app.py

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

# Load Llama 3.2-3B-Instruct model locally
model_name = "meta-llama/Llama-3.2-3B-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name, torch_dtype=torch.float16, device_map="auto"
)

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    # Format the conversation history
    messages = [{"role": "system", "content": system_message}]
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})
    prompt = "\n".join([msg["content"] for msg in messages])

    # Tokenize and generate response
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(
        **inputs,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
    )
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Gradio ChatInterface with controls for temperature, tokens, etc.
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
)

if __name__ == "__main__":
    demo.launch()