
"epoch": 0.12,
|
| 1237 |
+
"learning_rate": 7.113091308703498e-06,
|
| 1238 |
+
"loss": 0.0012,
|
| 1239 |
"step": 1020
|
| 1240 |
},
|
| 1241 |
{
|
| 1242 |
"epoch": 0.12,
|
| 1243 |
+
"learning_rate": 7.080081442945597e-06,
|
| 1244 |
+
"loss": 0.0014,
|
| 1245 |
"step": 1025
|
| 1246 |
},
|
| 1247 |
{
|
| 1248 |
"epoch": 0.12,
|
| 1249 |
+
"learning_rate": 7.046961572630463e-06,
|
| 1250 |
+
"loss": 0.0013,
|
| 1251 |
"step": 1030
|
| 1252 |
},
|
| 1253 |
{
|
| 1254 |
"epoch": 0.12,
|
| 1255 |
+
"learning_rate": 7.0137334492936875e-06,
|
| 1256 |
+
"loss": 0.0014,
|
| 1257 |
"step": 1035
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
"epoch": 0.12,
|
| 1261 |
+
"learning_rate": 6.980398830195785e-06,
|
| 1262 |
+
"loss": 0.0014,
|
| 1263 |
"step": 1040
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 0.12,
|
| 1267 |
+
"learning_rate": 6.946959478229277e-06,
|
| 1268 |
+
"loss": 0.0013,
|
| 1269 |
"step": 1045
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
"epoch": 0.12,
|
| 1273 |
+
"learning_rate": 6.913417161825449e-06,
|
| 1274 |
+
"loss": 0.001,
|
| 1275 |
"step": 1050
|
| 1276 |
},
|
| 1277 |
{
|
| 1278 |
"epoch": 0.12,
|
| 1279 |
+
"learning_rate": 6.8797736548608405e-06,
|
| 1280 |
+
"loss": 0.0019,
|
| 1281 |
"step": 1055
|
| 1282 |
},
|
| 1283 |
{
|
| 1284 |
"epoch": 0.12,
|
| 1285 |
+
"learning_rate": 6.8460307365634225e-06,
|
| 1286 |
+
"loss": 0.0014,
|
| 1287 |
"step": 1060
|
| 1288 |
},
|
| 1289 |
{
|
| 1290 |
"epoch": 0.12,
|
| 1291 |
+
"learning_rate": 6.812190191418508e-06,
|
| 1292 |
+
"loss": 0.0013,
|
| 1293 |
"step": 1065
|
| 1294 |
},
|
| 1295 |
{
|
| 1296 |
"epoch": 0.12,
|
| 1297 |
+
"learning_rate": 6.778253809074384e-06,
|
| 1298 |
+
"loss": 0.0009,
|
| 1299 |
"step": 1070
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
"epoch": 0.12,
|
| 1303 |
+
"learning_rate": 6.7442233842476545e-06,
|
| 1304 |
+
"loss": 0.0015,
|
| 1305 |
"step": 1075
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 0.12,
|
| 1309 |
+
"learning_rate": 6.710100716628345e-06,
|
| 1310 |
+
"loss": 0.0009,
|
| 1311 |
"step": 1080
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
"epoch": 0.12,
|
| 1315 |
+
"learning_rate": 6.675887610784708e-06,
|
| 1316 |
+
"loss": 0.0014,
|
| 1317 |
"step": 1085
|
| 1318 |
},
|
| 1319 |
{
|
| 1320 |
"epoch": 0.13,
|
| 1321 |
+
"learning_rate": 6.641585876067807e-06,
|
| 1322 |
+
"loss": 0.0015,
|
| 1323 |
"step": 1090
|
| 1324 |
},
|
| 1325 |
{
|
| 1326 |
"epoch": 0.13,
|
| 1327 |
+
"learning_rate": 6.607197326515808e-06,
|
| 1328 |
+
"loss": 0.001,
|
| 1329 |
"step": 1095
|
| 1330 |
},
|
| 1331 |
{
|
| 1332 |
"epoch": 0.13,
|
| 1333 |
+
"learning_rate": 6.572723780758069e-06,
|
| 1334 |
+
"loss": 0.0008,
|
| 1335 |
"step": 1100
|
| 1336 |
},
|
| 1337 |
{
|
| 1338 |
"epoch": 0.13,
|
| 1339 |
+
"learning_rate": 6.538167061918942e-06,
|
| 1340 |
+
"loss": 0.0011,
|
| 1341 |
"step": 1105
|
| 1342 |
},
|
| 1343 |
{
|
| 1344 |
"epoch": 0.13,
|
| 1345 |
+
"learning_rate": 6.503528997521365e-06,
|
| 1346 |
+
"loss": 0.0017,
|
| 1347 |
"step": 1110
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 0.13,
|
| 1351 |
+
"learning_rate": 6.468811419390222e-06,
|
| 1352 |
+
"loss": 0.0016,
|
| 1353 |
"step": 1115
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
"epoch": 0.13,
|
| 1357 |
+
"learning_rate": 6.434016163555452e-06,
|
| 1358 |
+
"loss": 0.0012,
|
| 1359 |
"step": 1120
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
"epoch": 0.13,
|
| 1363 |
+
"learning_rate": 6.399145070154962e-06,
|
| 1364 |
+
"loss": 0.0014,
|
| 1365 |
"step": 1125
|
| 1366 |
},
|
| 1367 |
{
|
| 1368 |
"epoch": 0.13,
|
| 1369 |
+
"learning_rate": 6.364199983337306e-06,
|
| 1370 |
+
"loss": 0.0012,
|
| 1371 |
"step": 1130
|
| 1372 |
},
|
| 1373 |
{
|
| 1374 |
"epoch": 0.13,
|
| 1375 |
+
"learning_rate": 6.329182751164164e-06,
|
| 1376 |
+
"loss": 0.0013,
|
| 1377 |
"step": 1135
|
| 1378 |
},
|
| 1379 |
{
|
| 1380 |
"epoch": 0.13,
|
| 1381 |
+
"learning_rate": 6.294095225512604e-06,
|
| 1382 |
+
"loss": 0.0011,
|
| 1383 |
"step": 1140
|
| 1384 |
},
|
| 1385 |
{
|
| 1386 |
"epoch": 0.13,
|
| 1387 |
+
"learning_rate": 6.2589392619771435e-06,
|
| 1388 |
+
"loss": 0.0012,
|
| 1389 |
"step": 1145
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 0.13,
|
| 1393 |
+
"learning_rate": 6.2237167197716195e-06,
|
| 1394 |
+
"loss": 0.0011,
|
| 1395 |
"step": 1150
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
"epoch": 0.13,
|
| 1399 |
+
"learning_rate": 6.188429461630866e-06,
|
| 1400 |
+
"loss": 0.0012,
|
| 1401 |
"step": 1155
|
| 1402 |
},
|
| 1403 |
{
|
| 1404 |
"epoch": 0.13,
|
| 1405 |
+
"learning_rate": 6.153079353712201e-06,
|
| 1406 |
+
"loss": 0.0016,
|
| 1407 |
"step": 1160
|
| 1408 |
},
|
| 1409 |
{
|
| 1410 |
"epoch": 0.13,
|
| 1411 |
+
"learning_rate": 6.117668265496738e-06,
|
| 1412 |
+
"loss": 0.0009,
|
| 1413 |
"step": 1165
|
| 1414 |
},
|
| 1415 |
{
|
| 1416 |
"epoch": 0.13,
|
| 1417 |
+
"learning_rate": 6.0821980696905145e-06,
|
| 1418 |
+
"loss": 0.0016,
|
| 1419 |
"step": 1170
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
"epoch": 0.13,
|
| 1423 |
+
"learning_rate": 6.046670642125461e-06,
|
| 1424 |
+
"loss": 0.0009,
|
| 1425 |
"step": 1175
|
| 1426 |
},
|
| 1427 |
{
|
| 1428 |
"epoch": 0.14,
|
| 1429 |
+
"learning_rate": 6.011087861660191e-06,
|
| 1430 |
+
"loss": 0.0015,
|
| 1431 |
"step": 1180
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 0.14,
|
| 1435 |
+
"learning_rate": 5.975451610080643e-06,
|
| 1436 |
+
"loss": 0.0015,
|
| 1437 |
"step": 1185
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
