app.py

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

url = "http://59.110.170.104:8085/chat_completion"


def respond(
    message,
    history: list[tuple[str, str]],
    do_sample: bool,
    seed: int,
    max_new_tokens,
    temperature,
    top_p,
    top_k,
    repetition_penalty
):
    messages = []

    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 = ""
    request_data = dict(
        messages=messages,
        max_new_tokens=max_new_tokens,
        do_sample=do_sample,
        seed=seed,
        top_p=top_p,
        top_k=top_k,
        temperature=temperature,
        repetition_penalty=repetition_penalty
    )
    print(request_data)
    with requests.post(url, json=request_data, stream=True, headers={"Authorization": f"Bearer {os.environ['HF_TOKEN']}"}) as r:
        # printing response of each stream
        for chunk in r.iter_content(1024):
            response += chunk.decode("utf8")
            yield response

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

demo = gr.ChatInterface(
    respond,
    chatbot=gr.Chatbot(height=600),
    additional_inputs=[
        # gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        gr.Checkbox(True, label="do sample"),
        gr.Number(42, precision=0, label="seed"),
        gr.Slider(minimum=1, maximum=2048, value=1024, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.01, maximum=4.0, value=0.7, step=0.01, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=1.0,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
        gr.Slider(
            minimum=0,
            maximum=100,
            value=0,
            step=1,
            label="Top-K (Top-K sampling)",
        ),
        gr.Slider(
            minimum=1,
            maximum=2,
            value=1.03,
            step=0.01,
            label="repetition penalty",
        ),
    ],
)


if __name__ == "__main__":
    demo.queue(default_concurrency_limit=2, max_size=10)
    demo.launch()