| 2025-03-08 23:01:33 - WARNING - __main__ - Process rank: 0, device: cuda:0, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:33 - INFO - __main__ - Model parameters ModelConfig(model_name_or_path='deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', model_revision='main', torch_dtype='bfloat16', trust_remote_code=False, attn_implementation='flash_attention_2', use_peft=False, lora_r=16, lora_alpha=32, lora_dropout=0.05, lora_target_modules=None, lora_modules_to_save=None, lora_task_type='CAUSAL_LM', use_rslora=False, load_in_8bit=False, load_in_4bit=False, bnb_4bit_quant_type='nf4', use_bnb_nested_quant=False) | |
| 2025-03-08 23:01:33 - INFO - __main__ - Script parameters GRPOScriptArguments(dataset_name='AI-MO/NuminaMath-TIR', dataset_config=None, dataset_train_split='train', dataset_test_split='test', gradient_checkpointing_use_reentrant=False, ignore_bias_buffers=False, reward_funcs=['accuracy', 'format', 'tag_count'], cosine_min_value_wrong=0.0, cosine_max_value_wrong=-0.5, cosine_min_value_correct=0.5, cosine_max_value_correct=1.0, cosine_max_len=1000, repetition_n_grams=3, repetition_max_penalty=-1.0, code_language='python') | |
| 2025-03-08 23:01:33 - INFO - __main__ - Training parameters GRPOConfig( | |
| _n_gpu=1, | |
| accelerator_config={'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, | |
| adafactor=False, | |
| adam_beta1=0.9, | |
| adam_beta2=0.999, | |
| adam_epsilon=1e-08, | |
| auto_find_batch_size=False, | |
| average_tokens_across_devices=False, | |
| batch_eval_metrics=False, | |
| benchmarks=[], | |
| beta=0.04, | |
| bf16=True, | |
| bf16_full_eval=False, | |
| callbacks=[], | |
| chat_template=None, | |
| data_seed=None, | |
| dataloader_drop_last=False, | |
| dataloader_num_workers=0, | |
| dataloader_persistent_workers=False, | |
| dataloader_pin_memory=True, | |
| dataloader_prefetch_factor=None, | |
| ddp_backend=None, | |
| ddp_broadcast_buffers=None, | |
| ddp_bucket_cap_mb=None, | |
| ddp_find_unused_parameters=None, | |
| ddp_timeout=1800, | |
| debug=[], | |
| deepspeed=None, | |
| disable_tqdm=False, | |
| dispatch_batches=None, | |
| do_eval=True, | |
| do_predict=False, | |
| do_train=False, | |
| ds3_gather_for_generation=True, | |
| epsilon=0.2, | |
| eval_accumulation_steps=None, | |
| eval_delay=0, | |
| eval_do_concat_batches=True, | |
| eval_on_start=False, | |
| eval_steps=100, | |
| eval_strategy=IntervalStrategy.STEPS, | |
| eval_use_gather_object=False, | |
| evaluation_strategy=None, | |
| fp16=False, | |
| fp16_backend=auto, | |
| fp16_full_eval=False, | |
| fp16_opt_level=O1, | |
| fsdp=[], | |
| fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, | |
| fsdp_min_num_params=0, | |
| fsdp_transformer_layer_cls_to_wrap=None, | |
| full_determinism=False, | |
| gradient_accumulation_steps=16, | |
| gradient_checkpointing=True, | |
| gradient_checkpointing_kwargs={'use_reentrant': False}, | |
| greater_is_better=None, | |
| group_by_length=False, | |
| half_precision_backend=auto, | |
| hub_always_push=False, | |
| hub_model_id=DeepSeek-R1-Distill-Qwen-7B-GRPO, | |
| hub_model_revision=main, | |
| hub_private_repo=None, | |
| hub_strategy=HubStrategy.EVERY_SAVE, | |
| hub_token=<HUB_TOKEN>, | |
| ignore_data_skip=False, | |
| include_for_metrics=[], | |
| include_inputs_for_metrics=False, | |
| include_num_input_tokens_seen=False, | |
| include_tokens_per_second=False, | |
| jit_mode_eval=False, | |
| label_names=None, | |
| label_smoothing_factor=0.0, | |
| learning_rate=2e-05, | |
| length_column_name=length, | |
| load_best_model_at_end=False, | |
| local_rank=0, | |
| log_completions=False, | |
| log_level=info, | |
| log_level_replica=warning, | |
| log_on_each_node=True, | |
| logging_dir=output/DeepSeek-R1-Distill-Qwen-7B/runs/Mar08_23-01-32_ablation-scale-0-0, | |
| logging_first_step=False, | |
