app.py

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

# Model ID
model_id = "apu20/Llama-3.2-3B-Instruct_Tele"

# Load quantized model (switch to 8-bit if needed)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,  # Use float16 for reduced memory footprint
    device_map="cpu"  # Force model to run on CPU
)

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_id)


def respond(message, history, system_message, max_tokens, temperature, top_p):
    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})

    # Tokenize input
    inputs = tokenizer(message, return_tensors="pt").to("cpu")  # Ensure inputs are on CPU

    # Generate response
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_length=max_tokens,
            temperature=temperature,
            top_p=top_p
        )

    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response


# Gradio Chat Interface
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()