app.py

import gradio as gr
import requests

# Replace with your actual ngrok URL from Colab
COLAB_BACKEND_URL = "https://64d0-34-124-237-140.ngrok-free.app/generate"

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    # Construct the prompt with chat history and system prompt
    full_prompt = system_message.strip() + "\n\n"
    for user_msg, bot_msg in history:
        if user_msg:
            full_prompt += f"User: {user_msg.strip()}\n"
        if bot_msg:
            full_prompt += f"AI: {bot_msg.strip()}\n"
    full_prompt += f"User: {message.strip()}\nAI:"

    try:
        # Send the prompt and generation parameters to the Colab backend
        response = requests.post(COLAB_BACKEND_URL, json={
            "prompt": full_prompt,
            "max_tokens": max_tokens,
            "temperature": temperature,
            "top_p": top_p,
        })
        reply = response.json().get("response", "")
        yield reply.strip()
    except Exception as e:
        yield f"[Error contacting backend: {str(e)}]"

# Gradio interface
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a flirty, romantic AI girlfriend.", 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=0.95, 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()