app.py

import gradio as gr
from huggingface_hub import InferenceClient

"""
For more information on `huggingface_hub` Inference API support, please check the docs: 
https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""

# Initialize the Hugging Face Inference Client
# Option 1: Use environment variable for authenticated access
import os
token = os.getenv("HUGGINGFACE_API_TOKE")

# Try different models - some may not require authentication
# client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=token)
# client = InferenceClient("microsoft/DialoGPT-medium", token=token)
# client = InferenceClient("google/flan-t5-base", token=token)

# If you have a token, use it:
if token:
    client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=token)
else:
    # Fallback to a model that might not require authentication
    print("No token found. Using alternative model...")
    client = InferenceClient("microsoft/DialoGPT-medium")

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    """
    Generate response using the Zephyr model
    """
    try:
        # Build the conversation history
        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]})
        
        # Add the current user message
        messages.append({"role": "user", "content": message})
        
        response = ""
        
        # Stream the response
        for message_chunk in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            if message_chunk.choices[0].delta.content is not None:
                token = message_chunk.choices[0].delta.content
                response += token
                yield response
                
    except Exception as e:
        yield f"Error: {str(e)}"

"""
For information on how to customize the ChatInterface, peruse the gradio docs: 
https://www.gradio.app/docs/chatinterface
"""

# Create the Gradio ChatInterface
demo = gr.ChatInterface(
    respond,
    title="🤖 Zephyr-7B Chatbot",
    description="Chat with Zephyr-7B model. Adjust the parameters below to customize the response behavior.",
    theme="soft",
    additional_inputs=[
        gr.Textbox(
            value="You are a friendly and helpful AI assistant.", 
            label="System message",
            placeholder="Enter system prompt to guide the assistant's behavior..."
        ),
        gr.Slider(
            minimum=1, 
            maximum=2048, 
            value=512, 
            step=1, 
            label="Max new tokens",
            info="Maximum number of tokens to generate"
        ),
        gr.Slider(
            minimum=0.1, 
            maximum=4.0, 
            value=0.7, 
            step=0.1, 
            label="Temperature",
            info="Controls randomness (lower = more focused)"
        ),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
            info="Controls diversity of responses"
        ),
    ]
)

if __name__ == "__main__":
    demo.launch(
        share=True,  # Set to True to create a public link
        server_name="0.0.0.0",  # Allow external connections
        server_port=7860,  # Default port
        show_error=True  # Show detailed error messages
    )