File size: 49,440 Bytes
fa1c045 |
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 |
========================
START TIME: Thu Jul 4 00:01:38 UTC 2024
python3 version = Python 3.10.14
========================
The token has not been saved to the git credentials helper. Pass `add_to_git_credential=True` in this function directly or `--add-to-git-credential` if using via `huggingface-cli` if you want to set the git credential as well.
Token is valid (permission: write).
Your token has been saved to /admin/home/ferdinand_mom/.cache/huggingface/token
Login successful
Already on 'bench_cluster'
M examples/config_tiny_llama.py
M examples/config_tiny_llama.yaml
M examples/train_tiny_llama.sh
M src/nanotron/models/llama.py
M src/nanotron/trainer.py
Your branch is up to date with 'origin/bench_cluster'.
Job status: RUNNING
W0704 00:01:41.514000 139693704775488 torch/distributed/run.py:757]
W0704 00:01:41.514000 139693704775488 torch/distributed/run.py:757] *****************************************
W0704 00:01:41.514000 139693704775488 torch/distributed/run.py:757] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
W0704 00:01:41.514000 139693704775488 torch/distributed/run.py:757] *****************************************
[default0]:07/04/2024 00:01:57 [WARNING|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Vocab Size Padding] Padded vocab (size: 50257) with 1 dummy tokens (new size: 50258)
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Config:
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Config(general=GeneralArgs(project='bench_cluster',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: run='%date_%jobid',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: seed=42,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: step=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: consumed_train_samples=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: benchmark_csv_path=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: ignore_sanity_checks=True),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: parallelism=ParallelismArgs(dp=2,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pp=2,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tp=2,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f2bf27a5090>,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tp_linear_async_communication=False,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: expert_parallel_size=1),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: model=ModelArgs(model_config=LlamaConfig(bos_token_id=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: eos_token_id=2,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hidden_act='silu',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hidden_size=2048,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: initializer_range=0.02,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: intermediate_size=4096,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: is_llama_config=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: max_position_embeddings=4096,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_attention_heads=32,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_hidden_layers=24,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_key_value_heads=32,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pad_token_id=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pretraining_tp=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rms_norm_eps=1e-05,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rope_scaling=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rope_theta=10000.0,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tie_word_embeddings=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: use_cache=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: vocab_size=50258),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: init_method=RandomInit(std=0.025),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: dtype=torch.bfloat16,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: make_vocab_size_divisible_by=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: ddp_bucket_cap_mb=25),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tokenizer=TokenizerArgs(tokenizer_name_or_path='openai-community/gpt2',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tokenizer_revision=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tokenizer_max_length=None),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: checkpoints=CheckpointsArgs(checkpoints_path=Path('/dev/null'),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: checkpoint_interval=100000,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: save_initial_state=False,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: resume_checkpoint_path=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: checkpoints_path_is_shared_file_system=False),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: logging=LoggingArgs(log_level='info',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: log_level_replica='info',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: iteration_step_info_interval=1),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tokens=TokensArgs(sequence_length=4096,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: train_steps=20,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: micro_batch_size=16,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: batch_accumulation_per_replica=32,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: val_check_interval=-1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: limit_val_batches=0,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: limit_test_batches=0),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: adam_beta1=0.9,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: adam_beta2=0.95,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: torch_adam_is_fused=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: name='adamW'),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: zero_stage=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: weight_decay=0.01,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: clip_grad=1.0,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: accumulate_grad_in_fp32=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0001,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lr_warmup_steps=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lr_warmup_style='linear',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lr_decay_style='linear',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lr_decay_steps=19,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lr_decay_starting_step=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: min_decay_lr=1e-05)),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: data_stages=[DatasetStageArgs(name='Training Stage',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: start_training_step=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: data=DataArgs(dataset=PretrainDatasetsArgs(hf_dataset_or_datasets='roneneldan/TinyStories',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hf_dataset_splits='train',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hf_dataset_config_name=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: dataset_processing_num_proc_per_process=64,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: dataset_overwrite_cache=False,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: text_column_name='text'),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: seed=42,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_loading_workers=0))],
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: profiler=ProfilerArgs(profiler_export_path=Path('/fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-2_pp-2_mbz-16')),
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: lighteval=None)
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Model Config:
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: LlamaConfig(bos_token_id=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: eos_token_id=2,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hidden_act='silu',
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: hidden_size=2048,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: initializer_range=0.02,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: intermediate_size=4096,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: is_llama_config=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: max_position_embeddings=4096,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_attention_heads=32,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_hidden_layers=24,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: num_key_value_heads=32,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pad_token_id=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: pretraining_tp=1,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rms_norm_eps=1e-05,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rope_scaling=None,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: rope_theta=10000.0,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: tie_word_embeddings=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: use_cache=True,
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: vocab_size=50258)
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Building model..