"epoch": 0.14,
|
| 1441 |
+
"learning_rate": 5.93976377200056e-06,
|
| 1442 |
+
"loss": 0.0013,
|
| 1443 |
"step": 1190
|
| 1444 |
},
|
| 1445 |
{
|
| 1446 |
"epoch": 0.14,
|
| 1447 |
+
"learning_rate": 5.904026234761827e-06,
|
| 1448 |
+
"loss": 0.001,
|
| 1449 |
"step": 1195
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
"epoch": 0.14,
|
| 1453 |
+
"learning_rate": 5.8682408883346535e-06,
|
| 1454 |
+
"loss": 0.0013,
|
| 1455 |
"step": 1200
|
| 1456 |
},
|
| 1457 |
{
|
| 1458 |
"epoch": 0.14,
|
| 1459 |
+
"learning_rate": 5.832409625217623e-06,
|
| 1460 |
+
"loss": 0.0014,
|
| 1461 |
"step": 1205
|
| 1462 |
},
|
| 1463 |
{
|
| 1464 |
"epoch": 0.14,
|
| 1465 |
+
"learning_rate": 5.796534340337614e-06,
|
| 1466 |
+
"loss": 0.0017,
|
| 1467 |
"step": 1210
|
| 1468 |
},
|
| 1469 |
{
|
| 1470 |
"epoch": 0.14,
|
| 1471 |
+
"learning_rate": 5.760616930949584e-06,
|
| 1472 |
+
"loss": 0.0015,
|
| 1473 |
"step": 1215
|
| 1474 |
},
|
| 1475 |
{
|
| 1476 |
"epoch": 0.14,
|
| 1477 |
+
"learning_rate": 5.724659296536234e-06,
|
| 1478 |
+
"loss": 0.001,
|
| 1479 |
"step": 1220
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
"epoch": 0.14,
|
| 1483 |
+
"learning_rate": 5.688663338707554e-06,
|
| 1484 |
+
"loss": 0.0013,
|
| 1485 |
"step": 1225
|
| 1486 |
},
|
| 1487 |
{
|
| 1488 |
"epoch": 0.14,
|
| 1489 |
+
"learning_rate": 5.65263096110026e-06,
|
| 1490 |
+
"loss": 0.001,
|
| 1491 |
"step": 1230
|
| 1492 |
},
|
| 1493 |
{
|
| 1494 |
"epoch": 0.14,
|
| 1495 |
+
"learning_rate": 5.616564069277111e-06,
|
| 1496 |
+
"loss": 0.0012,
|
| 1497 |
"step": 1235
|
| 1498 |
},
|
| 1499 |
{
|
| 1500 |
"epoch": 0.14,
|
| 1501 |
+
"learning_rate": 5.5804645706261515e-06,
|
| 1502 |
+
"loss": 0.0021,
|
| 1503 |
"step": 1240
|
| 1504 |
},
|
| 1505 |
{
|
| 1506 |
"epoch": 0.14,
|
| 1507 |
+
"learning_rate": 5.544334374259823e-06,
|
| 1508 |
+
"loss": 0.0012,
|
| 1509 |
"step": 1245
|
| 1510 |
},
|
| 1511 |
{
|
| 1512 |
"epoch": 0.14,
|
| 1513 |
+
"learning_rate": 5.50817539091401e-06,
|
| 1514 |
+
"loss": 0.0015,
|
| 1515 |
"step": 1250
|
| 1516 |
},
|
| 1517 |
{
|
| 1518 |
"epoch": 0.14,
|
| 1519 |
+
"learning_rate": 5.471989532846987e-06,
|
| 1520 |
+
"loss": 0.0008,
|
| 1521 |
"step": 1255
|
| 1522 |
},
|
| 1523 |
{
|
| 1524 |
"epoch": 0.14,
|
| 1525 |
+
"learning_rate": 5.435778713738292e-06,
|
| 1526 |
+
"loss": 0.0009,
|
| 1527 |
"step": 1260
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
"epoch": 0.15,
|
| 1531 |
+
"learning_rate": 5.3995448485875205e-06,
|
| 1532 |
+
"loss": 0.0013,
|
| 1533 |
"step": 1265
|
| 1534 |
},
|
| 1535 |
{
|
| 1536 |
"epoch": 0.15,
|
| 1537 |
+
"learning_rate": 5.363289853613054e-06,
|
| 1538 |
+
"loss": 0.0012,
|
| 1539 |
"step": 1270
|
| 1540 |
},
|
| 1541 |
{
|
| 1542 |
"epoch": 0.15,
|
| 1543 |
+
"learning_rate": 5.327015646150716e-06,
|
| 1544 |
+
"loss": 0.0017,
|
| 1545 |
"step": 1275
|
| 1546 |
},
|
| 1547 |
{
|
| 1548 |
"epoch": 0.15,
|
| 1549 |
+
"learning_rate": 5.290724144552379e-06,
|
| 1550 |
+
"loss": 0.0009,
|
| 1551 |
"step": 1280
|
| 1552 |
},
|
| 1553 |
{
|
| 1554 |
"epoch": 0.15,
|
| 1555 |
+
"learning_rate": 5.254417268084514e-06,
|
| 1556 |
+
"loss": 0.0009,
|
| 1557 |
"step": 1285
|
| 1558 |
},
|
| 1559 |
{
|
| 1560 |
"epoch": 0.15,
|
| 1561 |
+
"learning_rate": 5.218096936826681e-06,
|
| 1562 |
+
"loss": 0.0011,
|
| 1563 |
"step": 1290
|
| 1564 |
},
|
| 1565 |
{
|
| 1566 |
"epoch": 0.15,
|
| 1567 |
+
"learning_rate": 5.18176507157e-06,
|
| 1568 |
+
"loss": 0.0009,
|
| 1569 |
"step": 1295
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
"epoch": 0.15,
|
| 1573 |
+
"learning_rate": 5.145423593715558e-06,
|
| 1574 |
+
"loss": 0.0012,
|
| 1575 |
"step": 1300
|
| 1576 |
},
|
| 1577 |
{
|
| 1578 |
"epoch": 0.15,
|
| 1579 |
+
"learning_rate": 5.109074425172806e-06,
|
| 1580 |
+
"loss": 0.0016,
|
| 1581 |
"step": 1305
|
| 1582 |
},
|
| 1583 |
{
|
| 1584 |
"epoch": 0.15,
|
| 1585 |
+
"learning_rate": 5.072719488257915e-06,
|
| 1586 |
+
"loss": 0.0014,
|
| 1587 |
"step": 1310
|
| 1588 |
},
|
| 1589 |
{
|
| 1590 |
"epoch": 0.15,
|
| 1591 |
+
"learning_rate": 5.03636070559211e-06,
|
| 1592 |
+
"loss": 0.0014,
|
| 1593 |
"step": 1315
|
| 1594 |
},
|
| 1595 |
{
|
| 1596 |
"epoch": 0.15,
|
| 1597 |
+
"learning_rate": 5e-06,
|
| 1598 |
+
"loss": 0.001,
|
| 1599 |
"step": 1320
|
| 1600 |
},
|
| 1601 |
{
|
| 1602 |
"epoch": 0.15,
|
| 1603 |
+
"learning_rate": 4.963639294407893e-06,
|
| 1604 |
+
"loss": 0.0009,
|
| 1605 |
"step": 1325
|
| 1606 |
},
|
| 1607 |
{
|
| 1608 |
"epoch": 0.15,
|
| 1609 |
+
"learning_rate": 4.927280511742087e-06,
|
| 1610 |
+
"loss": 0.0011,
|
| 1611 |
"step": 1330
|
| 1612 |
},
|
| 1613 |
{
|
| 1614 |
"epoch": 0.15,
|
| 1615 |
+
"learning_rate": 4.890925574827195e-06,
|
| 1616 |
+
"loss": 0.001,
|
| 1617 |
"step": 1335
|
| 1618 |
},
|
| 1619 |
{
|
| 1620 |
"epoch": 0.15,
|
| 1621 |
+
"learning_rate": 4.854576406284443e-06,
|
| 1622 |
+
"loss": 0.0007,
|
| 1623 |
"step": 1340
|
| 1624 |
},
|
| 1625 |
{
|
| 1626 |
"epoch": 0.15,
|
| 1627 |
+
"learning_rate": 4.818234928430003e-06,
|
| 1628 |
+
"loss": 0.0009,
|
| 1629 |
"step": 1345
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 0.15,
|
| 1633 |
+
"learning_rate": 4.781903063173321e-06,
|
| 1634 |
+
"loss": 0.0014,
|
| 1635 |
"step": 1350
|
| 1636 |
},
|
| 1637 |
{
|
| 1638 |
"epoch": 0.16,
|
| 1639 |
+
"learning_rate": 4.745582731915488e-06,
|
| 1640 |
+
"loss": 0.0008,
|
| 1641 |
"step": 1355
|
| 1642 |
},
|
| 1643 |
{
|
| 1644 |
"epoch": 0.16,
|
| 1645 |
+
"learning_rate": 4.7092758554476215e-06,
|
| 1646 |
+
"loss": 0.0007,
|
| 1647 |
"step": 1360
|
