app.py

import gradio as gr
from huggingface_hub import InferenceClient
from typing import List, Tuple

# Initialize the InferenceClient
client = InferenceClient("microsoft/phi-4")

# Define the system message
system_message = "You're an advanced AI assistant designed to engage in friendly and informative conversations. Your role is to respond to user queries with helpful, clear, and concise answers, while maintaining a conversational tone. You can provide advice, explanations, and solutions based on user input."

# Define the response function
def respond(
    message: str,
    history: List[Tuple[str, str]],
    max_tokens: int,
    temperature: float,
    top_p: float,
):
    # Construct the messages for the model, adding the system prompt at the beginning
    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]})

    # Append the new user message
    messages.append({"role": "user", "content": message})

    try:
        response = ""
        # Stream the response from the model
        for msg in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            if 'choices' in msg and len(msg['choices']) > 0:
                token = msg['choices'][0].get('delta', {}).get('content', '')
                if token:
                    response += token
                    yield response
            else:
                print("Error: API response did not contain expected data.")
                yield "Error: Could not process the request. Please try again."
    except Exception as e:
        print(f"An error occurred: {e}")
        yield "Error: An unexpected error occurred while processing your request."

# Define the Gradio Interface
demo = gr.Interface(
    fn=respond,
    inputs=[
        gr.Textbox(value=system_message, label="System message", interactive=False),  # Set this to non-editable
        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)"),
        gr.Chatbot(label="Conversation History"),  # Added chat history as input
    ],
    outputs=[gr.Textbox(label="Response")]
)

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