app.py

import gradio as gr
from huggingface_hub import InferenceClient

# Define available models.
models = [
    {
        "name": "Tiny Model",
        "description": "A small chat model.",
        "id": "amusktweewt/tiny-model-500M-chat-v2",
        "enabled": True
    },
    {
        "name": "Another Model",
        "description": "A bigger chat model (disabled).",
        "id": "another-model",
        "enabled": False
    }
]

# Build the HTML for the custom dropdown.
dropdown_options = ""
for model in models:
    disabled_attr = "disabled" if not model["enabled"] else ""
    label = f"{model['name']}: {model['description']}"
    if not model["enabled"]:
        label = f"{model['name']} (Disabled): {model['description']}"
    dropdown_options += f'<option value="{model["id"]}" {disabled_attr}>{label}</option>\n'

dropdown_html = f"""
<div>
  <label for="model_select"><strong>Select Model:</strong></label>
  <select id="model_select" onchange="document.getElementById('hidden_model').value = this.value;">
    {dropdown_options}
  </select>
</div>
"""

def respond(message, history: list[tuple[str, str]], model_id, system_message, max_tokens, temperature, top_p):
    # Instantiate the InferenceClient using the selected model.
    client = InferenceClient(model_id)
    
    messages = []
    if system_message:
        messages.append({"role": "system", "content": system_message})
    if history:
        for user_msg, bot_msg in history:
            messages.append({"role": "user", "content": user_msg})
            messages.append({"role": "assistant", "content": bot_msg})
    messages.append({"role": "user", "content": message})
    messages.append({"role": "assistant", "content": ""})
    
    response_text = ""
    # Stream the response token-by-token.
    for resp in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = resp.choices[0].delta.content
        response_text += token
        yield response_text

# Build the interface using Gradio Blocks.
with gr.Blocks() as demo:
    # Display the custom dropdown.
    gr.HTML(value=dropdown_html)
    # Hidden textbox to capture the selected model ID.
    hidden_model = gr.Textbox(value=models[0]["id"], visible=False, elem_id="hidden_model")
    
    # Create the ChatInterface.
    chat_interface = gr.ChatInterface(
        fn=respond,
        additional_inputs=[
            # Pass the hidden model selector.
            hidden_model,
            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()