[default0]:07/04/2024 00:01:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Setting PP block ranks...
[default0]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Total number of parameters: 1.21G (2313.02MiB)
[default0]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Local number of parameters: 345M (658.27MiB)
[default0]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [After model building] Memory usage: 672.29MiB. Peak allocated: 674.32MiB Peak reserved: 690.00MiB
[default0]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: No checkpoint path provided.
[default0]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Parametrizing model parameters using StandardParametrizator
[default3]:07/04/2024 00:02:09 [INFO|DP=1|PP=0|TP=1|ip-26-0-169-139]: No checkpoint path provided.
[default2]:07/04/2024 00:02:09 [INFO|DP=1|PP=0|TP=0|ip-26-0-169-139]: No checkpoint path provided.
[default4]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=0|ip-26-0-169-139]: Local number of parameters: 261M (498.24MiB)
[default4]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=0|ip-26-0-169-139]: [After model building] Memory usage: 508.26MiB. Peak allocated: 510.29MiB Peak reserved: 526.00MiB
[default4]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=0|ip-26-0-169-139]: No checkpoint path provided.
[default1]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-139]: Local number of parameters: 345M (658.27MiB)
[default1]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-139]: [After model building] Memory usage: 672.29MiB. Peak allocated: 674.32MiB Peak reserved: 690.00MiB
[default1]:07/04/2024 00:02:09 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-139]: No checkpoint path provided.
[default5]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=1|ip-26-0-169-139]: Local number of parameters: 261M (498.24MiB)
[default5]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=1|ip-26-0-169-139]: [After model building] Memory usage: 508.26MiB. Peak allocated: 510.29MiB Peak reserved: 526.00MiB
[default5]:07/04/2024 00:02:09 [INFO|DP=0|PP=1|TP=1|ip-26-0-169-139]: No checkpoint path provided.
[default7]:07/04/2024 00:02:09 [INFO|DP=1|PP=1|TP=1|ip-26-0-169-139]: No checkpoint path provided.
[default6]:07/04/2024 00:02:09 [INFO|DP=1|PP=1|TP=0|ip-26-0-169-139]: No checkpoint path provided.
[default0]:07/04/2024 00:02:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Optimizer Building] Using LearningRateForSP as learning rate
[default0]:07/04/2024 00:02:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [ZeRO sharding] Size of optimizer params per rank:
[default0]:07/04/2024 00:02:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [ZeRO sharding] DP Rank 0 has 173M out of 345M (50.00%) params' optimizer states
[default0]:07/04/2024 00:02:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [ZeRO sharding] DP Rank 1 has 173M out of 345M (50.00%) params' optimizer states
[default0]:07/04/2024 00:02:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Training Plan] Stage Training Stage has 19 remaining training steps and has consumed 0 samples
[default0]:07/04/2024 00:02:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Using `datasets` library
[default0]:07/04/2024 00:02:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Loading tokenizer from openai-community/gpt2 and transformers/hf_hub versions ('4.41.2', '0.23.4')
[default0]:07/04/2024 00:02:13 [WARNING|DP=0|PP=0|TP=0|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default0]:Repo card metadata block was not found. Setting CardData to empty.
