diff --git "a/debug.log" "b/debug.log" new file mode 100644--- /dev/null +++ "b/debug.log" @@ -0,0 +1,1243 @@ +[2026-01-28 17:33:01,183] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:251] bf16 support detected, enabling for this configuration. +[2026-01-28 17:33:01,183] [WARNING] [py.warnings._showwarnmsg:110] [PID:251] /root/miniconda3/envs/py3.11/lib/python3.11/site-packages/torch/backends/__init__.py:46: UserWarning: Please use the new API settings to control TF32 behavior, such as torch.backends.cudnn.conv.fp32_precision = 'tf32' or torch.backends.cuda.matmul.fp32_precision = 'ieee'. Old settings, e.g, torch.backends.cuda.matmul.allow_tf32 = True, torch.backends.cudnn.allow_tf32 = True, allowTF32CuDNN() and allowTF32CuBLAS() will be deprecated after Pytorch 2.9. Please see https://pytorch.org/docs/main/notes/cuda.html#tensorfloat-32-tf32-on-ampere-and-later-devices (Triggered internally at /pytorch/aten/src/ATen/Context.cpp:80.) + self.setter(val) + +[2026-01-28 17:33:01,188] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:251] baseline 0.000GB () +[2026-01-28 17:33:01,189] [INFO] [axolotl.cli.config.load_cfg:259] [PID:251] config: +{ + "activation_offloading": false, + "adapter": "lora", + "axolotl_config_path": "/workspace/axolotl/configs/1.yml", + "base_model": "/cache/models/Qwen--Qwen2.5-3B-Instruct", + "base_model_config": "/cache/models/Qwen--Qwen2.5-3B-Instruct", + "batch_size": 64, + "bf16": true, + "capabilities": { + "bf16": true, + "compute_capability": "sm_90", + "fp8": true, + "n_gpu": 2, + "n_node": 1 + }, + "chat_template": "llama3", + "context_parallel_size": 1, + "dataloader_num_workers": 2, + "dataloader_pin_memory": true, + "dataloader_prefetch_factor": 256, + "dataset_num_proc": 52, + "datasets": [ + { + "data_files": [ + "1_train_data.json" + ], + "ds_type": "json", + "message_property_mappings": { + "content": "content", + "role": "role" + }, + "path": "/workspace/axolotl/data", + "split": "train", + "trust_remote_code": false + } + ], + "ddp": true, + "device": "cuda:0", + "device_map": { + "": 0 + }, + "dion_rank_fraction": 1.0, + "dion_rank_multiple_of": 1, + "eaft_alpha": 1.0, + "eaft_k": 20, + "env_capabilities": { + "torch_version": "2.9.0" + }, + "eval_batch_size": 2, + "eval_causal_lm_metrics": [ + "sacrebleu", + "comet", + "ter", + "chrf" + ], + "eval_max_new_tokens": 128, + "eval_strategy": "no", + "eval_table_size": 0, + "experimental_skip_move_to_device": true, + "flash_attention": false, + "fp16": false, + "gradient_accumulation_steps": 16, + "gradient_checkpointing": true, + "group_by_length": false, + "include_tkps": true, + "is_falcon_derived_model": false, + "is_llama_derived_model": false, + "is_mistral_derived_model": false, + "learning_rate": 2.5e-05, + "lisa_layers_attribute": "model.layers", + "load_best_model_at_end": false, + "load_in_4bit": false, + "load_in_8bit": false, + "local_rank": 0, + "logging_steps": 1, + "lora_alpha": 32, + "lora_dropout": 0.0, + "lora_r": 32, + "lora_target_linear": true, + "loraplus_lr_embedding": 1e-06, + "lr_scheduler": "cosine", + "max_grad_norm": 1.0, + "max_steps": 1000, + "mean_resizing_embeddings": false, + "micro_batch_size": 2, + "mlflow_experiment_name": "/workspace/axolotl/data/1_train_data.json", + "model_config_type": "qwen2", + "num_epochs": 1.0, + "optimizer": "adamw_bnb_8bit", + "otel_metrics_host": "localhost", + "otel_metrics_port": 8000, + "output_dir": "/app/checkpoints/1/game-v1.0.0", + "pad_to_sequence_len": true, + "pretrain_multipack_attn": true, + "profiler_steps_start": 0, + "qlora_sharded_model_loading": false, + "ray_num_workers": 1, + "resources_per_worker": { + "GPU": 1 + }, + "rl": "grpo", + "sample_packing": false, + "sample_packing_bin_size": 200, + "sample_packing_group_size": 100000, + "save_only_model": false, + "save_safetensors": true, + "save_steps": 20, + "save_total_limit": 1, + "sequence_len": 8192, + "shuffle_before_merging_datasets": false, + "shuffle_merged_datasets": true, + "skip_prepare_dataset": false, + "special_tokens": { + "bos_token": "<|im_end|>" + }, + "streaming_multipack_buffer_size": 10000, + "strict": false, + "tensor_parallel_size": 