app.py

import gradio as gr
from huggingface_hub import InferenceClient
import os
import requests

# Set up the inference API client
hf_client = InferenceClient("mistralai/Mistral-Nemo-Instruct-2407", token=os.getenv("HF_TOKEN"))

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):

    system_prefix = """
If the input language is Korean, respond in Korean. If it's English, respond in English.
    """
    
    messages = [{"role": "system", "content": f"{system_prefix} {system_message}"}]  # Add prefix

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    response = ""

    for message in hf_client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content
        if token is not None:
            response += token.strip("")  # Remove tokens
        yield response


theme = "Nymbo/Nymbo_Theme"

css = """
footer {
    visibility: hidden;
}
"""

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="""
You are an AI assistant.
""", label="System Prompt"),
        gr.Slider(minimum=1, maximum=2000, 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)",
        ),
    ],
    theme=theme,  # Apply theme
    css=css  # Apply CSS
)

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