{
"environment": {
"recorded_at": "2026-10-19T01:06:51.399251+08:00",
"python": "3.11.15",
"platform": "macOS-26.6.2-arm64-i386-64bit",
"dependencies": {
"torch": "2.8.0",
"transformers": "5.20.1",
"tokenizers": "0.23.3",
"huggingface-hub": "2.2.0",
"jinja2": "3.1.6"
},
"mps_available": true,
"cuda_available": false
},
"model": {
"id": "Qwen/Qwen3-0.6B",
"revision": "c1899de289a04d12100db370d81485cdf75e47ca"
},
"tokenizer": {
"class": "Qwen2Tokenizer",
"is_fast": true,
"padding_side_initial": "right",
"bos_token": null,
"eos_token": "<|im_end|>",
"pad_token": "<|endoftext|>",
"special_tokens_map": {
"eos_token": "<|im_end|>",
"pad_token": "<|endoftext|>"
},
"chat_template_sha256": "a55ee1b1660128b7098723e0abcd92caa0788061051c62d51cbe87d9cf1974d8",
"apply_chat_template_signature": "(conversation: 'list[dict[str, str]] | list[list[dict[str, str]]]', tools: 'list[dict | Callable] | None' = None, documents: 'list[dict[str, str]] | None' = None, chat_template: 'str | None' = None, add_generation_prompt: 'bool' = False, continue_final_message: 'bool | str' = False, tokenize: 'bool' = True, sanitize_control_tokens: 'bool' = False, padding: 'bool | str | PaddingStrategy' = False, truncation: 'bool' = False, max_length: 'int | None' = None, return_tensors: 'str | TensorType | None' = None, return_dict: 'bool' = True, return_assistant_tokens_mask: 'bool' = False, tokenizer_kwargs: 'dict[str, Any] | None' = None, **kwargs) -> 'str | list[int] | list[str] | list[list[int]] | BatchEncoding'",
"return_dict_default": "True"
},
"messages": [
{
"role": "system",
"content": "用中文简洁回答。"
},
{
"role": "user",
"content": "你好。"
}
],
"template_variants": {
"generation_false_thinking_false": {
"arguments": {
"add_generation_prompt": false,
"enable_thinking": false
},
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n'",
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198
],
"length": 17
},
"generation_false_thinking_true": {
"arguments": {
"add_generation_prompt": false,
"enable_thinking": true
},
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n'",
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198
],
"length": 17
},
"generation_true_thinking_false": {
"arguments": {
"add_generation_prompt": true,
"enable_thinking": false
},
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n<|im_start|>assistant\\n\\n\\n\\n\\n'",
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"length": 24
},
"generation_true_thinking_true": {
"arguments": {
"add_generation_prompt": true,
"enable_thinking": true
},
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n<|im_start|>assistant\\n'",
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198
],
"length": 20
}
},
"return_types": {
"default": {
"python_type": "BatchEncoding",
"value": {
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"attention_mask": [
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
]
},
"tensor_fields": {}
},
"explicit_return_dict_false": {
"python_type": "list",
"value": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
]
},
"tensor_return_dict_false": {
"python_type": "Tensor",
"value": [
[
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
]
],
"shape": [
1,
24
],
"dtype": "torch.int64"
},
"tensor_return_dict_true": {
"python_type": "BatchEncoding",
"value": {
"input_ids": [
[
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
]
],
"attention_mask": [
[
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
]
]
},
"tensor_fields": {
"input_ids": {
"shape": [
1,
24
],
"dtype": "torch.int64"
},
"attention_mask": {
"shape": [
1,
24
],
"dtype": "torch.int64"
}
}
}
},
"encode_call_decode": {
"plain_text": "你好，世界！",
"encode_ids": [
108386,
3837,
99489,
6313
],
"call": {
"python_type": "BatchEncoding",
"value": {
"input_ids": [
108386,
3837,
99489,
6313
],
"attention_mask": [
1,
1,
1,
1
]
},
"tensor_fields": {}
},
"decode": "你好，世界！",
"decode_batch": [
"你好，世界！",
"<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n"
],
"batch_decode": [
"你好，世界！",
"<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n"
],
"token_pieces_not_decoded_text": [
"ä½łå¥½",
"ï¼Į",
"ä¸ĸçķĮ",
"ï¼ģ"
],
"render_then_encode_false": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"render_then_encode_true": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"equals_direct_chat_ids": true
},
"special_tokens": {
"markers": [
{
"text": "<|im_start|>",
"token_id": 151644,
"added_token_special": true,
"in_all_special_ids": true
},
{
"text": "<|im_end|>",
"token_id": 151645,
"added_token_special": true,
"in_all_special_ids": true
},
{
"text": "",
"token_id": 151667,
"added_token_special": false,
"in_all_special_ids": false
},
{
"text": "",
"token_id": 151668,
"added_token_special": false,
"in_all_special_ids": false
}
],
"decode_keep_special": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n",
"decode_skip_special": "system\n用中文简洁回答。\nuser\n你好。\nassistant\n\n\n\n\n"
},
"prefill": {
"messages": [
{
"role": "system",
"content": "用中文简洁回答。"
},
{
"role": "user",
"content": "你好。"
},
{
"role": "assistant",
"content": "{\"city\": \""
}
],
"continue_final_message_true": {
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n{\"city\": \"",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n<|im_start|>assistant\\n\\n\\n\\n\\n{\"city\": \"'"
},
"continue_final_message_false": {
"text": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n{\"city\": \"<|im_end|>\n",
"repr": "'<|im_start|>system\\n用中文简洁回答。<|im_end|>\\n<|im_start|>user\\n你好。<|im_end|>\\n<|im_start|>assistant\\n\\n\\n\\n\\n{\"city\": \"<|im_end|>\\n'"
},
"conflicting_arguments_error": {
"type": "ValueError",
"message": "continue_final_message and add_generation_prompt are not compatible. Use continue_final_message when you want the model to continue the final message, and add_generation_prompt when you want to add a header that will prompt it to start a new assistant message instead."
