app.py

import os
import torch
from fastapi import FastAPI
from pydantic import BaseModel
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import uvicorn

# Define a Pydantic model for request validation
class Query(BaseModel):
    text: str

# Initialize FastAPI app
app = FastAPI(title="Financial Chatbot API")

# Load your fine-tuned model and tokenizer
model_name = "Phoenix21/meta-llama-Llama-3.2-3B-2025-03-13-checkpoints"
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
tokenizer.pad_token = tokenizer.eos_token

# Create a text-generation pipeline
chat_pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    max_new_tokens=256,
    temperature=0.7,
    top_p=0.95,
)

# Define an endpoint for generating responses
@app.post("/generate")
def generate(query: Query):
    prompt = f"Question: {query.text}\nAnswer: "
    response = chat_pipe(prompt)[0]["generated_text"]
    return {"response": response}

# Run the app using uvicorn. Hugging Face Spaces sets the PORT environment variable.
if __name__ == "__main__":
    port = int(os.environ.get("PORT", 8000))
    uvicorn.run(app, host="0.0.0.0", port=port)