| logging_nan_inf_filter=True, | |
| logging_steps=5, | |
| logging_strategy=IntervalStrategy.STEPS, | |
| lr_scheduler_kwargs={}, | |
| lr_scheduler_type=SchedulerType.COSINE, | |
| max_completion_length=1024, | |
| max_grad_norm=1.0, | |
| max_prompt_length=512, | |
| max_steps=-1, | |
| metric_for_best_model=None, | |
| model_init_kwargs=None, | |
| mp_parameters=, | |
| neftune_noise_alpha=None, | |
| no_cuda=False, | |
| num_generations=7, | |
| num_iterations=1, | |
| num_train_epochs=1, | |
| optim=OptimizerNames.ADAMW_TORCH, | |
| optim_args=None, | |
| optim_target_modules=None, | |
| output_dir=output/DeepSeek-R1-Distill-Qwen-7B, | |
| overwrite_hub_revision=False, | |
| overwrite_output_dir=True, | |
| past_index=-1, | |
| per_device_eval_batch_size=32, | |
| per_device_train_batch_size=16, | |
| prediction_loss_only=False, | |
| push_to_hub=True, | |
| push_to_hub_model_id=None, | |
| push_to_hub_organization=None, | |
| push_to_hub_revision=False, | |
| push_to_hub_token=<PUSH_TO_HUB_TOKEN>, | |
| ray_scope=last, | |
| ref_model_mixup_alpha=0.6, | |
| ref_model_sync_steps=512, | |
| remove_unused_columns=False, | |
| report_to=['wandb'], | |
| restore_callback_states_from_checkpoint=False, | |
| resume_from_checkpoint=None, | |
| reward_weights=None, | |
| run_name=output/DeepSeek-R1-Distill-Qwen-7B, | |
| save_on_each_node=False, | |
| save_only_model=False, | |
| save_safetensors=True, | |
| save_steps=500, | |
| save_strategy=SaveStrategy.NO, | |
| save_total_limit=None, | |
| seed=42, | |
| skip_memory_metrics=True, | |
| split_batches=None, | |
| sync_ref_model=False, | |
| system_prompt=None, | |
| temperature=0.9, | |
| tf32=None, | |
| torch_compile=False, | |
| torch_compile_backend=None, | |
| torch_compile_mode=None, | |
| torch_empty_cache_steps=None, | |
| torchdynamo=None, | |
| tpu_metrics_debug=False, | |
| tpu_num_cores=None, | |
| use_cpu=False, | |
| use_ipex=False, | |
| use_legacy_prediction_loop=False, | |
| use_liger_kernel=False, | |
| use_mps_device=False, | |
| use_vllm=True, | |
| vllm_device=auto, | |
| vllm_dtype=auto, | |
| vllm_enable_prefix_caching=True, | |
| vllm_gpu_memory_utilization=0.7, | |
| vllm_guided_decoding_regex=None, | |
| vllm_max_model_len=None, | |
| wandb_entity=None, | |
| wandb_project=None, | |
| warmup_ratio=0.1, | |
| warmup_steps=0, | |
| weight_decay=0.0, | |
| ) | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 3, device: cuda:3, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 4, device: cuda:4, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 2, device: cuda:2, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 6, device: cuda:6, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 1, device: cuda:1, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:34 - WARNING - __main__ - Process rank: 5, device: cuda:5, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-08 23:01:37 - INFO - datasets.builder - Found cached dataset numina_math-tir (/home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd) | |
| 2025-03-08 23:01:37 - INFO - datasets.info - Loading Dataset info from /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd | |
| 2025-03-08 23:01:39 - INFO - __main__ - Using Qwen model, set tokenizer.padding_side = 'left'. | |
| 2025-03-08 23:01:39 - INFO - datasets.arrow_dataset - Caching processed dataset at /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd/cache-ec164f4fa43b91e3.arrow | |
| 2025-03-08 23:01:42 - INFO - datasets.arrow_dataset - Caching processed dataset at /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd/cache-38ec81708e5e09b2.arrow | |
| 2025-03-08 23:01:42 - INFO - __main__ - *** Initializing model kwargs *** | |