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Training Plan] There are 1 training stages
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Stage Training Stage] start from step 1
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]:
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: [Start training] datetime: 2024-07-04 00:02:14.232785 | mbs: 16 | grad_accum: 32 | global_batch_size: 1024 | sequence_length: 4096 | train_steps: 20 | start_iteration_step: 0 | consumed_train_samples: 0
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Resuming training from stage Training Stage, it has trained for 0 samples and has 19 remaining train steps
[default0]:07/04/2024 00:02:14 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-139]: Memory usage: 2647.09MiB. Peak allocated 2647.09MiB. Peak reserved: 2668.00MiB
[default5]:07/04/2024 00:02:14 [WARNING|DP=0|PP=1|TP=1|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default5]:Repo card metadata block was not found. Setting CardData to empty.
[default3]:07/04/2024 00:02:14 [WARNING|DP=1|PP=0|TP=1|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default2]:07/04/2024 00:02:14 [WARNING|DP=1|PP=0|TP=0|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default4]:07/04/2024 00:02:14 [WARNING|DP=0|PP=1|TP=0|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default1]:07/04/2024 00:02:14 [WARNING|DP=0|PP=0|TP=1|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default4]:Repo card metadata block was not found. Setting CardData to empty.
[default2]:Repo card metadata block was not found. Setting CardData to empty.
[default6]:07/04/2024 00:02:14 [WARNING|DP=1|PP=1|TP=0|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default3]:Repo card metadata block was not found. Setting CardData to empty.
[default7]:07/04/2024 00:02:14 [WARNING|DP=1|PP=1|TP=1|ip-26-0-169-139]: Repo card metadata block was not found. Setting CardData to empty.
[default1]:Repo card metadata block was not found. Setting CardData to empty.
[default6]:Repo card metadata block was not found. Setting CardData to empty.
[default7]:Repo card metadata block was not found. Setting CardData to empty.
[default0]:[rank0]: Traceback (most recent call last):
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
[default0]:[rank0]: trainer.train(dataloader)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
[default0]:[rank0]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
[default0]:[rank0]: outputs = self.pipeline_engine.train_batch_iter(
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
[default0]:[rank0]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
[default0]:[rank0]: output = model(**micro_batch)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
[default0]:[rank0]: sharded_logits = self.model(
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
[default0]:[rank0]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
[default0]:[rank0]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
[default0]:[rank0]: output = self.pp_block(**new_kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 637, in forward
[default0]:[rank0]: hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 172, in forward
[default0]:[rank0]: hidden_states = self.down_proj(self.split_silu_mul(merged_states))
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 128, in forward
[default0]:[rank0]: return self.act(gate_states) * up_states
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default0]:[rank0]: return forward_call(*args, **kwargs)
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/nn/activations.py", line 149, in forward
[default0]:[rank0]: return nn.functional.silu(input)
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/functional.py", line 2102, in silu
[default0]:[rank0]: return torch._C._nn.silu(input)
[default0]:[rank0]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU
[default1]:[rank1]: Traceback (most recent call last):
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
[default1]:[rank1]: trainer.train(dataloader)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
[default1]:[rank1]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
[default1]:[rank1]: outputs = self.pipeline_engine.train_batch_iter(
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
[default1]:[rank1]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
[default1]:[rank1]: output = model(**micro_batch)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
[default1]:[rank1]: sharded_logits = self.model(
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
[default1]:[rank1]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
[default1]:[rank1]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
[default1]:[rank1]: output = self.pp_block(**new_kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 637, in forward
[default1]:[rank1]: hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 171, in forward
[default1]:[rank1]: merged_states = self.gate_up_proj(hidden_states)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default1]:[rank1]: return forward_call(*args, **kwargs)
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 87, in forward
[default1]:[rank1]: return column_linear(
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 359, in column_linear
[default1]:[rank1]: return F.linear(input, weight, bias)
[default1]:[rank1]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU has a total capacity of 79.33 GiB of which 497.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 67.60 GiB is allocated by PyTorch, and 301.87 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
[default3]:[rank3]: Traceback (most recent call last):
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
[default3]:[rank3]: trainer.train(dataloader)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
[default3]:[rank3]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