| 1648 |
},
|
| 1649 |
{
|
| 1650 |
"epoch": 0.16,
|
| 1651 |
+
"learning_rate": 4.672984353849285e-06,
|
| 1652 |
+
"loss": 0.0019,
|
| 1653 |
"step": 1365
|
| 1654 |
},
|
| 1655 |
{
|
| 1656 |
"epoch": 0.16,
|
| 1657 |
+
"learning_rate": 4.636710146386948e-06,
|
| 1658 |
+
"loss": 0.0012,
|
| 1659 |
"step": 1370
|
| 1660 |
},
|
| 1661 |
{
|
| 1662 |
"epoch": 0.16,
|
| 1663 |
+
"learning_rate": 4.600455151412482e-06,
|
| 1664 |
+
"loss": 0.0016,
|
| 1665 |
"step": 1375
|
| 1666 |
},
|
| 1667 |
{
|
| 1668 |
"epoch": 0.16,
|
| 1669 |
+
"learning_rate": 4.564221286261709e-06,
|
| 1670 |
+
"loss": 0.0014,
|
| 1671 |
"step": 1380
|
| 1672 |
},
|
| 1673 |
{
|
| 1674 |
"epoch": 0.16,
|
| 1675 |
+
"learning_rate": 4.528010467153015e-06,
|
| 1676 |
+
"loss": 0.0006,
|
| 1677 |
"step": 1385
|
| 1678 |
},
|
| 1679 |
{
|
| 1680 |
"epoch": 0.16,
|
| 1681 |
+
"learning_rate": 4.4918246090859905e-06,
|
| 1682 |
+
"loss": 0.0011,
|
| 1683 |
"step": 1390
|
| 1684 |
},
|
| 1685 |
{
|
| 1686 |
"epoch": 0.16,
|
| 1687 |
+
"learning_rate": 4.4556656257401786e-06,
|
| 1688 |
+
"loss": 0.0013,
|
| 1689 |
"step": 1395
|
| 1690 |
},
|
| 1691 |
{
|
| 1692 |
"epoch": 0.16,
|
| 1693 |
+
"learning_rate": 4.4195354293738484e-06,
|
| 1694 |
+
"loss": 0.0009,
|
| 1695 |
"step": 1400
|
| 1696 |
},
|
| 1697 |
{
|
| 1698 |
"epoch": 0.16,
|
| 1699 |
+
"learning_rate": 4.38343593072289e-06,
|
| 1700 |
+
"loss": 0.0009,
|
| 1701 |
"step": 1405
|
| 1702 |
},
|
| 1703 |
{
|
| 1704 |
"epoch": 0.16,
|
| 1705 |
+
"learning_rate": 4.347369038899744e-06,
|
| 1706 |
+
"loss": 0.0012,
|
| 1707 |
"step": 1410
|
| 1708 |
},
|
| 1709 |
{
|
| 1710 |
"epoch": 0.16,
|
| 1711 |
+
"learning_rate": 4.311336661292447e-06,
|
| 1712 |
+
"loss": 0.0007,
|
| 1713 |
"step": 1415
|
| 1714 |
},
|
| 1715 |
{
|
| 1716 |
"epoch": 0.16,
|
| 1717 |
+
"learning_rate": 4.275340703463767e-06,
|
| 1718 |
+
"loss": 0.0007,
|
| 1719 |
"step": 1420
|
| 1720 |
},
|
| 1721 |
{
|
| 1722 |
"epoch": 0.16,
|
| 1723 |
+
"learning_rate": 4.239383069050417e-06,
|
| 1724 |
+
"loss": 0.001,
|
| 1725 |
"step": 1425
|
| 1726 |
},
|
| 1727 |
{
|
| 1728 |
"epoch": 0.16,
|
| 1729 |
+
"learning_rate": 4.203465659662388e-06,
|
| 1730 |
+
"loss": 0.001,
|
| 1731 |
"step": 1430
|
| 1732 |
},
|
| 1733 |
{
|
| 1734 |
"epoch": 0.16,
|
| 1735 |
+
"learning_rate": 4.1675903747823795e-06,
|
| 1736 |
+
"loss": 0.0009,
|
| 1737 |
"step": 1435
|
| 1738 |
},
|
| 1739 |
{
|
| 1740 |
"epoch": 0.17,
|
| 1741 |
+
"learning_rate": 4.131759111665349e-06,
|
| 1742 |
+
"loss": 0.0012,
|
| 1743 |
"step": 1440
|
| 1744 |
},
|
| 1745 |
{
|
| 1746 |
"epoch": 0.17,
|
| 1747 |
+
"learning_rate": 4.0959737652381745e-06,
|
| 1748 |
+
"loss": 0.001,
|
| 1749 |
"step": 1445
|
| 1750 |
},
|
| 1751 |
{
|
| 1752 |
"epoch": 0.17,
|
| 1753 |
+
"learning_rate": 4.060236227999441e-06,
|
| 1754 |
+
"loss": 0.0012,
|
| 1755 |
"step": 1450
|
| 1756 |
},
|
| 1757 |
{
|
| 1758 |
"epoch": 0.17,
|
| 1759 |
+
"learning_rate": 4.02454838991936e-06,
|
| 1760 |
+
"loss": 0.0009,
|
| 1761 |
"step": 1455
|
| 1762 |
},
|
| 1763 |
{
|
| 1764 |
"epoch": 0.17,
|
| 1765 |
+
"learning_rate": 3.988912138339812e-06,
|
| 1766 |
+
"loss": 0.0016,
|
| 1767 |
"step": 1460
|
| 1768 |
},
|
| 1769 |
{
|
| 1770 |
"epoch": 0.17,
|
| 1771 |
+
"learning_rate": 3.95332935787454e-06,
|
| 1772 |
+
"loss": 0.0013,
|
| 1773 |
"step": 1465
|
| 1774 |
},
|
| 1775 |
{
|
| 1776 |
"epoch": 0.17,
|
| 1777 |
+
"learning_rate": 3.917801930309486e-06,
|
| 1778 |
+
"loss": 0.0012,
|
| 1779 |
"step": 1470
|
| 1780 |
},
|
| 1781 |
{
|
| 1782 |
"epoch": 0.17,
|
| 1783 |
+
"learning_rate": 3.882331734503263e-06,
|
| 1784 |
+
"loss": 0.0011,
|
| 1785 |
"step": 1475
|
| 1786 |
},
|
| 1787 |
{
|
| 1788 |
"epoch": 0.17,
|
| 1789 |
+
"learning_rate": 3.8469206462878e-06,
|
| 1790 |
+
"loss": 0.0011,
|
| 1791 |
"step": 1480
|
| 1792 |
},
|
| 1793 |
{
|
| 1794 |
"epoch": 0.17,
|
| 1795 |
+
"learning_rate": 3.8115705383691354e-06,
|
| 1796 |
+
"loss": 0.0012,
|
| 1797 |
"step": 1485
|
| 1798 |
},
|
| 1799 |
{
|
| 1800 |
"epoch": 0.17,
|
| 1801 |
+
"learning_rate": 3.776283280228381e-06,
|
| 1802 |
+
"loss": 0.001,
|
| 1803 |
"step": 1490
|
| 1804 |
},
|
| 1805 |
{
|
| 1806 |
"epoch": 0.17,
|
| 1807 |
+
"learning_rate": 3.741060738022858e-06,
|
| 1808 |
+
"loss": 0.0006,
|
| 1809 |
"step": 1495
|
| 1810 |
},
|
| 1811 |
{
|
| 1812 |
"epoch": 0.17,
|
| 1813 |
+
"learning_rate": 3.705904774487396e-06,
|
| 1814 |
+
"loss": 0.0009,
|
| 1815 |
"step": 1500
|
| 1816 |
},
|
| 1817 |
{
|
| 1818 |
"epoch": 0.17,
|
| 1819 |
+
"learning_rate": 3.6708172488358364e-06,
|
| 1820 |
+
"loss": 0.0011,
|
| 1821 |
"step": 1505
|
| 1822 |
},
|
| 1823 |
{
|
| 1824 |
"epoch": 0.17,
|
| 1825 |
+
"learning_rate": 3.6358000166626966e-06,
|
| 1826 |
+
"loss": 0.0012,
|
| 1827 |
"step": 1510
|
| 1828 |
},
|
| 1829 |
{
|
| 1830 |
"epoch": 0.17,
|
| 1831 |
+
"learning_rate": 3.6008549298450403e-06,
|
| 1832 |
+
"loss": 0.0011,
|
| 1833 |
"step": 1515
|
| 1834 |
},
|
| 1835 |
{
|
| 1836 |
"epoch": 0.17,
|
| 1837 |
+
"learning_rate": 3.5659838364445505e-06,
|
| 1838 |
+
"loss": 0.0011,
|
| 1839 |
"step": 1520
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
"epoch": 0.17,
|
| 1843 |
+
"learning_rate": 3.531188580609778e-06,
|
| 1844 |
+
"loss": 0.0015,
|
| 1845 |
"step": 1525
|
| 1846 |
},
|
| 1847 |
{
|
| 1848 |
"epoch": 0.18,
|
| 1849 |
+
"learning_rate": 3.4964710024786354e-06,
|
| 1850 |
+
"loss": 0.0009,
|
| 1851 |
"step": 1530
|
| 1852 |
},
|
| 1853 |
{
|
| 1854 |
"epoch": 0.18,
|
| 1855 |
+
"learning_rate": 3.461832938081059e-06,
|
| 1856 |
+
"loss": 0.0013,
|
| 1857 |
"step": 1535
|
