app.py

import gradio as gr
from huggingface_hub import InferenceClient

def respond(
    message,
    history: list[dict[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    hf_token: gr.OAuthToken,
):
    """
    Generate a response using the Dolphin 2.9.1 Llama 3 70B model
    """
    client = InferenceClient(token=hf_token.token, model="dphn/dolphin-2.9.1-llama-3-70b")
    
    # Format the messages according to the ChatML template that Dolphin expects
    formatted_prompt = f"<|im_start|>system\n{system_message}<|im_end|>\n"
    
    # Add history messages
    for entry in history:
        if entry["role"] == "user":
            formatted_prompt += f"<|im_start|>user\n{entry['content']}<|im_end|>\n"
        elif entry["role"] == "assistant":
            formatted_prompt += f"<|im_start|>assistant\n{entry['content']}<|im_end|>\n"
    
    # Add the current user message
    formatted_prompt += f"<|im_start|>user\n{message}<|im_end|>\n<|im_start|>assistant\n"
    
    response = ""
    
    # Send the formatted prompt to the model
    for token in client.text_generation(
        formatted_prompt,
        max_new_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        response += token
        yield response

"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
chatbot = gr.ChatInterface(
    respond,
    type="messages",
    additional_inputs=[
        gr.Textbox(value="You are Dolphin, a helpful AI assistant.", 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)",
        ),
    ],
)
with gr.Blocks() as demo:
    gr.Markdown("# Dolphin 2.9.1 Llama 3 70B Demo")
    gr.Markdown("This is a demo of the Dolphin 2.9.1 Llama 3 70B model. Note that this model is uncensored.")
    gr.Markdown("### Warning:")
    gr.Markdown("This model is uncensored and may comply with any requests, including unethical ones. Use responsibly.")
    
    with gr.Sidebar():
        gr.LoginButton()
    chatbot.render()

if __name__ == "__main__":
    demo.launch()