from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Load the model (TinyLlama)
model_name = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

def get_llm_reply(user_input: str) -> str:
    prompt = f"User: {user_input}\nAssistant:"
    inputs = tokenizer(prompt, return_tensors="pt")
    output = model.generate(
        **inputs,
        max_new_tokens=100,
        do_sample=True,
        temperature=0.7
    )
    response = tokenizer.decode(output[0], skip_special_tokens=True)
    return response.replace(prompt, "").strip()