|
from fastapi import FastAPI, Request, HTTPException |
|
import requests |
|
import os |
|
from pydantic import BaseModel |
|
|
|
app = FastAPI() |
|
|
|
|
|
API_URL = "https://api-inference.huggingface.co/models/meta-llama/Meta-Llama-3-8B-Instruct" |
|
HF_TOKEN = os.getenv("HF_API_KEY") |
|
headers = {"Authorization": f"Bearer {HF_TOKEN}"} |
|
|
|
|
|
class PromptRequest(BaseModel): |
|
prompt: str |
|
|
|
|
|
@app.post("/generate") |
|
async def generate_text(data: PromptRequest): |
|
try: |
|
response = requests.post(API_URL, headers=headers, json={"inputs": data.prompt}) |
|
response.raise_for_status() |
|
return response.json() |
|
except Exception as e: |
|
raise HTTPException(status_code=500, detail=str(e)) |
|
|