| 1858 |
},
|
| 1859 |
{
|
| 1860 |
"epoch": 0.18,
|
| 1861 |
+
"learning_rate": 3.427276219241933e-06,
|
| 1862 |
+
"loss": 0.0008,
|
| 1863 |
"step": 1540
|
| 1864 |
},
|
| 1865 |
{
|
| 1866 |
"epoch": 0.18,
|
| 1867 |
+
"learning_rate": 3.3928026734841935e-06,
|
| 1868 |
+
"loss": 0.0006,
|
| 1869 |
"step": 1545
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 0.18,
|
| 1873 |
+
"learning_rate": 3.3584141239321953e-06,
|
| 1874 |
+
"loss": 0.0008,
|
| 1875 |
"step": 1550
|
| 1876 |
},
|
| 1877 |
{
|
| 1878 |
"epoch": 0.18,
|
| 1879 |
+
"learning_rate": 3.3241123892152925e-06,
|
| 1880 |
+
"loss": 0.0009,
|
| 1881 |
"step": 1555
|
| 1882 |
},
|
| 1883 |
{
|
| 1884 |
"epoch": 0.18,
|
| 1885 |
+
"learning_rate": 3.289899283371657e-06,
|
| 1886 |
+
"loss": 0.0011,
|
| 1887 |
"step": 1560
|
| 1888 |
},
|
| 1889 |
{
|
| 1890 |
"epoch": 0.18,
|
| 1891 |
+
"learning_rate": 3.2557766157523467e-06,
|
| 1892 |
+
"loss": 0.0007,
|
| 1893 |
"step": 1565
|
| 1894 |
},
|
| 1895 |
{
|
| 1896 |
"epoch": 0.18,
|
| 1897 |
+
"learning_rate": 3.2217461909256186e-06,
|
| 1898 |
+
"loss": 0.0007,
|
| 1899 |
"step": 1570
|
| 1900 |
},
|
| 1901 |
{
|
| 1902 |
"epoch": 0.18,
|
| 1903 |
+
"learning_rate": 3.1878098085814926e-06,
|
| 1904 |
+
"loss": 0.0016,
|
| 1905 |
"step": 1575
|
| 1906 |
},
|
| 1907 |
{
|
| 1908 |
"epoch": 0.18,
|
| 1909 |
+
"learning_rate": 3.1539692634365788e-06,
|
| 1910 |
+
"loss": 0.0007,
|
| 1911 |
"step": 1580
|
| 1912 |
},
|
| 1913 |
{
|
| 1914 |
"epoch": 0.18,
|
| 1915 |
+
"learning_rate": 3.1202263451391603e-06,
|
| 1916 |
+
"loss": 0.0011,
|
| 1917 |
"step": 1585
|
| 1918 |
},
|
| 1919 |
{
|
| 1920 |
"epoch": 0.18,
|
| 1921 |
+
"learning_rate": 3.0865828381745515e-06,
|
| 1922 |
+
"loss": 0.0013,
|
| 1923 |
"step": 1590
|
| 1924 |
},
|
| 1925 |
{
|
| 1926 |
"epoch": 0.18,
|
| 1927 |
+
"learning_rate": 3.053040521770726e-06,
|
| 1928 |
+
"loss": 0.0009,
|
| 1929 |
"step": 1595
|
| 1930 |
},
|
| 1931 |
{
|
| 1932 |
"epoch": 0.18,
|
| 1933 |
+
"learning_rate": 3.019601169804216e-06,
|
| 1934 |
+
"loss": 0.0007,
|
| 1935 |
"step": 1600
|
| 1936 |
},
|
| 1937 |
{
|
| 1938 |
"epoch": 0.18,
|
| 1939 |
+
"learning_rate": 2.986266550706315e-06,
|
| 1940 |
+
"loss": 0.001,
|
| 1941 |
"step": 1605
|
| 1942 |
},
|
| 1943 |
{
|
| 1944 |
"epoch": 0.18,
|
| 1945 |
+
"learning_rate": 2.9530384273695373e-06,
|
| 1946 |
+
"loss": 0.0008,
|
| 1947 |
"step": 1610
|
| 1948 |
},
|
| 1949 |
{
|
| 1950 |
"epoch": 0.19,
|
| 1951 |
+
"learning_rate": 2.9199185570544054e-06,
|
| 1952 |
+
"loss": 0.0009,
|
| 1953 |
"step": 1615
|
| 1954 |
},
|
| 1955 |
{
|
| 1956 |
"epoch": 0.19,
|
| 1957 |
+
"learning_rate": 2.886908691296504e-06,
|
| 1958 |
+
"loss": 0.0007,
|
| 1959 |
"step": 1620
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
"epoch": 0.19,
|
| 1963 |
+
"learning_rate": 2.854010575813856e-06,
|
| 1964 |
+
"loss": 0.001,
|
| 1965 |
"step": 1625
|
| 1966 |
},
|
| 1967 |
{
|
| 1968 |
"epoch": 0.19,
|
| 1969 |
+
"learning_rate": 2.8212259504146045e-06,
|
| 1970 |
+
"loss": 0.0011,
|
| 1971 |
"step": 1630
|
| 1972 |
},
|
| 1973 |
{
|
| 1974 |
"epoch": 0.19,
|
| 1975 |
+
"learning_rate": 2.7885565489049948e-06,
|
| 1976 |
+
"loss": 0.0009,
|
| 1977 |
"step": 1635
|
| 1978 |
},
|
| 1979 |
{
|
| 1980 |
"epoch": 0.19,
|
| 1981 |
+
"learning_rate": 2.7560040989976894e-06,
|
| 1982 |
+
"loss": 0.0009,
|
| 1983 |
"step": 1640
|
| 1984 |
},
|
| 1985 |
{
|
| 1986 |
"epoch": 0.19,
|
| 1987 |
+
"learning_rate": 2.723570322220399e-06,
|
| 1988 |
+
"loss": 0.0009,
|
| 1989 |
"step": 1645
|
| 1990 |
},
|
| 1991 |
{
|
| 1992 |
"epoch": 0.19,
|
| 1993 |
+
"learning_rate": 2.6912569338248317e-06,
|
| 1994 |
+
"loss": 0.0012,
|
| 1995 |
"step": 1650
|
| 1996 |
},
|
| 1997 |
{
|
| 1998 |
"epoch": 0.19,
|
| 1999 |
+
"learning_rate": 2.6590656426959906e-06,
|
| 2000 |
+
"loss": 0.001,
|
| 2001 |
"step": 1655
|
| 2002 |
},
|
| 2003 |
{
|
| 2004 |
"epoch": 0.19,
|
| 2005 |
+
"learning_rate": 2.626998151261798e-06,
|
| 2006 |
+
"loss": 0.0011,
|
| 2007 |
"step": 1660
|
| 2008 |
},
|
| 2009 |
{
|
| 2010 |
"epoch": 0.19,
|
| 2011 |
+
"learning_rate": 2.595056155403063e-06,
|
| 2012 |
+
"loss": 0.0007,
|
| 2013 |
"step": 1665
|
| 2014 |
},
|
| 2015 |
{
|
| 2016 |
"epoch": 0.19,
|
| 2017 |
+
"learning_rate": 2.5632413443637887e-06,
|
| 2018 |
+
"loss": 0.001,
|
| 2019 |
"step": 1670
|
| 2020 |
},
|
| 2021 |
{
|
| 2022 |
"epoch": 0.19,
|
| 2023 |
+
"learning_rate": 2.5315554006618487e-06,
|
| 2024 |
+
"loss": 0.0012,
|
| 2025 |
"step": 1675
|
| 2026 |
},
|
| 2027 |
{
|
| 2028 |
"epoch": 0.19,
|
| 2029 |
+
"learning_rate": 2.5000000000000015e-06,
|
| 2030 |
+
"loss": 0.0009,
|
| 2031 |
"step": 1680
|
| 2032 |
},
|
| 2033 |
{
|
| 2034 |
"epoch": 0.19,
|
| 2035 |
+
"learning_rate": 2.4685768111772647e-06,
|
| 2036 |
+
"loss": 0.001,
|
| 2037 |
"step": 1685
|
| 2038 |
},
|
| 2039 |
{
|
| 2040 |
"epoch": 0.19,
|
| 2041 |
+
"learning_rate": 2.437287496000674e-06,
|
| 2042 |
+
"loss": 0.0011,
|
| 2043 |
"step": 1690
|
| 2044 |
},
|
| 2045 |
{
|
| 2046 |
"epoch": 0.19,
|
| 2047 |
+
"learning_rate": 2.406133709197392e-06,
|
| 2048 |
+
"loss": 0.0007,
|
| 2049 |
"step": 1695
|
| 2050 |
},
|
| 2051 |
{
|
| 2052 |
"epoch": 0.2,
|
| 2053 |
+
"learning_rate": 2.3751170983272e-06,
|
| 2054 |
+
"loss": 0.001,
|
| 2055 |
"step": 1700
|
| 2056 |
},
|
| 2057 |
{
|
| 2058 |
"epoch": 0.2,
|
| 2059 |
+
"learning_rate": 2.3442393036953614e-06,
|
| 2060 |
+
"loss": 0.0006,
|
| 2061 |
"step": 1705
|
| 2062 |
},
|
| 2063 |
{
|
| 2064 |
"epoch": 0.2,
|
| 2065 |
+
"learning_rate": 2.3135019582658803e-06,
|
| 2066 |
+
"loss": 0.0007,
|
| 2067 |