1, + "tf32": false, + "tiled_mlp_use_original_mlp": true, + "tokenizer_config": "/cache/models/Qwen--Qwen2.5-3B-Instruct", + "tokenizer_save_jinja_files": true, + "tokenizer_type": "AutoTokenizer", + "torch_dtype": "torch.bfloat16", + "train_on_inputs": false, + "trl": { + "beta": 0.001, + "log_completions": false, + "mask_truncated_completions": false, + "max_completion_length": 256, + "num_generations": 8, + "ref_model_mixup_alpha": 0.9, + "ref_model_sync_steps": 64, + "reward_funcs": [ + "affine_game.rollout_reward_func" + ], + "reward_weights": [ + 1.0 + ], + "rollout_func": "affine_game.rollout_first_prompt_and_completion", + "scale_rewards": true, + "sync_ref_model": false, + "temperature": 0.7, + "use_vllm": true, + "vllm_enable_sleep_mode": false, + "vllm_mode": "colocate", + "vllm_server_host": "0.0.0.0", + "vllm_server_port": 8000 + }, + "trust_remote_code": true, + "type_of_model": "AutoModelForCausalLM", + "use_mlflow": true, + "use_otel_metrics": false, + "use_ray": false, + "use_wandb": true, + "val_set_size": 0.0, + "vllm": { + "device": "auto", + "dtype": "auto", + "enable_prefix_caching": true, + "gpu_memory_utilization": 0.3, + "host": "0.0.0.0", + "max_model_len": 8192, + "port": 8000, + "tensor_parallel_size": 1 + }, + "wandb_mode": "online", + "wandb_name": "1_game-v1.0.0", + "wandb_project": "Gradients-Rollout", + "warmup_steps": 10, + "weight_decay": 0.0, + "world_size": 2 +} +[2026-01-28 17:33:01,190] [WARNING] [axolotl.cli.checks.check_user_token:46] [PID:251] Error verifying HuggingFace token. Remember to log in using `hf auth login` and get your access token from https://huggingface.co/settings/tokens if you want to use gated models or datasets. +[2026-01-28 17:33:01,421] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:251] EOS: 151645 / <|im_end|> +[2026-01-28 17:33:01,422] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:251] BOS: 151645 / <|im_end|> +[2026-01-28 17:33:01,422] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:251] PAD: 151643 / <|endoftext|> +[2026-01-28 17:33:01,422] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:251] UNK: None / None +[2026-01-28 17:33:08,215] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:481] [PID:251] Unable to find prepared dataset in last_run_prepared/99e92ccac4c32db35c2b52b88285f0d2 +[2026-01-28 17:33:08,600] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:251] EOS: 151645 / <|im_end|> +[2026-01-28 17:33:08,600] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:251] BOS: 151645 / <|im_end|> +[2026-01-28 17:33:08,600] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:251] PAD: 151643 / <|endoftext|> +[2026-01-28 17:33:08,600] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:251] UNK: None / None + Saving the dataset (0/3 shards): 0%| | 0/1000 [00:00 +[2026-01-28 17:33:09,170] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:251] BOS: 151645 / <|im_end|> +[2026-01-28 17:33:09,170] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:251] PAD: 151643 / <|endoftext|> +[2026-01-28 17:33:09,170] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:251] UNK: None / None +[2026-01-28 17:33:09,170] [DEBUG] [axolotl.train.setup_model_and_tokenizer:82] [PID:251] Loading model +[2026-01-28 17:33:09,179] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:251] Patched Trainer.evaluation_loop with nanmean loss calculation +[2026-01-28 17:33:09,181] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:251] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation + Loading checkpoint shards: 0%| | 0/2 [00:00:246: FutureWarning: The `max_prompt_length` argument is deprecated and will be removed in version 0.28.0. You should instead filter your dataset before training to ensure that prompts do not exceed your desired length. + +[2026-01-28 17:33:19,837] [WARNING] [py.warnings._showwarnmsg:110] [PID:251] /workspace/axolotl/src/axolotl/core/trainers/mixins/optimizer.py:209: UserWarning: You are importing from 'rollout_func', which is an experimental feature. This API may change or be removed at any time without prior notice. Silence this warning by setting environment variable TRL_EXPERIMENTAL_SILENCE=1. + super().__init__(*args, **kwargs) + + Loading safetensors checkpoint shards: 0% Completed | 0/2 [00:00