python from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Qwen/Qwen2.5-7B-Instruct"

加载 tokenizer 和模型（使用 bfloat16 节省显存）

tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto" )

使用 ChatML 格式构造对话

messages = [ {"role": "system", "content": "你是一个有用的助手。"}, {"role": "user", "content": "请用三种语言介绍 Qwen 模型系列。"} ]

应用聊天模板

text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True )

生成回复

inputs = tokenizer(text, return_tensors="pt").to(model.device) outputs = model.generate(**inputs, max_new_tokens=512) response = tokenizer.decode(outputs[0][len(inputs.input_ids[0]):], skip_special_tokens=True) print(response)

如果显存不足，可尝试使用 4-bit 量化加载：

```python
from transformers import BitsAndBytesConfig

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_use_double_quant=True,
)

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    quantization_config=quantization_config,
    device_map="auto"
)