File size: 14,888 Bytes
409a825 6660cc4 315a9a7 b703d56 0354e7d 7f48aac |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 |
0%| | 0/478 [00:00<?, ?it/s][WARNING|modeling_utils.py:1188] 2024-04-26 15:57:21,671 >> Could not estimate the number of tokens of the input, floating-point operations will not be computed
0%| | 2/478 [00:03<12:43, 1.60s/it]
5%|▌ | 25/478 [00:32<09:39, 1.28s/it]
10%|█ | 50/478 [01:05<09:13, 1.29s/it]
16%|█▌ | 75/478 [01:37<08:37, 1.28s/it]
21%|██ | 100/478 [02:09<08:04, 1.28s/it][INFO|trainer.py:3614] 2024-04-26 15:59:29,412 >> ***** Running Evaluation *****
[INFO|trainer.py:3616] 2024-04-26 15:59:29,412 >> Num examples = 2000
[INFO|trainer.py:3619] 2024-04-26 15:59:29,412 >> Batch size = 8
6%|▋ | 2/32 [00:00<00:03, 8.89it/s]
[INFO|configuration_utils.py:471] 2024-04-26 15:59:37,711 >> Configuration saved in ./checkpoint-100/config.json
[INFO|configuration_utils.py:697] 2024-04-26 15:59:37,713 >> Configuration saved in ./checkpoint-100/generation_config.json
{'eval_loss': 0.6759119629859924, 'eval_runtime': 8.2733, 'eval_samples_per_second': 241.742, 'eval_steps_per_second': 3.868, 'eval_rewards/chosen': 0.0017230990342795849, 'eval_rewards/rejected': -0.03281649947166443, 'eval_rewards/accuracies': 0.62890625, 'eval_rewards/margins': 0.0345395989716053, 'eval_logps/rejected': -407.8036804199219, 'eval_logps/chosen': -423.0196533203125, 'eval_logits/rejected': -3.2565112113952637, 'eval_logits/chosen': -3.313567638397217, 'epoch': 0.21}
[INFO|modeling_utils.py:2598] 2024-04-26 15:59:47,330 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./checkpoint-100/model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 15:59:47,344 >> tokenizer config file saved in ./checkpoint-100/tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 15:59:47,384 >> Special tokens file saved in ./checkpoint-100/special_tokens_map.json
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:00:07,103 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:00:07,105 >> Special tokens file saved in ./special_tokens_map.json
26%|██▋ | 126/478 [03:19<07:28, 1.27s/it]
32%|███▏ | 151/478 [03:51<06:56, 1.28s/it]
37%|███▋ | 176/478 [04:23<06:28, 1.29s/it]
42%|████▏ | 200/478 [04:54<05:52, 1.27s/it][INFO|trainer.py:3614] 2024-04-26 16:02:14,470 >> ***** Running Evaluation *****
[INFO|trainer.py:3616] 2024-04-26 16:02:14,470 >> Num examples = 2000
[INFO|trainer.py:3619] 2024-04-26 16:02:14,470 >> Batch size = 8
19%|█▉ | 6/32 [00:01<00:06, 4.22it/s]
[INFO|configuration_utils.py:471] 2024-04-26 16:02:22,770 >> Configuration saved in ./checkpoint-200/config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:02:22,773 >> Configuration saved in ./checkpoint-200/generation_config.json
{'eval_loss': 0.6533502340316772, 'eval_runtime': 8.2763, 'eval_samples_per_second': 241.653, 'eval_steps_per_second': 3.866, 'eval_rewards/chosen': -0.06664139777421951, 'eval_rewards/rejected': -0.16173213720321655, 'eval_rewards/accuracies': 0.64453125, 'eval_rewards/margins': 0.09509073942899704, 'eval_logps/rejected': -420.6952209472656, 'eval_logps/chosen': -429.85614013671875, 'eval_logits/rejected': -3.2240023612976074, 'eval_logits/chosen': -3.2767982482910156, 'epoch': 0.42}
[INFO|modeling_utils.py:2598] 2024-04-26 16:02:32,167 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./checkpoint-200/model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:02:32,170 >> tokenizer config file saved in ./checkpoint-200/tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:02:32,172 >> Special tokens file saved in ./checkpoint-200/special_tokens_map.json
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:02:50,674 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:02:50,676 >> Special tokens file saved in ./special_tokens_map.json
[INFO|trainer.py:3397] 2024-04-26 16:02:50,704 >> Deleting older checkpoint [checkpoint-100] due to args.save_total_limit
47%|████▋ | 226/478 [06:05<05:20, 1.27s/it]
52%|█████▏ | 250/478 [06:36<05:00, 1.32s/it]
58%|█████▊ | 275/478 [07:09<04:25, 1.31s/it]