"step": 1710
|
| 2068 |
},
|
| 2069 |
{
|
| 2070 |
"epoch": 0.2,
|
| 2071 |
+
"learning_rate": 2.282906687575151e-06,
|
| 2072 |
+
"loss": 0.0016,
|
| 2073 |
"step": 1715
|
| 2074 |
},
|
| 2075 |
{
|
| 2076 |
"epoch": 0.2,
|
| 2077 |
+
"learning_rate": 2.2524551096459703e-06,
|
| 2078 |
+
"loss": 0.0013,
|
| 2079 |
"step": 1720
|
| 2080 |
},
|
| 2081 |
{
|
| 2082 |
"epoch": 0.2,
|
| 2083 |
+
"learning_rate": 2.2221488349019903e-06,
|
| 2084 |
+
"loss": 0.0013,
|
| 2085 |
"step": 1725
|
| 2086 |
},
|
| 2087 |
{
|
| 2088 |
"epoch": 0.2,
|
| 2089 |
+
"learning_rate": 2.1919894660825362e-06,
|
| 2090 |
+
"loss": 0.0009,
|
| 2091 |
"step": 1730
|
| 2092 |
},
|
| 2093 |
{
|
| 2094 |
"epoch": 0.2,
|
| 2095 |
+
"learning_rate": 2.161978598157857e-06,
|
| 2096 |
+
"loss": 0.0007,
|
| 2097 |
"step": 1735
|
| 2098 |
},
|
| 2099 |
{
|
| 2100 |
"epoch": 0.2,
|
| 2101 |
+
"learning_rate": 2.132117818244771e-06,
|
| 2102 |
+
"loss": 0.0013,
|
| 2103 |
"step": 1740
|
| 2104 |
},
|
| 2105 |
{
|
| 2106 |
"epoch": 0.2,
|
| 2107 |
+
"learning_rate": 2.102408705522725e-06,
|
| 2108 |
+
"loss": 0.0009,
|
| 2109 |
"step": 1745
|
| 2110 |
},
|
| 2111 |
{
|
| 2112 |
"epoch": 0.2,
|
| 2113 |
+
"learning_rate": 2.0728528311502977e-06,
|
| 2114 |
+
"loss": 0.0009,
|
| 2115 |
"step": 1750
|
| 2116 |
},
|
| 2117 |
{
|
| 2118 |
"epoch": 0.2,
|
| 2119 |
+
"learning_rate": 2.0434517581820893e-06,
|
| 2120 |
+
"loss": 0.0008,
|
| 2121 |
"step": 1755
|
| 2122 |
},
|
| 2123 |
{
|
| 2124 |
"epoch": 0.2,
|
| 2125 |
+
"learning_rate": 2.0142070414860704e-06,
|
| 2126 |
+
"loss": 0.001,
|
| 2127 |
"step": 1760
|
| 2128 |
},
|
| 2129 |
{
|
| 2130 |
"epoch": 0.2,
|
| 2131 |
+
"learning_rate": 1.9851202276613524e-06,
|
| 2132 |
+
"loss": 0.0008,
|
| 2133 |
"step": 1765
|
| 2134 |
},
|
| 2135 |
{
|
| 2136 |
"epoch": 0.2,
|
| 2137 |
+
"learning_rate": 1.956192854956397e-06,
|
| 2138 |
+
"loss": 0.0008,
|
| 2139 |
"step": 1770
|
| 2140 |
},
|
| 2141 |
{
|
| 2142 |
"epoch": 0.2,
|
| 2143 |
+
"learning_rate": 1.927426453187663e-06,
|
| 2144 |
+
"loss": 0.0009,
|
| 2145 |
"step": 1775
|
| 2146 |
},
|
| 2147 |
{
|
| 2148 |
"epoch": 0.2,
|
| 2149 |
+
"learning_rate": 1.8988225436587005e-06,
|
| 2150 |
+
"loss": 0.0008,
|
| 2151 |
"step": 1780
|
| 2152 |
},
|
| 2153 |
{
|
| 2154 |
"epoch": 0.2,
|
| 2155 |
+
"learning_rate": 1.8703826390797047e-06,
|
| 2156 |
+
"loss": 0.0006,
|
| 2157 |
"step": 1785
|
| 2158 |
},
|
| 2159 |
{
|
| 2160 |
"epoch": 0.21,
|
| 2161 |
+
"learning_rate": 1.8421082434875133e-06,
|
| 2162 |
+
"loss": 0.0014,
|
| 2163 |
"step": 1790
|
| 2164 |
},
|
| 2165 |
{
|
| 2166 |
"epoch": 0.21,
|
| 2167 |
+
"learning_rate": 1.814000852166059e-06,
|
| 2168 |
+
"loss": 0.001,
|
| 2169 |
"step": 1795
|
| 2170 |
},
|
| 2171 |
{
|
| 2172 |
"epoch": 0.21,
|
| 2173 |
+
"learning_rate": 1.7860619515673034e-06,
|
| 2174 |
+
"loss": 0.0005,
|
| 2175 |
"step": 1800
|
| 2176 |
},
|
| 2177 |
{
|
| 2178 |
"epoch": 0.21,
|
| 2179 |
+
"learning_rate": 1.7582930192326187e-06,
|
| 2180 |
+
"loss": 0.0011,
|
| 2181 |
"step": 1805
|
| 2182 |
},
|
| 2183 |
{
|
| 2184 |
"epoch": 0.21,
|
| 2185 |
+
"learning_rate": 1.7306955237146523e-06,
|
| 2186 |
+
"loss": 0.0015,
|
| 2187 |
"step": 1810
|
| 2188 |
},
|
| 2189 |
{
|
| 2190 |
"epoch": 0.21,
|
| 2191 |
+
"learning_rate": 1.7032709244996559e-06,
|
| 2192 |
+
"loss": 0.0007,
|
| 2193 |
"step": 1815
|
| 2194 |
},
|
| 2195 |
{
|
| 2196 |
"epoch": 0.21,
|
| 2197 |
+
"learning_rate": 1.6760206719303107e-06,
|
| 2198 |
+
"loss": 0.0011,
|
| 2199 |
"step": 1820
|
| 2200 |
},
|
| 2201 |
{
|
| 2202 |
"epoch": 0.21,
|
| 2203 |
+
"learning_rate": 1.6489462071290213e-06,
|
| 2204 |
+
"loss": 0.0004,
|
| 2205 |
"step": 1825
|
| 2206 |
},
|
| 2207 |
{
|
| 2208 |
"epoch": 0.21,
|
| 2209 |
+
"learning_rate": 1.6220489619216988e-06,
|
| 2210 |
+
"loss": 0.001,
|
| 2211 |
"step": 1830
|
| 2212 |
},
|
| 2213 |
{
|
| 2214 |
"epoch": 0.21,
|
| 2215 |
+
"learning_rate": 1.5953303587620472e-06,
|
| 2216 |
"loss": 0.0011,
|
| 2217 |
"step": 1835
|
| 2218 |
},
|
| 2219 |
{
|
| 2220 |
"epoch": 0.21,
|
| 2221 |
+
"learning_rate": 1.5687918106563326e-06,
|
| 2222 |
+
"loss": 0.0009,
|
| 2223 |
"step": 1840
|
| 2224 |
},
|
| 2225 |
{
|
| 2226 |
"epoch": 0.21,
|
| 2227 |
+
"learning_rate": 1.5424347210886538e-06,
|
| 2228 |
+
"loss": 0.0005,
|
| 2229 |
"step": 1845
|
| 2230 |
},
|
| 2231 |
{
|
| 2232 |
"epoch": 0.21,
|
| 2233 |
+
"learning_rate": 1.5162604839467265e-06,
|
| 2234 |
+
"loss": 0.0013,
|
| 2235 |
"step": 1850
|
| 2236 |
},
|
| 2237 |
{
|
| 2238 |
"epoch": 0.21,
|
| 2239 |
+
"learning_rate": 1.4902704834481585e-06,
|
| 2240 |
+
"loss": 0.0011,
|
| 2241 |
"step": 1855
|
| 2242 |
},
|
| 2243 |
{
|
| 2244 |
"epoch": 0.21,
|
| 2245 |
+
"learning_rate": 1.4644660940672628e-06,
|
| 2246 |
+
"loss": 0.0006,
|
| 2247 |
"step": 1860
|
| 2248 |
},
|
| 2249 |
{
|
| 2250 |
"epoch": 0.21,
|
| 2251 |
+
"learning_rate": 1.4388486804623464e-06,
|
| 2252 |
+
"loss": 0.0007,
|
| 2253 |
"step": 1865
|
| 2254 |
},
|
| 2255 |
{
|
| 2256 |
"epoch": 0.21,
|
| 2257 |
+
"learning_rate": 1.4134195974035525e-06,
|
| 2258 |
+
"loss": 0.0006,
|
| 2259 |
"step": 1870
|
| 2260 |
},
|
| 2261 |
{
|
| 2262 |
"epoch": 0.22,
|
| 2263 |
+
"learning_rate": 1.3881801897012225e-06,
|
| 2264 |
+
"loss": 0.0014,
|
| 2265 |
"step": 1875
|
| 2266 |
},
|
| 2267 |
{
|
| 2268 |
"epoch": 0.22,
|
| 2269 |
+
"learning_rate": 1.3631317921347564e-06,
|
| 2270 |
+
"loss": 0.0012,
|
| 2271 |
"step": 1880
|
| 2272 |
},
|
| 2273 |
{
|
| 2274 |
"epoch": 0.22,
|
| 2275 |
+
"learning_rate": 1.3382757293820408e-06,
|
| 2276 |
+
"loss": 0.0012,
|
| 2277 |
"step": 1885
|
