diff --git "a/debug.log" "b/debug.log" new file mode 100644--- /dev/null +++ "b/debug.log" @@ -0,0 +1,2223 @@ +[2025-10-31 00:01:31,161] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:7760] bf16 support detected, enabling for this configuration. +[2025-10-31 00:01:31,543] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:7760] baseline 0.000GB () +[2025-10-31 00:01:31,544] [INFO] [axolotl.cli.config.load_cfg:248] [PID:7760] config: +{ + "activation_offloading": false, + "axolotl_config_path": "my-config.yml", + "base_model": "Qwen/Qwen3-4B", + "base_model_config": "Qwen/Qwen3-4B", + "batch_size": 4, + "bf16": true, + "capabilities": { + "bf16": true, + "compute_capability": "sm_90", + "fp8": false, + "n_gpu": 1, + "n_node": 1 + }, + "chat_template": "chatml", + "context_parallel_size": 1, + "dataloader_num_workers": 1, + "dataloader_pin_memory": true, + "dataloader_prefetch_factor": 256, + "dataset_num_proc": 16, + "dataset_prepared_path": "prepared_data2", + "datasets": [ + { + "chat_template": "tokenizer_default", + "field_messages": "messages", + "message_property_mappings": { + "content": "content", + "role": "role" + }, + "path": "joeyzero/OpenThought-144k-Backfill-0.2", + "trust_remote_code": false, + "type": "chat_template" + }, + { + "chat_template": "tokenizer_default", + "field_messages": "messages", + "message_property_mappings": { + "content": "content", + "role": "role" + }, + "path": "joeyzero/dolphin-r1-backfill-0.0.2", + "trust_remote_code": false, + "type": "chat_template" + } + ], + "ddp": false, + "device": "cuda:0", + "dion_rank_fraction": 1.0, + "dion_rank_multiple_of": 1, + "env_capabilities": { + "torch_version": "2.7.1" + }, + "eval_batch_size": 2, + "eval_causal_lm_metrics": [ + "sacrebleu", + "comet", + "ter", + "chrf" + ], + "eval_max_new_tokens": 128, + "eval_sample_packing": true, + "eval_table_size": 0, + "experimental_skip_move_to_device": true, + "flash_attention": true, + "fp16": false, + "gradient_accumulation_steps": 2, + "gradient_checkpointing": false, + "hf_use_auth_token": true, + "hub_model_id": "joeyzero/Qwen3-4B-Reasoning-Backfill-V0.1", + "include_tkps": true, + "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, + "lora_dropout": 0.0, + "loraplus_lr_embedding": 1e-06, + "lr_scheduler": "cosine", + "max_grad_norm": 1.0, + "mean_resizing_embeddings": false, + "micro_batch_size": 2, + "model_config_type": "qwen3", + "num_epochs": 4.0, + "optimizer": "adamw_bnb_8bit", + "otel_metrics_host": "localhost", + "otel_metrics_port": 8000, + "output_dir": "./thinking-backfill-0.1.17", + "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 + }, + "sample_packing": true, + "sample_packing_bin_size": 200, + "sample_packing_group_size": 100000, + "save_only_model": false, + "save_safetensors": true, + "save_steps": 0.5, + "sequence_len": 1024, + "shuffle_before_merging_datasets": false, + "shuffle_merged_datasets": true, + "skip_prepare_dataset": false, + "streaming_multipack_buffer_size": 10000, + "strict": false, + "tensor_parallel_size": 1, + "tf32": false, + "tiled_mlp_use_original_mlp": true, + "tokenizer_config": "Qwen/Qwen3-4B", + "tokenizer_save_jinja_files": true, + "torch_dtype": "torch.bfloat16", + "train_on_inputs": false, + "trl": { + "log_completions": false, + "mask_truncated_completions": false, + "ref_model_mixup_alpha": 0.9, + "ref_model_sync_steps": 64, + "scale_rewards": true, + "sync_ref_model": false, + "use_vllm": false, + "vllm_server_host": "0.0.0.0", + "vllm_server_port": 8000 + }, + "use_otel_metrics": false, + "use_ray": false, + "use_wandb": true, + "val_set_size": 0.0, + "vllm": { + "device": "auto", + "dtype": "auto", + "gpu_memory_utilization": 0.9, + "host": "0.0.0.0", + "port": 8000 + }, + "wandb_name": "reasoning-backfill-attempt-04", + "wandb_project": "reasoning-backfill", + "warmup_steps": 40, + "weight_decay": 0.02, + "world_size": 1 +} +[2025-10-31 00:01:32,578] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:7760] EOS: 151645 / <|im_end|> +[2025-10-31 