app.py

from huggingface_hub import InferenceClient
import gradio as gr
import torch

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

def format_prompt(message, history):
  prompt = "<s>"
  for user_prompt, bot_response in history:
    prompt += f"[INST] {user_prompt} [/INST]"
    prompt += f" {bot_response}</s> "
  prompt += f"[INST] {message} [/INST]"
  return prompt

def generate(
    prompt, history, max_new_tokens, temperature, repetition_penalty, top_p, top_k, seed,
):

    if seed == 0:
        seed = random.randint(1, 100000)
        torch.manual_seed(seed)
    else:
        torch.manual_seed(seed)

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

    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


additional_inputs=[
    gr.Slider(
        label="Max new tokens",
        value=1000,
        minimum=100,
        maximum=32768,
        step=64,
        interactive=True,
        info="The maximum numbers of new tokens, controls how long is the output",
    ),
    gr.Slider(
        label="Temperature",
        value=0.9,
        minimum=0.0,
        maximum=1.0,
        step=0.05,
        interactive=True,
        info="Higher values produce more diverse outputs",
    ),
    gr.Slider(
        label="Repetition penalty",
        value=1.2,
        minimum=1.0,
        maximum=2.0,
        step=0.05,
        interactive=True,
        info="Penalize repeated tokens, making the AI repeat less itself",
    ),
    gr.Slider(
        label="Top-p (nucleus sampling)",
        value=0.90,
        minimum=0.0,
        maximum=1,
        step=0.05,
        interactive=True,
        info="Higher values sample more low-probability tokens",
    ),
    gr.Slider(
        label="Top-k",
        value=1,
        minimum=0,
        maximum=100,
        step=1,
        interactive=True,
        info="Higher k means more diverse outputs by considering a range of tokens",
    ),
    gr.Number(
        label="Seed",
        value=42,
        minimum=1,
        info="Use an integer starting point to initiate the generation process, put 0 for a random",
    ),
]


gr.ChatInterface(
    fn=generate,
    chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
    additional_inputs=additional_inputs,
    title="Mixtral 8x7b Instruct v0.1",
    description="Chatbot Hugging Face space made by [Nick088](https://linktr.ee/Nick088) with costumizable options for model: https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1<br>If you get an erorr, you putted a too much high Max_New_Tokens or your prompt is too long, shorten up one of these",
).launch(show_api=False, share=True)