app.py

from huggingface_hub import InferenceClient
import gradio as gr
import os 

client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")

secret_prompt = os.getenv("SECRET_PROMPT")

def format_prompt(new_message, history):
    prompt = secret_prompt
    for user_msg, bot_msg in history:
        prompt += f"[INST] {user_msg} [/INST]"
        prompt += f" {bot_msg}</s> "
    prompt += f"[INST] {new_message} [/INST]"
    return prompt

def generate(prompt, history,
             temperature=0.25,
             max_new_tokens=512,
             top_p=0.95,
             repetition_penalty=1.0):
                 
    temperature = float(temperature)
                 
    if temperature < 1e-2:
        temperature = 1e-2
        
    top_p = float(top_p)

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=727,
    )

    formatted_prompt = format_prompt(prompt, history)

    stream = client.text_generation(formatted_prompt,
                                    **generate_kwargs,
                                    stream=True,
                                    details=True,
                                    return_full_text=False)
    output = ""

    for response in stream:
        output += response.token.text
        yield output
    return output

samir_chatbot = gr.Chatbot(avatar_images=["./user.png", "./bot.png"],
                           bubble_full_width=False,
                           show_label=False, 
                           show_copy_button=True,
                           likeable=True,)


# Gradio-Demo konfigurieren
theme = 'syddharth/gray-minimal'
demo = gr.ChatInterface(fn=generate, 
                        chatbot=samir_chatbot,
                        title="Ailexs Mixtral 8x7b Chat",
                        theme=theme)

demo.queue().launch(show_api=False)