[2026-03-29 17:06:48,176] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:1870] bf16 support detected, enabling for this configuration. [2026-03-29 17:06:48,413] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:1870] baseline 0.000GB () [2026-03-29 17:06:48,417] [INFO] [axolotl.cli.config.load_cfg:248] [PID:1870] config: { "activation_offloading": false, "adapter": "lora", "axolotl_config_path": "config.yaml", "base_model": "dphn/Dolphin3.0-Llama3.2-3B", "base_model_config": "dphn/Dolphin3.0-Llama3.2-3B", "batch_size": 16, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_90", "fp8": false, "n_gpu": 1, "n_node": 1 }, "chat_template": "tokenizer_default", "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_processes": 26, "datasets": [ { "chat_template": "tokenizer_default", "data_files": [ "train.jsonl" ], "field_messages": "messages", "message_property_mappings": { "content": "content", "role": "role" }, "path": "karan11/defender-judge-fine-tune", "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": 4, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": false, "eval_steps": 50, "eval_table_size": 0, "experimental_skip_move_to_device": true, "flash_attention": true, "fp16": false, "gradient_accumulation_steps": 4, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": true }, "include_tkps": true, "is_llama_derived_model": true, "learning_rate": 0.00015, "lisa_layers_attribute": "model.layers", "load_best_model_at_end": false, "load_in_4bit": false, "load_in_8bit": true, "local_rank": 0, "logging_steps": 5, "lora_alpha": 64, "lora_dropout": 0.1, "lora_r": 32, "lora_target_modules": [ "q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj" ], "loraplus_lr_embedding": 1e-06, "lr_scheduler": "cosine", "max_grad_norm": 1.0, "mean_resizing_embeddings": false, "micro_batch_size": 4, "model_config_type": "llama", "neftune_noise_alpha": 5.0, "num_epochs": 3.0, "optimizer": "adamw_bnb_8bit", "output_dir": "./outputs/defender-judge", "pad_to_sequence_len": true, "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "ray_num_workers": 1, "remove_unused_columns": false, "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": 50, "sequence_len": 8192, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": false, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "dphn/Dolphin3.0-Llama3.2-3B", "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_ray": false, "val_set_size": 0.0, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "warmup_ratio": 0.1, "weight_decay": 0.05, "world_size": 1 } [2026-03-29 17:06:48,431] [WARNING] [axolotl.cli.checks.check_user_token:46] [PID:1870] Error verifying HuggingFace token. Remember to log in using `huggingface-cli login` and get your access token from https://huggingface.co/settings/tokens if you want to use gated models or datasets. [2026-03-29 17:06:49,360] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:278] [PID:1870] EOS: 128256 / <|im_end|> [2026-03-29 17:06:49,362] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:1870] BOS: 128000 / <|begin_of_text|> [2026-03-29 17:06:49,364] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1870] PAD: 128001 / <|end_of_text|> [2026-03-29 17:06:49,365] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1870] UNK: None / None [2026-03-29 17:06:49,368] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:476] [PID:1870] Unable to find prepared dataset in last_run_prepared/b72df10b48d5f00efef65cbe761c72f7 [2026-03-29 17:06:49,370] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:1870] Loading raw datasets... [2026-03-29 17:06:49,372] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:1870] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`. [2026-03-29 17:06:50,431] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:1870] Loading dataset: karan11/defender-judge-fine-tune with base_type: chat_template and prompt_style: None [2026-03-29 17:06:50,440] [INFO] [axolotl.prompt_strategies.chat_template.__call__:969] [PID:1870] Using chat template: --- {%- if tools %} {{- '<|im_start|>system\n' }} {%- if messages[0]['role'] == 'system' %} {{- messages[0]['content'] }} {%- else %} {{- 'You are Dolphin, created by Eric Hartford. You are a helpful assistant.' }} {%- endif %} {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }} {%- for tool in tools %} {{- "\n" }} {{- tool | tojson }} {%- endfor %} {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }} {%- else %} {%- if messages[0]['role'] == 'system' %} {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }} {%- else %} {{- '<|im_start|>system\nYou are Dolphin, created by Eric Hartford. You are a helpful assistant.<|im_end|>\n' }} {%- endif %} {%- endif %} {%- for message in messages %} {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %} {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }} {%- elif message.role == "assistant" %} {{- '<|im_start|>' + message.role }} {%- if message.content %} {{- '\n' + message.content }} {%- endif %} {%- for tool_call in message.tool_calls %} {%- if tool_call.function is defined %} {%- set tool_call = tool_call.function %} {%- endif %} {{- '\n\n{"name": "' }} {{- tool_call.name }} {{- '", "arguments": ' }} {{- tool_call.arguments | tojson }} {{- '}\n' }} {%- endfor %} {{- '<|im_end|>\n' }} {%- elif message.role == "tool" %} {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %} {{- '<|im_start|>user' }} {%- endif %} {{- '\n\n' }} {{- message.content }} {{- '\n' }} {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %} {{- '<|im_end|>\n' }} {%- endif %} {%- endif %} {%- endfor %} {%- if add_generation_prompt %} {{- '<|im_start|>assistant\n' }} {%- endif %} --- Tokenizing Prompts (num_proc=26): 0%| | 0/2700 [00:008192) (num_proc=26): 0%| | 0/2700 [00:008192) (num_proc=26): 4%|███▌ | 104/2700 [00:01<00:26, 99.21 examples/s] Dropping Long Sequences (>8192) (num_proc=26): 23%|████████████████████▊ | 624/2700 [00:01<00:02, 701.67 examples/s] Dropping Long Sequences (>8192) (num_proc=26): 39%|█████████████████████████████████▉ | 1040/2700 [00:01<00:01, 1190.54 examples/s] Dropping Long Sequences (>8192) (num_proc=26): 100%|████████████████████████████████████████████████████████████████████████████████████████| 2700/2700 [00:01<00:00, 1481.80 examples/s] Drop Samples with Zero Trainable Tokens (num_proc=26): 0%| | 0/2700 [00:00 [2026-03-29 17:07:06,639] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:1870] BOS: 128000 / <|begin_of_text|> [2026-03-29 17:07:06,641] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1870] PAD: 128001 / <|end_of_text|> [2026-03-29 17:07:06,642] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1870] UNK: None / None [2026-03-29 17:07:06,644] [DEBUG] [axolotl.train.setup_model_and_tokenizer:74] [PID:1870] Loading model [2026-03-29 17:07:06,747] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:1870] Patched Trainer.evaluation_loop with nanmean loss calculation [2026-03-29 17:07:06,752] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:1870] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2026-03-29 17:07:06,755] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:301] [PID:1870] Applying multipack dataloader patch for sample packing... model.safetensors.index.json: 0.00B [00:00, ?B/s] model.safetensors.index.json: 20.9kB [00:00, 78.6MB/s] model-00001-of-00002.safetensors: 0%| | 0.00/4.97G [00:00