import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Load your Llama model and tokenizer.
# Replace the model identifier with the path or model name you want to use.
model_name = "meta-llama/Meta-Llama-3.1-70B-Instruct"  # or another available checkpoint

tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")

def generate_response(prompt: str) -> str:
    """
    Given an input prompt, generate a response using the Llama model.
    """
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(
        **inputs,
        max_new_tokens=150,
        do_sample=True,
        temperature=0.7,
        top_p=0.9
    )
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Create a Gradio interface for interactive usage.
interface = gr.Interface(
    fn=generate_response,
    inputs=gr.inputs.Textbox(lines=5, placeholder="Enter your prompt here..."),
    outputs="text",
    title="Llama Interactive Generator",
    description="Enter a prompt to see the Llama model generate a response."
)

if __name__ == "__main__":
    interface.launch(debug=True)