from contextlib import asynccontextmanager from fastapi import FastAPI from app.model_loader import load_model from app.routes import router @asynccontextmanager async def lifespan(app: FastAPI): tokenizer, model = load_model() app.state.tokenizer = tokenizer app.state.model = model yield app.state.tokenizer = None app.state.model = None app = FastAPI( title="Qwen Bash Tool Calling API", lifespan=lifespan, ) app.include_router(router)