只需几步！用 vLLM 快速上手 DeepSeek 大模型部署

# 配置清华镜像源加速下载
python -m pip install --upgrade pip
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
# 安装核心依赖包
pip install modelscope==1.22.3
pip install openai==1.61.0
pip install vllm==0.7.1
1.
2.
3.
4.
5.
6.
7.

from modelscope.hub.snapshot_download import snapshot_download
# 下载预训练模型并指定缓存路径
model_dir = snapshot_download('deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', cache_dir='/root/autodl-tmp', revision='master')
1.
2.
3.

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer
import os
import json


# 自动下载模型时，指定使用modelscope; 否则，会从HuggingFace下载
os.environ['VLLM_USE_MODELSCOPE']='True'
1.
2.
3.
4.
5.
6.
7.
8.

def get_completion(prompts, model, tokenizer=None, max_tokens=8192, temperature=0.6, top_p=0.95, max_model_len=2048):
    stop_token_ids = [151329, 151336, 151338]
    # 创建采样参数。temperature 控制生成文本的多样性，top_p 控制核心采样的概率
    sampling_params = SamplingParams(temperature=temperature, top_p=top_p, max_tokens=max_tokens, stop_token_ids=stop_token_ids)
    # 初始化 vLLM 推理引擎
    llm = LLM(model=model, tokenizer=tokenizer, max_model_len=max_model_len,trust_remote_code=True)
    outputs = llm.generate(prompts, sampling_params)
    return outputs
1.
2.
3.
4.
5.
6.
7.
8.

# 初始化 vLLM 推理引擎
model='/root/autodl-tmp/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B' # 指定模型路径
# model="deepseek-ai/DeepSeek-R1-Distill-Qwen-7B" # 指定模型名称，自动下载模型
tokenizer = None
# 加载分词器后传入vLLM 模型，但不是必要的。
# tokenizer = AutoTokenizer.from_pretrained(model, use_fast=False) 


text = ["请帮我推荐一种深圳到北京的出行方式<think>\n", ]
1.
2.
3.
4.
5.
6.
7.
8.
9.

# 思考需要输出更多的 Token 数，max_tokens 设为 8K，根据 DeepSeek 官方的建议，temperature应在 0.5-0.7，推荐 0.6
outputs = get_completion(text, model, tokenizer=tokenizer, max_tokens=8192, temperature=0.6, top_p=0.95, max_model_len=2048) 


# 输出是一个包含 prompt、生成文本和其他信息的 RequestOutput 对象列表。
# 打印输出。
for output in outputs:
    prompt = output.prompt
    generated_text = output.outputs[0].text
    if r"</think>" in generated_text:
        think_content, answer_content = generated_text.split(r"</think>")
    else:
        think_content = ""
        answer_content = generated_text
print(f"Prompt: {prompt!r}, Think: {think_content!r}, Answer: {answer_content!r}")
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.

python -m vllm.entrypoints.openai.api_server \
  --model /root/autodl-tmp/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B \
  --served-model-name DeepSeek-R1-Distill-Qwen-7B \
  --max-model-len=2048
1.
2.
3.
4.

from openai import OpenAI
 
client = OpenAI(base_url="http://localhost:8000/v1", api_key="sk-xxx")
response = client.completions.create(
    model="DeepSeek-R1-Distill-Qwen-7B",
    prompt="简单介绍一下deepseek？<think>\n",
    max_tokens=1024
)
print(response.choices[0].text)
1.
2.
3.
4.
5.
6.
7.
8.
9.

chat_response = client.chat.completions.create(
    model="DeepSeek-R1-Distill-Qwen-7B",
    messages=[{"role": "user", "content": "我买了10个苹果，给了你3个，自己吃了1个，还剩几个？"}]
)
print(chat_response.choices[0].message.content)
1.
2.
3.
4.
5.

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "DeepSeek-R1-Distill-Qwen-7B",
        "prompt": "我买了10个苹果，给了你3个，自己吃了1个，还剩几个？<think>\n",
        "max_tokens": 1024,
        "temperature": 0
}'
1.
2.
3.
4.
5.
6.
7.
8.