63%|██████▎ | 300/478 [07:42<03:53, 1.31s/it][INFO|trainer.py:3614] 2024-04-26 16:05:02,769 >> ***** Running Evaluation *****
[INFO|trainer.py:3616] 2024-04-26 16:05:02,769 >> Num examples = 2000
[INFO|trainer.py:3619] 2024-04-26 16:05:02,769 >> Batch size = 8
12%|█▎ | 4/32 [00:00<00:05, 5.09it/s]
[INFO|configuration_utils.py:471] 2024-04-26 16:05:11,146 >> Configuration saved in ./checkpoint-300/config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:05:11,149 >> Configuration saved in ./checkpoint-300/generation_config.json
{'eval_loss': 0.6438009142875671, 'eval_runtime': 8.3559, 'eval_samples_per_second': 239.351, 'eval_steps_per_second': 3.83, 'eval_rewards/chosen': -0.10771973431110382, 'eval_rewards/rejected': -0.24101632833480835, 'eval_rewards/accuracies': 0.62109375, 'eval_rewards/margins': 0.13329659402370453, 'eval_logps/rejected': -428.6236572265625, 'eval_logps/chosen': -433.9639892578125, 'eval_logits/rejected': -3.2049574851989746, 'eval_logits/chosen': -3.2553329467773438, 'epoch': 0.63}
[INFO|modeling_utils.py:2598] 2024-04-26 16:05:20,618 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./checkpoint-300/model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:05:20,621 >> tokenizer config file saved in ./checkpoint-300/tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:05:20,623 >> Special tokens file saved in ./checkpoint-300/special_tokens_map.json
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:05:39,112 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:05:39,114 >> Special tokens file saved in ./special_tokens_map.json
[INFO|trainer.py:3397] 2024-04-26 16:05:39,143 >> Deleting older checkpoint [checkpoint-200] due to args.save_total_limit
68%|██████▊ | 325/478 [08:53<03:20, 1.31s/it]
73%|███████▎ | 351/478 [09:27<02:47, 1.32s/it]
79%|███████▊ | 376/478 [10:01<02:20, 1.37s/it]
84%|████████▎ | 400/478 [10:34<01:45, 1.35s/it][INFO|trainer.py:3614] 2024-04-26 16:07:54,737 >> ***** Running Evaluation *****
[INFO|trainer.py:3616] 2024-04-26 16:07:54,737 >> Num examples = 2000
[INFO|trainer.py:3619] 2024-04-26 16:07:54,737 >> Batch size = 8
12%|█▎ | 4/32 [00:00<00:07, 3.96it/s]
[INFO|configuration_utils.py:471] 2024-04-26 16:08:03,486 >> Configuration saved in ./checkpoint-400/config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:08:03,490 >> Configuration saved in ./checkpoint-400/generation_config.json
{'eval_loss': 0.6415477395057678, 'eval_runtime': 8.7287, 'eval_samples_per_second': 229.13, 'eval_steps_per_second': 3.666, 'eval_rewards/chosen': -0.10007989406585693, 'eval_rewards/rejected': -0.24366310238838196, 'eval_rewards/accuracies': 0.62109375, 'eval_rewards/margins': 0.14358317852020264, 'eval_logps/rejected': -428.88836669921875, 'eval_logps/chosen': -433.20001220703125, 'eval_logits/rejected': -3.204622507095337, 'eval_logits/chosen': -3.254263401031494, 'epoch': 0.84}
[INFO|modeling_utils.py:2598] 2024-04-26 16:08:13,074 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./checkpoint-400/model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:08:13,077 >> tokenizer config file saved in ./checkpoint-400/tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:08:13,079 >> Special tokens file saved in ./checkpoint-400/special_tokens_map.json
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:08:31,839 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:08:31,841 >> Special tokens file saved in ./special_tokens_map.json
[INFO|trainer.py:3397] 2024-04-26 16:08:31,870 >> Deleting older checkpoint [checkpoint-300] due to args.save_total_limit
89%|████████▉ | 425/478 [11:46<01:11, 1.35s/it]
94%|█████████▍| 450/478 [12:20<00:37, 1.35s/it]
99%|█████████▉| 475/478 [12:54<00:04, 1.35s/it]
100%|██████████| 478/478 [12:58<00:00, 1.35s/it][INFO|trainer.py:2316] 2024-04-26 16:10:19,036 >>
Training completed. Do not forget to share your model on huggingface.co/models =)
100%|██████████| 478/478 [12:58<00:00, 1.63s/it]
[INFO|trainer.py:3614] 2024-04-26 16:10:19,102 >> ***** Running Evaluation *****
[INFO|trainer.py:3616] 2024-04-26 16:10:19,102 >> Num examples = 2000
[INFO|trainer.py:3619] 2024-04-26 16:10:19,102 >> Batch size = 8
