app.py

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

app = FastAPI()

class UserInput(BaseModel):
    prompt: str

# Load model and tokenizer
model_name = "NousResearch/Nous-Hermes-llama2-13b"
tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto",
    trust_remote_code=True
)

generator = pipeline("text-generation", model=model, tokenizer=tokenizer)

@app.post("/chat")
async def chat(data: UserInput):
    output = generator(
        data.prompt,
        max_new_tokens=512,
        do_sample=True,
        temperature=0.8,
        top_k=50,
        top_p=0.95
    )
    return {"response": output[0]["generated_text"]}