[2026-02-04 03:22:54,042] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:23602] bf16 support detected, enabling for this configuration. [2026-02-04 03:22:54,243] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:23602] baseline 0.000GB () [2026-02-04 03:22:54,245] [INFO] [axolotl.cli.config.load_cfg:259] [PID:23602] config: { "activation_offloading": false, "axolotl_config_path": "T2J_SFT_4B.yaml", "base_model": "Qwen/Qwen2.5-1.5B-Instruct", "base_model_config": "Qwen/Qwen2.5-1.5B-Instruct", "batch_size": 256, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_90", "fp8": true, "n_gpu": 1, "n_node": 1 }, "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_num_proc": 192, "dataset_prepared_path": "preprocess", "datasets": [ { "chat_template": "tokenizer_default", "message_property_mappings": { "content": "content", "role": "role" }, "path": "train_T2J.jsonl", "trust_remote_code": false, "type": "chat_template" } ], "ddp": false, "device": "cuda:0", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "eaft_alpha": 1.0, "eaft_k": 20, "env_capabilities": { "torch_version": "2.9.1" }, "eval_batch_size": 2, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": false, "eval_table_size": 0, "evals_per_epoch": 0, "experimental_skip_move_to_device": true, "flash_attention": true, "fp16": false, "gradient_accumulation_steps": 128, "hub_model_id": "amphora/FC-T2J-SFT-1_5B", "include_tkps": true, "learning_rate": 2e-05, "liger_fused_linear_cross_entropy": true, "liger_glu_activation": true, "liger_rms_norm": true, "liger_rope": true, "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_dropout": 0.0, "loraplus_lr_embedding": 1e-06, "lr_scheduler": "cosine", "mean_resizing_embeddings": false, "micro_batch_size": 2, "model_config_type": "qwen2", "num_epochs": 3.0, "optimizer": "adamw_torch_fused", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "./outputs", "pad_to_sequence_len": false, "plugins": [ "axolotl.integrations.liger.LigerPlugin" ], "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "ray_num_workers": 1, "resources_per_worker": { "GPU": 1 }, "sample_packing": false, "sample_packing_bin_size": 200, "sample_packing_group_size": 100000, "save_only_model": false, "save_safetensors": true, "save_steps": 0.3333333333333333, "saves_per_epoch": 1, "sequence_len": 16384, "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/Qwen2.5-1.5B-Instruct", "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": 0.01, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "wandb_project": "FC-T2J", "warmup_ratio": 0.05, "weight_decay": 0.01, "world_size": 1 } [2026-02-04 03:22:55,706] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:285] [PID:23602] EOS: 151645 / <|im_end|> [2026-02-04 03:22:55,707] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:23602] BOS: None / None [2026-02-04 03:22:55,707] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:23602] PAD: 151643 / <|endoftext|> [2026-02-04 03:22:55,707] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:23602] UNK: None / None [2026-02-04 03:22:55,709] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:475] [PID:23602] Loading prepared dataset from disk at preprocess/d897fb18203056becd1259947bafc7e4... Loading dataset from disk: 0%| | 0/192 [00:00 [2026-02-04 03:23:11,868] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:286] [PID:23602] BOS: None / None [2026-02-04 03:23:11,868] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:287] [PID:23602] PAD: 151643 / <|endoftext|> [2026-02-04 03:23:11,869] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:288] [PID:23602] UNK: None / None [2026-02-04 03:23:11,869] [DEBUG] [axolotl.train.setup_model_and_tokenizer:82] [PID:23602] Loading model [2026-02-04 03:23:12,011] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:23602] Patched Trainer.evaluation_loop with nanmean loss calculation [2026-02-04 03:23:12,016] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:23602] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2026-02-04 03:23:12,201] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:98] [PID:23602] Applying LIGER to qwen2 with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': True, 'rms_norm': True, 'swiglu': True} Loading weights: 0%| | 0/338 [00:00:39] [PID:23602] Skipping import of cpp extensions due to incompatible torch version 2.9.1+cu128 for torchao version 0.13.0 [2026-02-04 03:23:22,826] [INFO] [axolotl.train.save_initial_configs:406] [PID:23602] Pre-saving tokenizer to ./outputs... [2026-02-04 03:23:23,029] [INFO] [axolotl.train.save_initial_configs:411] [PID:23602] Pre-saving model config to ./outputs... [2026-02-04 03:23:23,032] [INFO] [axolotl.train.execute_training:207] [PID:23602] Starting trainer... wandb: [wandb.login()] Loaded credentials for https://api.wandb.ai from /root/.netrc. wandb: Currently logged in as: guijinson to https://api.wandb.ai. Use `wandb login --relogin` to force relogin wandb: ⢿ Waiting for wandb.init()... m wandb: ⣻ setting up run l5xo86c3 (0.1s) m wandb: ⣽ setting up run l5xo86c3 (0.1s) m wandb: Tracking run with wandb version 0.24.1 wandb: Run data is saved locally in /workspace/axolotl/wandb/run-20260204_032324-l5xo86c3 wandb: Run `wandb offline` to turn off syncing. wandb: Syncing run fallen-puddle-3 wandb: ⭐️ View project at https://wandb.ai/guijinson/FC-T2J wandb: 🚀 View run at https://wandb.ai/guijinson/FC-T2J/runs/l5xo86c3 wandb: Detected [huggingface_hub.inference] in use. wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script. wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/ wandb: WARNING Saving files without folders. If you want to preserve subdirectories pass base_path to wandb.save, i.e. wandb.save("/mnt/folder/file.h5", base_path="/mnt") wandb: WARNING Symlinked 1 file into the W&B run directory; call wandb.save again to sync new files. [2026-02-04 03:23:27,200] [INFO] [axolotl.utils.callbacks.on_train_begin:757] [PID:23602] The Axolotl config has been saved to the WandB run under files. 0%| | 0/1751 [00:00