| 2025-03-08 23:04:16 - INFO - __main__ - *** Train *** | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 0, device: cuda:0, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - INFO - __main__ - Model parameters ModelConfig(model_name_or_path='deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', model_revision='main', torch_dtype='bfloat16', trust_remote_code=False, attn_implementation='flash_attention_2', use_peft=False, lora_r=16, lora_alpha=32, lora_dropout=0.05, lora_target_modules=None, lora_modules_to_save=None, lora_task_type='CAUSAL_LM', use_rslora=False, load_in_8bit=False, load_in_4bit=False, bnb_4bit_quant_type='nf4', use_bnb_nested_quant=False) | |
| 2025-03-09 20:08:16 - INFO - __main__ - Script parameters GRPOScriptArguments(dataset_name='AI-MO/NuminaMath-TIR', dataset_config=None, dataset_train_split='train', dataset_test_split='test', gradient_checkpointing_use_reentrant=False, ignore_bias_buffers=False, reward_funcs=['accuracy', 'format', 'tag_count'], cosine_min_value_wrong=0.0, cosine_max_value_wrong=-0.5, cosine_min_value_correct=0.5, cosine_max_value_correct=1.0, cosine_max_len=1000, repetition_n_grams=3, repetition_max_penalty=-1.0, code_language='python') | |
| 2025-03-09 20:08:16 - INFO - __main__ - Training parameters GRPOConfig( | |
| _n_gpu=1, | |
| accelerator_config={'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, | |
| adafactor=False, | |
| adam_beta1=0.9, | |
| adam_beta2=0.999, | |
| adam_epsilon=1e-08, | |
| auto_find_batch_size=False, | |
| average_tokens_across_devices=False, | |
| batch_eval_metrics=False, | |
| benchmarks=[], | |
| beta=0.04, | |
| bf16=True, | |
| bf16_full_eval=False, | |
| callbacks=[], | |
| chat_template=None, | |
| data_seed=None, | |
| dataloader_drop_last=False, | |
| dataloader_num_workers=0, | |
| dataloader_persistent_workers=False, | |
| dataloader_pin_memory=True, | |
| dataloader_prefetch_factor=None, | |
| ddp_backend=None, | |
| ddp_broadcast_buffers=None, | |
| ddp_bucket_cap_mb=None, | |
| ddp_find_unused_parameters=None, | |
| ddp_timeout=1800, | |
| debug=[], | |
| deepspeed=None, | |
| disable_tqdm=False, | |
| dispatch_batches=None, | |
| do_eval=True, | |
| do_predict=False, | |
| do_train=False, | |
| ds3_gather_for_generation=True, | |
| epsilon=0.2, | |
| eval_accumulation_steps=None, | |
| eval_delay=0, | |
| eval_do_concat_batches=True, | |
| eval_on_start=False, | |
| eval_steps=100, | |
| eval_strategy=IntervalStrategy.STEPS, | |
| eval_use_gather_object=False, | |
| evaluation_strategy=None, | |
| fp16=False, | |
| fp16_backend=auto, | |
| fp16_full_eval=False, | |
| fp16_opt_level=O1, | |
| fsdp=[], | |
| fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, | |
| fsdp_min_num_params=0, | |
| fsdp_transformer_layer_cls_to_wrap=None, | |
| full_determinism=False, | |
| gradient_accumulation_steps=16, | |
| gradient_checkpointing=True, | |
| gradient_checkpointing_kwargs={'use_reentrant': False}, | |
| greater_is_better=None, | |
| group_by_length=False, | |
| half_precision_backend=auto, | |
| hub_always_push=False, | |
| hub_model_id=DeepSeek-R1-Distill-Qwen-7B-GRPO, | |
| hub_model_revision=main, | |
| hub_private_repo=None, | |
| hub_strategy=HubStrategy.EVERY_SAVE, | |
| hub_token=<HUB_TOKEN>, | |
| ignore_data_skip=False, | |
| include_for_metrics=[], | |
| include_inputs_for_metrics=False, | |
| include_num_input_tokens_seen=False, | |
| include_tokens_per_second=False, | |
| jit_mode_eval=False, | |
| label_names=None, | |
| label_smoothing_factor=0.0, | |
| learning_rate=2e-05, | |
| length_column_name=length, | |
| load_best_model_at_end=False, | |
| local_rank=0, | |
| log_completions=False, | |
| log_level=info, | |
| log_level_replica=warning, | |
| log_on_each_node=True, | |
| logging_dir=output/DeepSeek-R1-Distill-Qwen-7B/runs/Mar09_20-08-15_job-2025-03-09t05-35-43-0-0, | |
| logging_first_step=False, | |
| logging_nan_inf_filter=True, | |
| logging_steps=5, | |