| 2278 |
},
|
| 2279 |
{
|
| 2280 |
"epoch": 0.22,
|
| 2281 |
+
"learning_rate": 1.3136133159493803e-06,
|
| 2282 |
+
"loss": 0.0008,
|
| 2283 |
"step": 1890
|
| 2284 |
},
|
| 2285 |
{
|
| 2286 |
"epoch": 0.22,
|
| 2287 |
+
"learning_rate": 1.2891458561019914e-06,
|
| 2288 |
+
"loss": 0.0006,
|
| 2289 |
"step": 1895
|
| 2290 |
},
|
| 2291 |
{
|
| 2292 |
"epoch": 0.22,
|
| 2293 |
+
"learning_rate": 1.264874643795021e-06,
|
| 2294 |
+
"loss": 0.0009,
|
| 2295 |
"step": 1900
|
| 2296 |
},
|
| 2297 |
{
|
| 2298 |
"epoch": 0.22,
|
| 2299 |
+
"learning_rate": 1.2408009626051137e-06,
|
| 2300 |
+
"loss": 0.0013,
|
| 2301 |
"step": 1905
|
| 2302 |
},
|
| 2303 |
{
|
| 2304 |
"epoch": 0.22,
|
| 2305 |
+
"learning_rate": 1.2169260856625358e-06,
|
| 2306 |
+
"loss": 0.0011,
|
| 2307 |
"step": 1910
|
| 2308 |
},
|
| 2309 |
{
|
| 2310 |
"epoch": 0.22,
|
| 2311 |
+
"learning_rate": 1.1932512755838448e-06,
|
| 2312 |
+
"loss": 0.001,
|
| 2313 |
"step": 1915
|
| 2314 |
},
|
| 2315 |
{
|
| 2316 |
"epoch": 0.22,
|
| 2317 |
+
"learning_rate": 1.1697777844051105e-06,
|
| 2318 |
+
"loss": 0.0009,
|
| 2319 |
"step": 1920
|
| 2320 |
},
|
| 2321 |
{
|
| 2322 |
"epoch": 0.22,
|
| 2323 |
+
"learning_rate": 1.1465068535157098e-06,
|
| 2324 |
+
"loss": 0.0012,
|
| 2325 |
"step": 1925
|
| 2326 |
},
|
| 2327 |
{
|
| 2328 |
"epoch": 0.22,
|
| 2329 |
+
"learning_rate": 1.1234397135926705e-06,
|
| 2330 |
+
"loss": 0.0011,
|
| 2331 |
"step": 1930
|
| 2332 |
},
|
| 2333 |
{
|
| 2334 |
"epoch": 0.22,
|
| 2335 |
+
"learning_rate": 1.100577584535592e-06,
|
| 2336 |
+
"loss": 0.001,
|
| 2337 |
"step": 1935
|
| 2338 |
},
|
| 2339 |
{
|
| 2340 |
"epoch": 0.22,
|
| 2341 |
+
"learning_rate": 1.0779216754021215e-06,
|
| 2342 |
+
"loss": 0.001,
|
| 2343 |
"step": 1940
|
| 2344 |
},
|
| 2345 |
{
|
| 2346 |
"epoch": 0.22,
|
| 2347 |
+
"learning_rate": 1.0554731843440275e-06,
|
| 2348 |
+
"loss": 0.0013,
|
| 2349 |
"step": 1945
|
| 2350 |
},
|
| 2351 |
{
|
| 2352 |
"epoch": 0.22,
|
| 2353 |
+
"learning_rate": 1.0332332985438248e-06,
|
| 2354 |
+
"loss": 0.0007,
|
| 2355 |
"step": 1950
|
| 2356 |
},
|
| 2357 |
{
|
| 2358 |
"epoch": 0.22,
|
| 2359 |
+
"learning_rate": 1.0112031941519934e-06,
|
| 2360 |
+
"loss": 0.0013,
|
| 2361 |
"step": 1955
|
| 2362 |
},
|
| 2363 |
{
|
| 2364 |
"epoch": 0.22,
|
| 2365 |
+
"learning_rate": 9.893840362247809e-07,
|
| 2366 |
+
"loss": 0.0008,
|
| 2367 |
"step": 1960
|
| 2368 |
},
|
| 2369 |
{
|
| 2370 |
"epoch": 0.23,
|
| 2371 |
+
"learning_rate": 9.677769786625869e-07,
|
| 2372 |
+
"loss": 0.0011,
|
| 2373 |
"step": 1965
|
| 2374 |
},
|
| 2375 |
{
|
| 2376 |
"epoch": 0.23,
|
| 2377 |
+
"learning_rate": 9.463831641489391e-07,
|
| 2378 |
+
"loss": 0.0009,
|
| 2379 |
"step": 1970
|
| 2380 |
},
|
| 2381 |
{
|
| 2382 |
"epoch": 0.23,
|
| 2383 |
+
"learning_rate": 9.252037240900618e-07,
|
| 2384 |
+
"loss": 0.0012,
|
| 2385 |
"step": 1975
|
| 2386 |
},
|
| 2387 |
{
|
| 2388 |
"epoch": 0.23,
|
| 2389 |
+
"learning_rate": 9.042397785550405e-07,
|
| 2390 |
+
"loss": 0.0006,
|
| 2391 |
"step": 1980
|
| 2392 |
},
|
| 2393 |
{
|
| 2394 |
"epoch": 0.23,
|
| 2395 |
+
"learning_rate": 8.834924362165992e-07,
|
| 2396 |
+
"loss": 0.0008,
|
| 2397 |
"step": 1985
|
| 2398 |
},
|
| 2399 |
{
|
| 2400 |
"epoch": 0.23,
|
| 2401 |
+
"learning_rate": 8.629627942924473e-07,
|
| 2402 |
+
"loss": 0.0012,
|
| 2403 |
"step": 1990
|
| 2404 |
},
|
| 2405 |
{
|
| 2406 |
"epoch": 0.23,
|
| 2407 |
+
"learning_rate": 8.426519384872733e-07,
|
| 2408 |
+
"loss": 0.0008,
|
| 2409 |
"step": 1995
|
| 2410 |
},
|
| 2411 |
{
|
| 2412 |
"epoch": 0.23,
|
| 2413 |
+
"learning_rate": 8.225609429353187e-07,
|
| 2414 |
+
"loss": 0.0006,
|
| 2415 |
"step": 2000
|
| 2416 |
},
|
| 2417 |
{
|
| 2418 |
"epoch": 0.23,
|
| 2419 |
+
"learning_rate": 8.026908701435681e-07,
|
| 2420 |
+
"loss": 0.0005,
|
| 2421 |
"step": 2005
|
| 2422 |
},
|
| 2423 |
{
|
| 2424 |
"epoch": 0.23,
|
| 2425 |
+
"learning_rate": 7.830427709355726e-07,
|
| 2426 |
+
"loss": 0.0014,
|
| 2427 |
"step": 2010
|
| 2428 |
},
|
| 2429 |
{
|
| 2430 |
"epoch": 0.23,
|
| 2431 |
+
"learning_rate": 7.636176843958599e-07,
|
| 2432 |
+
"loss": 0.0009,
|
| 2433 |
"step": 2015
|
| 2434 |
},
|
| 2435 |
{
|
| 2436 |
"epoch": 0.23,
|
| 2437 |
+
"learning_rate": 7.444166378150014e-07,
|
| 2438 |
+
"loss": 0.0007,
|
| 2439 |
"step": 2020
|
| 2440 |
},
|
| 2441 |
{
|
| 2442 |
"epoch": 0.23,
|
| 2443 |
+
"learning_rate": 7.254406466352682e-07,
|
| 2444 |
+
"loss": 0.0008,
|
| 2445 |
"step": 2025
|
| 2446 |
},
|
| 2447 |
{
|
| 2448 |
"epoch": 0.23,
|
| 2449 |
+
"learning_rate": 7.066907143969353e-07,
|
| 2450 |
+
"loss": 0.0011,
|
| 2451 |
"step": 2030
|
| 2452 |
},
|
| 2453 |
{
|
| 2454 |
"epoch": 0.23,
|
| 2455 |
+
"learning_rate": 6.881678326852137e-07,
|
| 2456 |
+
"loss": 0.0008,
|
| 2457 |
"step": 2035
|
| 2458 |
},
|
| 2459 |
{
|
| 2460 |
"epoch": 0.23,
|
| 2461 |
+
"learning_rate": 6.698729810778065e-07,
|
| 2462 |
+
"loss": 0.001,
|
| 2463 |
"step": 2040
|
| 2464 |
},
|
| 2465 |
{
|
| 2466 |
"epoch": 0.23,
|
| 2467 |
+
"learning_rate": 6.518071270931059e-07,
|
| 2468 |
+
"loss": 0.0007,
|
| 2469 |
"step": 2045
|
| 2470 |
},
|
| 2471 |
{
|
| 2472 |
"epoch": 0.24,
|
| 2473 |
+
"learning_rate": 6.339712261390213e-07,
|
| 2474 |
+
"loss": 0.0011,
|
| 2475 |
"step": 2050
|
| 2476 |
},
|
| 2477 |
{
|
| 2478 |
"epoch": 0.24,
|
| 2479 |
+
"learning_rate": 6.163662214624616e-07,
|
| 2480 |
+
"loss": 0.0011,
|
| 2481 |
"step": 2055
|
| 2482 |
},
|
| 2483 |
{
|
| 2484 |
"epoch": 0.24,
|
| 2485 |
+
"learning_rate": 5.989930440994451e-07,
|
| 2486 |
+
"loss": 0.0008,
|
| 2487 |
"step": 2060
|
| 2488 |
},