[default3]:[rank3]: outputs = self.pipeline_engine.train_batch_iter(
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
[default3]:[rank3]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
[default3]:[rank3]: output = model(**micro_batch)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
[default3]:[rank3]: sharded_logits = self.model(
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
[default3]:[rank3]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
[default3]:[rank3]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
[default3]:[rank3]: output = self.pp_block(**new_kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 637, in forward
[default3]:[rank3]: hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 171, in forward
[default3]:[rank3]: merged_states = self.gate_up_proj(hidden_states)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default3]:[rank3]: return self._call_impl(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default3]:[rank3]: return forward_call(*args, **kwargs)
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 87, in forward
[default3]:[rank3]: return column_linear(
[default3]:[rank3]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 359, in column_linear
[default3]:[rank3]: return F.linear(input, weight, bias)
[default3]:[rank3]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU has a total capacity of 79.33 GiB of which 497.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 67.60 GiB is allocated by PyTorch, and 301.87 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
[default2]:[rank2]: Traceback (most recent call last):
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
[default2]:[rank2]: trainer.train(dataloader)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
[default2]:[rank2]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
[default2]:[rank2]: outputs = self.pipeline_engine.train_batch_iter(
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
[default2]:[rank2]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
[default2]:[rank2]: output = model(**micro_batch)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
[default2]:[rank2]: sharded_logits = self.model(
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
[default2]:[rank2]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
[default2]:[rank2]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
[default2]:[rank2]: output = self.pp_block(**new_kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 637, in forward
[default2]:[rank2]: hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 171, in forward
[default2]:[rank2]: merged_states = self.gate_up_proj(hidden_states)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
[default2]:[rank2]: return self._call_impl(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
[default2]:[rank2]: return forward_call(*args, **kwargs)
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 87, in forward
[default2]:[rank2]: return column_linear(
[default2]:[rank2]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 359, in column_linear
[default2]:[rank2]: return F.linear(input, weight, bias)
[default2]:[rank2]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU has a total capacity of 79.33 GiB of which 497.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 67.60 GiB is allocated by PyTorch, and 301.87 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
W0704 00:02:26.667000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677683 closing signal SIGTERM
W0704 00:02:26.668000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677684 closing signal SIGTERM
W0704 00:02:26.668000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677685 closing signal SIGTERM
W0704 00:02:26.668000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677686 closing signal SIGTERM
W0704 00:02:26.668000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677687 closing signal SIGTERM
W0704 00:02:26.668000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677688 closing signal SIGTERM
W0704 00:02:26.670000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 677689 closing signal SIGTERM
E0704 00:02:28.584000 139693704775488 torch/distributed/elastic/multiprocessing/api.py:826] failed (exitcode: 1) local_rank: 0 (pid: 677682) of binary: /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10
Traceback (most recent call last):
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/torchrun", line 8, in <module>
sys.exit(main())
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 347, in wrapper
return f(*args, **kwargs)
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/run.py", line 879, in main
run(args)
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/run.py", line 870, in run
elastic_launch(
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 132, in __call__
return launch_agent(self._config, self._entrypoint, list(args))
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 263, in launch_agent
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
============================================================
/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py FAILED
------------------------------------------------------------
Failures:
<NO_OTHER_FAILURES>
------------------------------------------------------------
Root Cause (first observed failure):
[0]:
time : 2024-07-04_00:02:26
host : ip-26-0-169-139.ec2.internal
rank : 0 (local_rank: 0)
exitcode : 1 (pid: 677682)
error_file: <N/A>
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
============================================================
srun: error: ip-26-0-169-139: task 0: Exited with exit code 1
Consider using `hf_transfer` for faster uploads. This solution comes with some limitations. See https://huggingface.co/docs/huggingface_hub/hf_transfer for more details.
|