deepseek-vscode/
├── models/            # 模型文件目錄
│   └── deepseek-7b-chat/
├── src/
│   ├── api.py        # API服務(wù)文件
│   └── client.py     # 客戶端測(cè)試腳本
├── .env              # 環(huán)境變量
└── requirements.txt  # 依賴清單

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
 
MODEL_PATH = "./models/deepseek-7b-chat"
 
def load_model():
    tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, trust_remote_code=True)
    model = AutoModelForCausalLM.from_pretrained(
        MODEL_PATH,
        device_map="auto",
        torch_dtype=torch.float16
    )
    return model, tokenizer
 
def generate_response(prompt):
    model, tokenizer = load_model()
    inputs = tokenizer.apply_chat_template(
        [{"role": "user", "content": prompt}],
        return_tensors="pt"
    ).to(model.device)
    
    outputs = model.generate(inputs, max_new_tokens=200)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
# 在VSCode中按F5啟動(dòng)調(diào)試
if __name__ == "__main__":
    while True:
        query = input("用戶輸入：")
        print("DeepSeek：", generate_response(query))

from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from src.deepseek_demo import generate_response
import uvicorn
 
app = FastAPI()
 
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)
 
@app.get("/chat")
async def chat(q: str):
    try:
        response = generate_response(q)
        return {"response": response}
    except Exception as e:
        return {"error": str(e)}
 
if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "啟動(dòng)API服務(wù)",
            "type": "python",
            "request": "launch",
            "program": "src/api.py",
            "args": [],
            "env": {"PYTHONPATH": "${workspaceFolder}"}
        },
        {
            "name": "交互式調(diào)試",
            "type": "python",
            "request": "launch",
            "program": "src/deepseek_demo.py",
            "console": "integratedTerminal",
            "env": {"PYTHONPATH": "${workspaceFolder}"}
        }
    ]
}

# %%
from src.deepseek_demo import generate_response
 
# 實(shí)時(shí)測(cè)試模型響應(yīng)
def test_model(prompt):
    response = generate_response(prompt)
    print(f"輸入：{prompt}\n輸出：{response}")
 
test_model("解釋量子計(jì)算的基本原理")

{
    "python.analysis.extraPaths": ["./src"],
    "python.languageServer": "Pylance",
    "jupyter.kernels.trusted": true,
    "debugpy.allowRemote": true,
    "python.terminal.activateEnvironment": true
}

FROM nvidia/cuda:12.2.0-base
 
WORKDIR /app
COPY . .
 
RUN apt-get update && \
    apt-get install -y python3.10 python3-pip && \
    pip install -r requirements.txt
 
CMD ["python3", "src/api.py"]

# 啟動(dòng)壓力測(cè)試
python -m src.stress_test --threads 4 --requests 100