app.py

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from fastapi import FastAPI
from pydantic import BaseModel
import torch

tokenizer = AutoTokenizer.from_pretrained("cmlang/somali-flan-t5")
model = AutoModelForSeq2SeqLM.from_pretrained("cmlang/somali-flan-t5")

app = FastAPI()

class Message(BaseModel):
    text: str

@app.post("/chatbot")
async def chatbot(msg: Message):
    inputs = tokenizer(msg.text, return_tensors="pt")
    outputs = model.generate(**inputs, max_length=150)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return {"response": response}