|
| 2489 |
{
|
| 2490 |
"epoch": 0.24,
|
| 2491 |
+
"learning_rate": 5.818526128258622e-07,
|
| 2492 |
+
"loss": 0.0009,
|
| 2493 |
"step": 2065
|
| 2494 |
},
|
| 2495 |
{
|
| 2496 |
"epoch": 0.24,
|
| 2497 |
+
"learning_rate": 5.649458341088915e-07,
|
| 2498 |
+
"loss": 0.0011,
|
| 2499 |
"step": 2070
|
| 2500 |
},
|
| 2501 |
{
|
| 2502 |
"epoch": 0.24,
|
| 2503 |
+
"learning_rate": 5.482736020590551e-07,
|
| 2504 |
+
"loss": 0.0005,
|
| 2505 |
"step": 2075
|
| 2506 |
},
|
| 2507 |
{
|
| 2508 |
"epoch": 0.24,
|
| 2509 |
+
"learning_rate": 5.318367983829393e-07,
|
| 2510 |
+
"loss": 0.0012,
|
| 2511 |
"step": 2080
|
| 2512 |
},
|
| 2513 |
{
|
| 2514 |
"epoch": 0.24,
|
| 2515 |
+
"learning_rate": 5.188574484306829e-07,
|
| 2516 |
+
"loss": 0.0016,
|
| 2517 |
"step": 2085
|
| 2518 |
},
|
| 2519 |
{
|
| 2520 |
"epoch": 0.24,
|
| 2521 |
+
"learning_rate": 5.028465979562792e-07,
|
| 2522 |
+
"loss": 0.0005,
|
| 2523 |
"step": 2090
|
| 2524 |
},
|
| 2525 |
{
|
| 2526 |
"epoch": 0.24,
|
| 2527 |
+
"learning_rate": 4.87073578250698e-07,
|
| 2528 |
+
"loss": 0.001,
|
| 2529 |
"step": 2095
|
| 2530 |
},
|
| 2531 |
{
|
| 2532 |
"epoch": 0.24,
|
| 2533 |
+
"learning_rate": 4.7153922346591554e-07,
|
| 2534 |
+
"loss": 0.0009,
|
| 2535 |
"step": 2100
|
| 2536 |
},
|
| 2537 |
{
|
| 2538 |
"epoch": 0.24,
|
| 2539 |
+
"learning_rate": 4.562443551321788e-07,
|
| 2540 |
+
"loss": 0.001,
|
| 2541 |
"step": 2105
|
| 2542 |
},
|
| 2543 |
{
|
| 2544 |
"epoch": 0.24,
|
| 2545 |
+
"learning_rate": 4.4118978211455723e-07,
|
| 2546 |
+
"loss": 0.0008,
|
| 2547 |
"step": 2110
|
| 2548 |
},
|
| 2549 |
{
|
| 2550 |
"epoch": 0.24,
|
| 2551 |
+
"learning_rate": 4.263763005701649e-07,
|
| 2552 |
+
"loss": 0.0009,
|
| 2553 |
"step": 2115
|
| 2554 |
},
|
| 2555 |
{
|
| 2556 |
"epoch": 0.24,
|
| 2557 |
+
"learning_rate": 4.118046939060566e-07,
|
| 2558 |
+
"loss": 0.0004,
|
| 2559 |
"step": 2120
|
| 2560 |
},
|
| 2561 |
{
|
| 2562 |
"epoch": 0.24,
|
| 2563 |
+
"learning_rate": 3.9747573273779816e-07,
|
| 2564 |
+
"loss": 0.0014,
|
| 2565 |
"step": 2125
|
| 2566 |
},
|
| 2567 |
{
|
| 2568 |
"epoch": 0.24,
|
| 2569 |
+
"learning_rate": 3.833901748487151e-07,
|
| 2570 |
+
"loss": 0.0012,
|
| 2571 |
"step": 2130
|
| 2572 |
},
|
| 2573 |
{
|
| 2574 |
"epoch": 0.24,
|
| 2575 |
+
"learning_rate": 3.6954876514981084e-07,
|
| 2576 |
"loss": 0.0013,
|
| 2577 |
"step": 2135
|
| 2578 |
},
|
| 2579 |
{
|
| 2580 |
"epoch": 0.25,
|
| 2581 |
+
"learning_rate": 3.5595223564037884e-07,
|
| 2582 |
+
"loss": 0.0011,
|
| 2583 |
"step": 2140
|
| 2584 |
},
|
| 2585 |
{
|
| 2586 |
"epoch": 0.25,
|
| 2587 |
+
"learning_rate": 3.426013053692878e-07,
|
| 2588 |
+
"loss": 0.0009,
|
| 2589 |
"step": 2145
|
| 2590 |
},
|
| 2591 |
{
|
| 2592 |
"epoch": 0.25,
|
| 2593 |
+
"learning_rate": 3.294966803969574e-07,
|
| 2594 |
+
"loss": 0.0009,
|
| 2595 |
"step": 2150
|
| 2596 |
},
|
| 2597 |
{
|
| 2598 |
"epoch": 0.25,
|
| 2599 |
+
"learning_rate": 3.166390537580122e-07,
|
| 2600 |
+
"loss": 0.0008,
|
| 2601 |
"step": 2155
|
| 2602 |
},
|
| 2603 |
{
|
| 2604 |
"epoch": 0.25,
|
| 2605 |
+
"learning_rate": 3.0402910542463915e-07,
|
| 2606 |
+
"loss": 0.0009,
|
| 2607 |
"step": 2160
|
| 2608 |
},
|
| 2609 |
{
|
| 2610 |
"epoch": 0.25,
|
| 2611 |
+
"learning_rate": 2.916675022706239e-07,
|
| 2612 |
+
"loss": 0.0013,
|
| 2613 |
"step": 2165
|
| 2614 |
},
|
| 2615 |
{
|
| 2616 |
"epoch": 0.25,
|
| 2617 |
+
"learning_rate": 2.7955489803607907e-07,
|
| 2618 |
+
"loss": 0.0013,
|
| 2619 |
"step": 2170
|
| 2620 |
},
|
| 2621 |
{
|
| 2622 |
"epoch": 0.25,
|
| 2623 |
+
"learning_rate": 2.676919332928785e-07,
|
| 2624 |
"loss": 0.001,
|
| 2625 |
"step": 2175
|
| 2626 |
},
|
| 2627 |
{
|
| 2628 |
"epoch": 0.25,
|
| 2629 |
+
"learning_rate": 2.560792354107777e-07,
|
| 2630 |
+
"loss": 0.0009,
|
| 2631 |
"step": 2180
|
| 2632 |
},
|
| 2633 |
{
|
| 2634 |
"epoch": 0.25,
|
| 2635 |
+
"learning_rate": 2.447174185242324e-07,
|
| 2636 |
"loss": 0.0011,
|
| 2637 |
"step": 2185
|
| 2638 |
},
|
| 2639 |
{
|
| 2640 |
"epoch": 0.25,
|
| 2641 |
+
"learning_rate": 2.33607083499926e-07,
|
| 2642 |
+
"loss": 0.0009,
|
| 2643 |
"step": 2190
|
| 2644 |
},
|
| 2645 |
{
|
| 2646 |
"epoch": 0.25,
|
| 2647 |
+
"learning_rate": 2.2274881790498914e-07,
|
| 2648 |
+
"loss": 0.0007,
|
| 2649 |
"step": 2195
|
| 2650 |
},
|
| 2651 |
{
|
| 2652 |
"epoch": 0.25,
|
| 2653 |
+
"learning_rate": 2.1214319597592792e-07,
|
| 2654 |
+
"loss": 0.0008,
|
| 2655 |
"step": 2200
|
| 2656 |
},
|
| 2657 |
{
|
| 2658 |
"epoch": 0.25,
|
| 2659 |
+
"learning_rate": 2.0179077858825445e-07,
|
| 2660 |
+
"loss": 0.001,
|
| 2661 |
"step": 2205
|
| 2662 |
},
|
| 2663 |
{
|
| 2664 |
"epoch": 0.25,
|
| 2665 |
+
"learning_rate": 1.916921132268229e-07,
|
| 2666 |
+
"loss": 0.0015,
|
| 2667 |
"step": 2210
|
| 2668 |
},
|
| 2669 |
{
|
| 2670 |
"epoch": 0.25,
|
| 2671 |
+
"learning_rate": 1.8184773395688527e-07,
|
| 2672 |
+
"loss": 0.0009,
|
| 2673 |
"step": 2215
|
| 2674 |
},
|
| 2675 |
{
|
| 2676 |
"epoch": 0.25,
|
| 2677 |
+
"learning_rate": 1.7225816139583407e-07,
|
| 2678 |
+
"loss": 0.0009,
|
| 2679 |
"step": 2220
|
| 2680 |
},
|
| 2681 |
{
|
| 2682 |
"epoch": 0.26,
|
| 2683 |
+
"learning_rate": 1.6292390268568103e-07,
|
| 2684 |
+
"loss": 0.001,
|
| 2685 |
"step": 2225
|
| 2686 |
},
|
| 2687 |
{
|
| 2688 |
"epoch": 0.26,
|
| 2689 |
+
"learning_rate": 1.5384545146622854e-07,
|
| 2690 |
+
"loss": 0.0008,
|
| 2691 |
"step": 2230
|
| 2692 |
},
|
| 2693 |
{
|
| 2694 |
"epoch": 0.26,
|
| 2695 |
+
"learning_rate": 1.450232878489699e-07,
|
| 2696 |
+
"loss": 0.0008,
|
| 2697 |
"step": 2235
|
| 2698 |
},
|
| 2699 |
{
|
