app.py

import gradio as gr
from huggingfacehub import InferenceClient

"""
or more information on `huggingfacehub` Inference API support, please check the docs: https://huggingface.co/docs/huggingfacehub/v0.22.2/en/guides/inference
"""
client = InferenceClient("bigscience/bloom")


def respond(
    message,
    history: list[tuple[str, str]],
    systemmessage,
    maxtokens,
    temperature,
    topp,
):
    messages = [{"role": "system", "content": systemmessage}]

    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 client.chatcompletion(
        messages,
        maxtokens=maxtokens,
        stream=rue,
        temperature=temperature,
        topp=topp,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

"""
or information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    additionalinputs=[
        gr.extbox(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="emperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="op-p (nucleus sampling)",
        ),
    ],
)


if name == "main":
    demo.launch()