app.py

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

# Load model and tokenizer
model_name = "martinbravo/llama_finetuned_test"
base_model_name = "unsloth/llama-3.2-1b-instruct-bnb-4bit"

# Load tokenizer and model locally
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",  # Automatically maps model to GPU/CPU
    trust_remote_code=True,  # If model uses custom implementations
)

# Create a text-generation pipeline
generator = pipeline("text-generation", model=model, tokenizer=tokenizer, device=0)


def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    # Build input prompt
    prompt = system_message + "\n"
    for user_input, assistant_response in history:
        prompt += f"User: {user_input}\nAssistant: {assistant_response}\n"
    prompt += f"User: {message}\nAssistant:"

    # Generate response
    response = generator(
        prompt,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,  # Sampling for diverse responses
    )[0]["generated_text"]

    # Extract the assistant's response
    assistant_response = response[len(prompt) :].strip()
    yield assistant_response


# Gradio 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()