| 2700 |
"epoch": 0.26,
|
| 2701 |
+
"learning_rate": 1.3645787839169755e-07,
|
| 2702 |
+
"loss": 0.001,
|
| 2703 |
"step": 2240
|
| 2704 |
},
|
| 2705 |
{
|
| 2706 |
"epoch": 0.26,
|
| 2707 |
+
"learning_rate": 1.2814967607382433e-07,
|
| 2708 |
+
"loss": 0.0011,
|
| 2709 |
"step": 2245
|
| 2710 |
},
|
| 2711 |
{
|
| 2712 |
"epoch": 0.26,
|
| 2713 |
+
"learning_rate": 1.2009912027243386e-07,
|
| 2714 |
+
"loss": 0.0009,
|
| 2715 |
"step": 2250
|
| 2716 |
},
|
| 2717 |
{
|
| 2718 |
"epoch": 0.26,
|
| 2719 |
+
"learning_rate": 1.123066367390424e-07,
|
| 2720 |
"loss": 0.001,
|
| 2721 |
"step": 2255
|
| 2722 |
},
|
| 2723 |
{
|
| 2724 |
"epoch": 0.26,
|
| 2725 |
+
"learning_rate": 1.0477263757708078e-07,
|
| 2726 |
+
"loss": 0.0011,
|
| 2727 |
"step": 2260
|
| 2728 |
},
|
| 2729 |
{
|
| 2730 |
"epoch": 0.26,
|
| 2731 |
+
"learning_rate": 9.749752122010347e-08,
|
| 2732 |
+
"loss": 0.0006,
|
| 2733 |
"step": 2265
|
| 2734 |
},
|
| 2735 |
{
|
| 2736 |
"epoch": 0.26,
|
| 2737 |
+
"learning_rate": 9.048167241071548e-08,
|
| 2738 |
+
"loss": 0.0006,
|
| 2739 |
"step": 2270
|
| 2740 |
},
|
| 2741 |
{
|
| 2742 |
"epoch": 0.26,
|
| 2743 |
+
"learning_rate": 8.372546218022747e-08,
|
| 2744 |
+
"loss": 0.0007,
|
| 2745 |
"step": 2275
|
| 2746 |
},
|
| 2747 |
{
|
| 2748 |
"epoch": 0.26,
|
| 2749 |
+
"learning_rate": 7.722924782902985e-08,
|
| 2750 |
+
"loss": 0.0015,
|
| 2751 |
"step": 2280
|
| 2752 |
},
|
| 2753 |
{
|
| 2754 |
"epoch": 0.26,
|
| 2755 |
+
"learning_rate": 7.09933729077017e-08,
|
| 2756 |
+
"loss": 0.0008,
|
| 2757 |
"step": 2285
|
| 2758 |
},
|
| 2759 |
{
|
| 2760 |
"epoch": 0.26,
|
| 2761 |
+
"learning_rate": 6.501816719884091e-08,
|
| 2762 |
+
"loss": 0.0006,
|
| 2763 |
"step": 2290
|
| 2764 |
},
|
| 2765 |
{
|
| 2766 |
"epoch": 0.26,
|
| 2767 |
+
"learning_rate": 5.9303946699620365e-08,
|
| 2768 |
+
"loss": 0.0011,
|
| 2769 |
"step": 2295
|
| 2770 |
},
|
| 2771 |
{
|
| 2772 |
"epoch": 0.26,
|
| 2773 |
+
"learning_rate": 5.3851013605080717e-08,
|
| 2774 |
+
"loss": 0.0009,
|
| 2775 |
"step": 2300
|
| 2776 |
},
|
| 2777 |
{
|
| 2778 |
"epoch": 0.26,
|
| 2779 |
+
"learning_rate": 4.865965629214819e-08,
|
| 2780 |
+
"loss": 0.001,
|
| 2781 |
"step": 2305
|
| 2782 |
},
|
| 2783 |
{
|
| 2784 |
"epoch": 0.27,
|
| 2785 |
+
"learning_rate": 4.37301493043818e-08,
|
| 2786 |
+
"loss": 0.0012,
|
| 2787 |
"step": 2310
|
| 2788 |
},
|
| 2789 |
{
|
| 2790 |
"epoch": 0.27,
|
| 2791 |
+
"learning_rate": 3.9062753337454354e-08,
|
| 2792 |
+
"loss": 0.0016,
|
| 2793 |
"step": 2315
|
| 2794 |
},
|
| 2795 |
{
|
| 2796 |
"epoch": 0.27,
|
| 2797 |
+
"learning_rate": 3.465771522536854e-08,
|
| 2798 |
+
"loss": 0.0014,
|
| 2799 |
"step": 2320
|
| 2800 |
},
|
| 2801 |
{
|
| 2802 |
"epoch": 0.27,
|
| 2803 |
+
"learning_rate": 3.0515267927400116e-08,
|
| 2804 |
+
"loss": 0.0006,
|
| 2805 |
"step": 2325
|
| 2806 |
},
|
| 2807 |
{
|
| 2808 |
"epoch": 0.27,
|
| 2809 |
+
"learning_rate": 2.6635630515779996e-08,
|
| 2810 |
+
"loss": 0.0012,
|
| 2811 |
"step": 2330
|
| 2812 |
},
|
| 2813 |
{
|
| 2814 |
"epoch": 0.27,
|
| 2815 |
+
"learning_rate": 2.301900816410574e-08,
|
| 2816 |
+
"loss": 0.0009,
|
| 2817 |
"step": 2335
|
| 2818 |
},
|
| 2819 |
{
|
| 2820 |
"epoch": 0.27,
|
| 2821 |
+
"learning_rate": 1.966559213649577e-08,
|
| 2822 |
+
"loss": 0.0007,
|
| 2823 |
"step": 2340
|
| 2824 |
},
|
| 2825 |
{
|
| 2826 |
"epoch": 0.27,
|
| 2827 |
+
"learning_rate": 1.657555977746972e-08,
|
| 2828 |
"loss": 0.0015,
|
| 2829 |
"step": 2345
|
| 2830 |
},
|
| 2831 |
{
|
| 2832 |
"epoch": 0.27,
|
| 2833 |
+
"learning_rate": 1.3749074502572012e-08,
|
| 2834 |
+
"loss": 0.0009,
|
| 2835 |
"step": 2350
|
| 2836 |
},
|
| 2837 |
{
|
| 2838 |
"epoch": 0.27,
|
| 2839 |
+
"learning_rate": 1.1186285789728247e-08,
|
| 2840 |
+
"loss": 0.0006,
|
| 2841 |
"step": 2355
|
| 2842 |
},
|
| 2843 |
{
|
| 2844 |
"epoch": 0.27,
|
| 2845 |
+
"learning_rate": 8.887329171343717e-09,
|
| 2846 |
+
"loss": 0.001,
|
| 2847 |
"step": 2360
|
| 2848 |
},
|
| 2849 |
{
|
| 2850 |
"epoch": 0.27,
|
| 2851 |
+
"learning_rate": 6.852326227130835e-09,
|
| 2852 |
+
"loss": 0.001,
|
| 2853 |
"step": 2365
|
| 2854 |
},
|
| 2855 |
{
|
| 2856 |
"epoch": 0.27,
|
| 2857 |
+
"learning_rate": 5.08138457768148e-09,
|
| 2858 |
+
"loss": 0.0011,
|
| 2859 |
"step": 2370
|
| 2860 |
},
|
| 2861 |
{
|
| 2862 |
"epoch": 0.27,
|
| 2863 |
+
"learning_rate": 3.574597878777675e-09,
|
| 2864 |
+
"loss": 0.0011,
|
| 2865 |
"step": 2375
|
| 2866 |
},
|
| 2867 |
{
|
| 2868 |
"epoch": 0.27,
|
| 2869 |
+
"learning_rate": 2.3320458164355352e-09,
|
| 2870 |
+
"loss": 0.0006,
|
| 2871 |
"step": 2380
|
| 2872 |
},
|
| 2873 |
{
|
| 2874 |
"epoch": 0.27,
|
| 2875 |
+
"learning_rate": 1.3537941026914302e-09,
|
| 2876 |
+
"loss": 0.0006,
|
| 2877 |
"step": 2385
|
| 2878 |
},
|
| 2879 |
{
|
| 2880 |
"epoch": 0.27,
|
| 2881 |
+
"learning_rate": 6.398944721297539e-10,
|
| 2882 |
+
"loss": 0.0012,
|
| 2883 |
"step": 2390
|
| 2884 |
},
|
| 2885 |
{
|
| 2886 |
"epoch": 0.27,
|
| 2887 |
+
"learning_rate": 1.903846791434516e-10,
|
| 2888 |
+
"loss": 0.0007,
|
| 2889 |
"step": 2395
|
| 2890 |
},
|
| 2891 |
{
|
| 2892 |
"epoch": 0.28,
|
| 2893 |
+
"learning_rate": 5.288495938948757e-12,
|
| 2894 |
+
"loss": 0.001,
|
| 2895 |
"step": 2400
|
| 2896 |
},
|
| 2897 |
{
|
| 2898 |
"epoch": 0.28,
|
| 2899 |
"step": 2400,
|
| 2900 |
"total_flos": 3.3948129315510026e+18,
|
| 2901 |
+
"train_loss": 0.025274057275091764,
|
| 2902 |
+
"train_runtime": 26505.156,
|
| 2903 |
+
"train_samples_per_second": 1.449,
|
| 2904 |
+
"train_steps_per_second": 0.091
|
| 2905 |
}
|
| 2906 |
],
|
| 2907 |
"logging_steps": 5,
|