Docker vLLM 快速部署 Qwen2.5

Dockerfile：

FROM kevinchina/deeplearning:llamafactory20241027

# 设置工作目录
WORKDIR /app

# 暴露端口
EXPOSE 8000
EXPOSE 7860

# 使用 JSON 格式的 ENTRYPOINT，指定要执行的命令
# ENTRYPOINT ["llamafactory-cli", "api", "--model_name_or_path", "/data/xiedong/Qwen2.5-72B-Instruct-GPTQ-Int4", "--template", "qwen", "--infer_dtype", "bfloat16"]


ENTRYPOINT ["vllm", "serve"