app.py

import gradio as gr
from huggingface_hub import InferenceClient

# Initialize the InferenceClient with the model ID from Hugging Face
client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta")

def respond(
    message: str,
    history: list[tuple[str, str]],
    system_message: str,
    max_tokens: int,
    temperature: float,
    top_p: float,
):
    """
    Generates a response from the AI model based on the user's message and chat history.
    
    Args:
        message (str): The user's input message.
        history (list): A list of tuples representing the conversation history (user, assistant).
        system_message (str): A system-level message guiding the AI's behavior.
        max_tokens (int): The maximum number of tokens for the output.
        temperature (float): Sampling temperature for controlling the randomness.
        top_p (float): Top-p (nucleus sampling) for controlling diversity.

    Yields:
        str: The AI's response as it is generated.
    """
    
    # Prepare the conversation history for the API call
    messages = [{"role": "system", "content": system_message}]
    
    for user_input, assistant_response in history:
        if user_input:
            messages.append({"role": "user", "content": user_input})
        if assistant_response:
            messages.append({"role": "assistant", "content": assistant_response})

    # Add the latest user message to the conversation
    messages.append({"role": "user", "content": message})

    # Initialize an empty response
    response = ""

    try:
        # Generate a response from the model with streaming
        for message in client.chat_completion(
            messages=messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = message.choices[0].delta.content
            response += token
            yield response

    except Exception as e:
        yield f"An error occurred: {str(e)}"

# Define the UI layout with a more user-friendly design
with gr.Blocks() as demo:
    gr.Markdown("# 🧠 AI Chatbot Interface")
    gr.Markdown("### Customize your AI Chatbot's behavior and responses.")
    
    with gr.Row():
        chatbot = gr.Chatbot()
        with gr.Column():
            system_message = gr.Textbox(value="You are a friendly Chatbot.", label="System message", lines=2)
            max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens")
            temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
            top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
    
    with gr.Row():
        message = gr.Textbox(label="Your message:", lines=1)
        submit_btn = gr.Button("Send")
    
    # Update the chatbot with the new message and response
    submit_btn.click(respond, 
                     inputs=[message, chatbot, system_message, max_tokens, temperature, top_p], 
                     outputs=[chatbot],
                     show_progress=True)

# Launch the Gradio interface
if __name__ == "__main__":
    demo.launch()