12%|█▎ | 4/32 [00:00<00:05, 5.08it/s]
{'train_runtime': 784.6622, 'train_samples_per_second': 77.913, 'train_steps_per_second': 0.609, 'train_loss': 0.6571792745689967, 'epoch': 1.0}
***** train metrics *****
epoch = 1.0
total_flos = 0GF
train_loss = 0.6572
train_runtime = 0:13:04.66
train_samples = 61135
train_samples_per_second = 77.913
train_steps_per_second = 0.609
2024-04-26 16:10:19 - INFO - __main__ - *** Training complete ***
100%|██████████| 32/32 [00:08<00:00, 3.97it/s]
[INFO|trainer.py:3305] 2024-04-26 16:10:27,430 >> Saving model checkpoint to ./
[INFO|configuration_utils.py:471] 2024-04-26 16:10:27,432 >> Configuration saved in ./config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:10:27,434 >> Configuration saved in ./generation_config.json
***** eval metrics *****
epoch = 1.0
eval_logits/chosen = -3.2544
eval_logits/rejected = -3.2047
eval_logps/chosen = -433.6304
eval_logps/rejected = -429.4582
eval_loss = 0.6412
eval_rewards/accuracies = 0.6445
eval_rewards/chosen = -0.1044
eval_rewards/margins = 0.145
eval_rewards/rejected = -0.2494
eval_runtime = 0:00:08.29
eval_samples = 2000
eval_samples_per_second = 241.204
eval_steps_per_second = 3.859
2024-04-26 16:10:27 - INFO - __main__ - *** Save model ***
[INFO|modeling_utils.py:2598] 2024-04-26 16:10:37,122 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:10:37,133 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:10:37,135 >> Special tokens file saved in ./special_tokens_map.json
[INFO|trainer.py:3305] 2024-04-26 16:10:37,190 >> Saving model checkpoint to ./
[INFO|configuration_utils.py:471] 2024-04-26 16:10:37,192 >> Configuration saved in ./config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:10:37,194 >> Configuration saved in ./generation_config.json
[INFO|modeling_utils.py:2598] 2024-04-26 16:10:48,100 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:10:48,103 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:10:48,105 >> Special tokens file saved in ./special_tokens_map.json
[INFO|modelcard.py:450] 2024-04-26 16:10:48,202 >> Dropping the following result as it does not have all the necessary fields:
{'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}}
events.out.tfevents.1714147827.ip-26-0-160-225.711598.1: 100%|██████████| 828/828 [00:00<00:00, 5.29kB/s]
events.out.tfevents.1714147034.ip-26-0-160-225.711598.0: 100%|██████████| 21.8k/21.8k [00:00<00:00, 108kB/s]
model-00001-of-00002.safetensors: 1%| | 32.0M/4.99G [00:00<02:33, 32.3MB/s]
events.out.tfevents.1714147034.ip-26-0-160-225.711598.0: 0%| | 0.00/21.8k [00:00<?, ?B/s]
Upload 4 LFS files: 100%|██████████| 4/4 [01:26<00:00, 21.75s/it]0:24<00:00, 57.8MB/s]:00<?, ?B/s]
[INFO|modelcard.py:450] 2024-04-26 16:12:22,694 >> Dropping the following result as it does not have all the necessary fields:
{'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}, 'dataset': {'name': 'HuggingFaceH4/ultrafeedback_binarized', 'type': 'HuggingFaceH4/ultrafeedback_binarized'}}
[INFO|configuration_utils.py:471] 2024-04-26 16:12:22,700 >> Configuration saved in ./config.json
[INFO|trainer.py:3305] 2024-04-26 16:12:22,704 >> Saving model checkpoint to ./
[INFO|configuration_utils.py:471] 2024-04-26 16:12:22,706 >> Configuration saved in ./config.json
[INFO|configuration_utils.py:697] 2024-04-26 16:12:22,708 >> Configuration saved in ./generation_config.json
2024-04-26 16:12:22 - INFO - __main__ - Model saved to ./
2024-04-26 16:12:22 - INFO - __main__ - Pushing to hub...
[INFO|modeling_utils.py:2598] 2024-04-26 16:12:33,635 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 2 checkpoint shards. You can find where each parameters has been saved in the index located at ./model.safetensors.index.json.
[INFO|tokenization_utils_base.py:2488] 2024-04-26 16:12:33,638 >> tokenizer config file saved in ./tokenizer_config.json
[INFO|tokenization_utils_base.py:2497] 2024-04-26 16:12:33,640 >> Special tokens file saved in ./special_tokens_map.json
[INFO|modelcard.py:450] 2024-04-26 16:12:33,786 >> Dropping the following result as it does not have all the necessary fields:
|