| logging_strategy=IntervalStrategy.STEPS, | |
| lr_scheduler_kwargs={}, | |
| lr_scheduler_type=SchedulerType.COSINE, | |
| max_completion_length=1024, | |
| max_grad_norm=1.0, | |
| max_prompt_length=512, | |
| max_steps=-1, | |
| metric_for_best_model=None, | |
| model_init_kwargs=None, | |
| mp_parameters=, | |
| neftune_noise_alpha=None, | |
| no_cuda=False, | |
| num_generations=7, | |
| num_iterations=1, | |
| num_train_epochs=1, | |
| optim=OptimizerNames.ADAMW_TORCH, | |
| optim_args=None, | |
| optim_target_modules=None, | |
| output_dir=output/DeepSeek-R1-Distill-Qwen-7B, | |
| overwrite_hub_revision=False, | |
| overwrite_output_dir=True, | |
| past_index=-1, | |
| per_device_eval_batch_size=32, | |
| per_device_train_batch_size=16, | |
| prediction_loss_only=False, | |
| push_to_hub=True, | |
| push_to_hub_model_id=None, | |
| push_to_hub_organization=None, | |
| push_to_hub_revision=False, | |
| push_to_hub_token=<PUSH_TO_HUB_TOKEN>, | |
| ray_scope=last, | |
| ref_model_mixup_alpha=0.6, | |
| ref_model_sync_steps=512, | |
| remove_unused_columns=False, | |
| report_to=['wandb'], | |
| restore_callback_states_from_checkpoint=False, | |
| resume_from_checkpoint=None, | |
| reward_weights=None, | |
| run_name=output/DeepSeek-R1-Distill-Qwen-7B, | |
| save_on_each_node=False, | |
| save_only_model=False, | |
| save_safetensors=True, | |
| save_steps=500, | |
| save_strategy=SaveStrategy.NO, | |
| save_total_limit=None, | |
| seed=42, | |
| skip_memory_metrics=True, | |
| split_batches=None, | |
| sync_ref_model=False, | |
| system_prompt=None, | |
| temperature=0.9, | |
| tf32=None, | |
| torch_compile=False, | |
| torch_compile_backend=None, | |
| torch_compile_mode=None, | |
| torch_empty_cache_steps=None, | |
| torchdynamo=None, | |
| tpu_metrics_debug=False, | |
| tpu_num_cores=None, | |
| use_cpu=False, | |
| use_ipex=False, | |
| use_legacy_prediction_loop=False, | |
| use_liger_kernel=False, | |
| use_mps_device=False, | |
| use_vllm=True, | |
| vllm_device=auto, | |
| vllm_dtype=auto, | |
| vllm_enable_prefix_caching=True, | |
| vllm_gpu_memory_utilization=0.7, | |
| vllm_guided_decoding_regex=None, | |
| vllm_max_model_len=None, | |
| wandb_entity=None, | |
| wandb_project=None, | |
| warmup_ratio=0.1, | |
| warmup_steps=0, | |
| weight_decay=0.0, | |
| ) | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 5, device: cuda:5, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 6, device: cuda:6, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 4, device: cuda:4, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 1, device: cuda:1, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 2, device: cuda:2, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:16 - WARNING - __main__ - Process rank: 3, device: cuda:3, n_gpu: 1 distributed training: True, 16-bits training: False | |
| 2025-03-09 20:08:20 - INFO - datasets.builder - Found cached dataset numina_math-tir (/home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd) | |
| 2025-03-09 20:08:20 - INFO - datasets.info - Loading Dataset info from /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd | |
| 2025-03-09 20:08:22 - INFO - __main__ - Using Qwen model, set tokenizer.padding_side = 'left'. | |
| 2025-03-09 20:08:22 - INFO - datasets.arrow_dataset - Caching processed dataset at /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd/cache-6329233ef4bcb2f8.arrow | |
| 2025-03-09 20:08:26 - INFO - datasets.arrow_dataset - Caching processed dataset at /home/yizhouw/.cache/huggingface/datasets/AI-MO___numina_math-tir/default/0.0.0/77a91d7b7a1a98ac4b1beb7d86c09d156b935dcd/cache-531ceb5bbf98770c.arrow | |
| 2025-03-09 20:08:26 - INFO - __main__ - *** Initializing model kwargs *** | |
| 2025-03-09 20:11:10 - INFO - __main__ - *** Train *** | |
| 2025-03-11 05:21:06 - INFO - __main__ - *** Save model *** | |