app.py

from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from transformers import AutoModelForCausalLM, AutoTokenizer

app = FastAPI()

model_name = "Hadeel11/fine-tuned-model"
model = AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

class Query(BaseModel):
    question: str
    contexts: list

@app.post("/predict/")
async def predict(query: Query):
    inputs = tokenizer(query.question, return_tensors="pt")
    outputs = model.generate(**inputs)
    answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return {"answer": answer}