[2025-10-12 01:51:15,011] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:4367] bf16 support detected, enabling for this configuration. config.json: 0%| | 0.00/766 [00:00 [2025-10-12 01:51:17,401] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:4367] BOS: None / None [2025-10-12 01:51:17,401] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:4367] PAD: 151665 / <|PAD_TOKEN|> [2025-10-12 01:51:17,401] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:4367] UNK: None / None [2025-10-12 01:51:17,402] [INFO] [axolotl.utils.data.shared.load_preprocessed_dataset:476] [PID:4367] Unable to find prepared dataset in last_run_prepared/912502a23fd453b552daefcb3e93cedf [2025-10-12 01:51:17,402] [INFO] [axolotl.utils.data.sft._load_raw_datasets:320] [PID:4367] Loading raw datasets... [2025-10-12 01:51:17,402] [WARNING] [axolotl.utils.data.sft._load_raw_datasets:322] [PID:4367] Processing datasets during training can lead to VRAM instability. Please pre-process your dataset using `axolotl preprocess path/to/config.yml`. [2025-10-12 01:51:17,496] [INFO] [axolotl.utils.data.wrappers.get_dataset_wrapper:87] [PID:4367] Loading dataset: /workspace/axolotl/sgtr-em-dev/data/finetuning/aesthetic_preferences_unpopular.jsonl with base_type: chat_template and prompt_style: None [2025-10-12 01:51:17,498] [INFO] [axolotl.prompt_strategies.chat_template.__call__:969] [PID:4367] Using chat template: --- {%- if tools %} {{- '<|im_start|>system\n' }} {%- if messages[0]['role'] == 'system' %} {{- messages[0]['content'] }} {%- else %} {{- 'You are Qwen, created by Alibaba Cloud. 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 Qwen, created by Alibaba Cloud. 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=12): 0%| | 0/5000 [00:002048) (num_proc=12): 0%| | 0/5000 [00:002048) (num_proc=12): 8%|████▊ | 417/5000 [00:00<00:05, 866.84 examples/s] Dropping Long Sequences (>2048) (num_proc=12): 100%|████████████████████████████████████████████████████████| 5000/5000 [00:00<00:00, 7217.37 examples/s] Saving the dataset (0/12 shards): 0%| | 0/5000 [00:00 [2025-10-12 01:51:22,325] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:279] [PID:4367] BOS: None / None [2025-10-12 01:51:22,325] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:4367] PAD: 151665 / <|PAD_TOKEN|> [2025-10-12 01:51:22,325] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:4367] UNK: None / None [2025-10-12 01:51:22,325] [DEBUG] [axolotl.train.setup_model_and_tokenizer:74] [PID:4367] Loading model [2025-10-12 01:51:22,409] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:4367] Patched Trainer.evaluation_loop with nanmean loss calculation [2025-10-12 01:51:22,410] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:4367] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2025-10-12 01:51:23,491] [INFO] [axolotl.monkeypatch.lora_kernels.patch_self_attn_lora:232] [PID:4367] Patched attention class with LoRA optims: Qwen2Attention model.safetensors.index.json: 0.00B [00:00, ?B/s] model.safetensors.index.json: 63.2kB [00:00, 57.2MB/s] model-00001-of-00014.safetensors: 0%| | 0.00/4.89G [00:00