00:01:32,578] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:7760] BOS: None / None +[2025-10-31 00:01:32,578] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:7760] PAD: 151643 / <|endoftext|> +[2025-10-31 00:01:32,578] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:7760] UNK: None / None +[2025-10-31 00:01:32,579] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:475] [PID:7760] Loading prepared dataset from disk at prepared_data2/00beaf69a2639c1250888269a8624992... +[2025-10-31 00:01:32,787] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:406] [PID:7760] total_num_tokens: 37_608_602 +[2025-10-31 00:01:33,446] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:424] [PID:7760] `total_supervised_tokens: 17_899_199` +[2025-10-31 00:01:34,255] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:7760] Using single process for pack_parallel, running sequentially. +[2025-10-31 00:01:35,344] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:7760] Using single process for pack_parallel, running sequentially. +[2025-10-31 00:01:35,864] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:7760] generate_batches time: 0.5274832248687744 +[2025-10-31 00:01:35,871] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:7760] Using single process for pack_parallel, running sequentially. +[2025-10-31 00:01:36,274] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:7760] generate_batches time: 0.40900588035583496 +[2025-10-31 00:01:36,280] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:7760] Using single process for pack_parallel, running sequentially. +[2025-10-31 00:01:36,729] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:7760] generate_batches time: 0.4532792568206787 +[2025-10-31 00:01:36,735] [DEBUG] [axolotl.utils.samplers.multipack.pack_parallel:177] [PID:7760] Using single process for pack_parallel, running sequentially. +[2025-10-31 00:01:37,134] [DEBUG] [axolotl.utils.samplers.multipack.__len__:462] [PID:7760] generate_batches time: 0.40355348587036133 +[2025-10-31 00:01:37,176] [INFO] [axolotl.utils.samplers.multipack.calc_min_len:438] [PID:7760] gather_len_batches: [20040] +[2025-10-31 00:01:37,176] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:483] [PID:7760] data_loader_len: 10020 +[2025-10-31 00:01:37,176] [INFO] [axolotl.utils.trainer.calc_sample_packing_eff_est:499] [PID:7760] sample_packing_eff_est across ranks: [0.9165747539462191] +[2025-10-31 00:01:37,176] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:511] [PID:7760] sample_packing_eff_est: 0.92 +[2025-10-31 00:01:37,176] [DEBUG] [axolotl.utils.trainer.calculate_total_num_steps:522] [PID:7760] total_num_steps: 40080 +[2025-10-31 00:01:37,180] [INFO] [axolotl.utils.data.sft._prepare_standard_dataset:121] [PID:7760] Maximum number of steps set at 40080 +[2025-10-31 00:01:37,218] [DEBUG] [axolotl.train.setup_model_and_tokenizer:65] [PID:7760] Loading tokenizer... Qwen/Qwen3-4B +[2025-10-31 00:01:37,972] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:7760] EOS: 151645 / <|im_end|> +[2025-10-31 00:01:37,972] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:7760] BOS: None / None +[2025-10-31 00:01:37,972] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:7760] PAD: 151643 / <|endoftext|> +[2025-10-31 00:01:37,972] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:7760] UNK: None / None +[2025-10-31 00:01:37,972] [DEBUG] [axolotl.train.setup_model_and_tokenizer:74] [PID:7760] Loading model +[2025-10-31 00:01:38,153] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:7760] Patched Trainer.evaluation_loop with nanmean loss calculation +[2025-10-31 00:01:38,156] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:7760] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation +[2025-10-31 00:01:38,156] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:301] [PID:7760] Applying multipack dataloader patch for sample packing... + model.safetensors.index.json: 0.00B [00:00, ?B/s] model.safetensors.index.json: 32.8kB [00:00, 40.2MB/s] + model-00001-of-00003.safetensors: 0%| | 0.00/3.96G [00:00