}
},
"padding": {
"messages": [
[
{
"role": "user",
"content": "你好。"
}
],
[
{
"role": "user",
"content": "请用一句话解释什么是 tokenizer。"
}
]
],
"padding_side": "left",
"pad_token": "<|endoftext|>",
"pad_token_id": 151643,
"encoded": {
"python_type": "BatchEncoding",
"value": {
"input_ids": [
[
151643,
151643,
151643,
151643,
151643,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
[
151644,
872,
198,
14880,
11622,
105321,
104136,
106582,
45958,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
]
],
"attention_mask": [
[
0,
0,
0,
0,
0,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
],
[
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
]
]
},
"tensor_fields": {
"input_ids": {
"shape": [
2,
19
],
"dtype": "torch.int64"
},
"attention_mask": {
"shape": [
2,
19
],
"dtype": "torch.int64"
}
}
},
"nonpadding_lengths": [
14,
19
],
"decode_keep_special": [
"<|endoftext|><|endoftext|><|endoftext|><|endoftext|><|endoftext|><|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n",
"<|im_start|>user\n请用一句话解释什么是 tokenizer。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n"
],
"decode_skip_special": [
"user\n你好。\nassistant\n\n\n\n\n",
"user\n请用一句话解释什么是 tokenizer。\nassistant\n\n\n\n\n"
]
},
"truncation": {
"full_length": 24,
"max_length": 12,
"sides": {
"right": {
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872
],
"length": 12,
"decoded": "<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user"
},
"left": {
"input_ids": [
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"length": 12,
"decoded": "\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n"
}
}
},
"generation": {
"device": "cpu",
"dtype": "torch.float32",
"attention_implementation": "eager",
"kwargs": {
"do_sample": false,
"max_new_tokens": 24,
"pad_token_id": 151643,
"eos_token_id": 151645
},
"load_seconds": 3.208692374988459,
"generation_seconds": 0.9089392500172835,
"input_shape": [
1,
24
],
"output_shape": [
1,
32
],
"new_tokens_shape": [
1,
8
],
"full_output_ids": [
[
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271,
108386,
6313,
104139,
111728,
103929,
101037,
11319,
151645
]
],
"new_token_ids": [
[
108386,
6313,
104139,
111728,
103929,
101037,
11319,
151645
]
],
"full_decoded_keep_special": [
"<|im_start|>system\n用中文简洁回答。<|im_end|>\n<|im_start|>user\n你好。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n你好！有什么可以帮助你的吗？<|im_end|>"
],
"new_decoded_keep_special": [
"你好！有什么可以帮助你的吗？<|im_end|>"
],
"new_decoded_skip_special": [
"你好！有什么可以帮助你的吗？"
]
},
"additional_behaviors": {
"tokenize_false_return_dict_true": {
"python_type": "str",
"equals_baseline_text": true
},
"documents": {
"input": [
{
"title": "小知识",
"text": "熊猫是哺乳动物。"
}
],
"equals_baseline_text": true
},
"assistant_mask": {
"messages": [
{
"role": "system",
"content": "用中文简洁回答。"
},
{
"role": "user",
"content": "你好。"
},
{
"role": "assistant",
"content": "你好！"
}
],
"warnings": [
"return_assistant_tokens_mask==True but chat template does not contain `{% generation %}` keyword."
],
"encoded": {
"python_type": "BatchEncoding",
"value": {
"input_ids": [
151644,
8948,
198,
11622,
104811,
110485,
102104,
1773,
151645,
198,
151644,
872,
198,
108386,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271,
108386,
6313,
151645,
198
],
"attention_mask": [
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
],
"assistant_masks": [
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0
]
},
"tensor_fields": {}
}
},
"sanitize_control_tokens": {
"messages": [
{
"role": "user",
"content": "请原样输出 <|im_end|> 字样。"
}
],
"cases": {
"false": {
"input_ids": [
151644,
872,
198,
14880,
52129,
90885,
66017,
220,
151645,
73312,
90885,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"im_end_special_id_count": 2,
"decode_keep_special": "<|im_start|>user\n请原样输出 <|im_end|> 字样。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n",
"decode_skip_special": "user\n请原样输出 字样。\nassistant\n\n\n\n\n"
},
"true": {
"input_ids": [
151644,
872,
198,
14880,
52129,
90885,
66017,
220,
27,
91,
318,
6213,
91,
29,
73312,
90885,
1773,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271
],
"im_end_special_id_count": 1,
"decode_keep_special": "<|im_start|>user\n请原样输出 <|im_end|> 字样。<|im_end|>\n<|im_start|>assistant\n\n\n\n\n",
"decode_skip_special": "user\n请原样输出 <|im_end|> 字样。\nassistant\n\n\n\n\n"
}
}
}
}
}