app.py

import os
import subprocess
from threading import Thread

import random
import torch
import spaces
import gradio as gr
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    BitsAndBytesConfig,
    TextIteratorStreamer,
)

subprocess.run(
    "pip install flash-attn --no-build-isolation",
    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
    shell=True,
)

MODEL_ID = "speakleash/Bielik-7B-Instruct-v0.1"
CHAT_TEMPLATE = "ChatML"
MODEL_NAME = MODEL_ID.split("/")[-1]
CONTEXT_LENGTH = 1024
COLOR = os.environ.get("COLOR")
EMOJI = os.environ.get("EMOJI")
DESCRIPTION = os.environ.get("DESCRIPTION")

# Load model
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
quantization_config = BitsAndBytesConfig(
    load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16
)
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
tokenizer.pad_token = tokenizer.eos_token
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    device_map="auto",
    torch_dtype="auto",
    attn_implementation="flash_attention_2",
)


@spaces.GPU()
def generate(
    instruction,
    stop_tokens,
    temperature,
    max_new_tokens,
    top_k,
    repetition_penalty,
    top_p,
):
    streamer = TextIteratorStreamer(
        tokenizer, skip_prompt=True, skip_special_tokens=True
    )
    enc = tokenizer([instruction], return_tensors="pt", padding=True, truncation=True)
    input_ids, attention_mask = enc.input_ids, enc.attention_mask

    if input_ids.shape[1] > CONTEXT_LENGTH:
        input_ids = input_ids[:, -CONTEXT_LENGTH:]

    generate_kwargs = dict(
        {
            "input_ids": input_ids.to(device),
            "attention_mask": attention_mask.to(device),
        },
        streamer=streamer,
        do_sample=True if temperature else False,
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_k=top_k,
        repetition_penalty=repetition_penalty,
        top_p=top_p,
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()
    outputs = []
    for new_token in streamer:
        outputs.append(new_token)
        if new_token in stop_tokens:
            break
        yield "".join(outputs)


def predict(
    message,
    history,
    system_prompt,
    temperature,
    max_new_tokens,
    top_k,
    repetition_penalty,
    top_p,
):
    repetition_penalty = float(repetition_penalty)
    print(
        "LLL",
        [
            message,
            history,
            system_prompt,
            temperature,
            max_new_tokens,
            top_k,
            repetition_penalty,
            top_p,
        ],
    )
    # Format history with a given chat template
    if CHAT_TEMPLATE == "ChatML":
        stop_tokens = ["<|endoftext|>", "<|im_end|>"]
        instruction = "<|im_start|>system\n" + system_prompt + "\n<|im_end|>\n"
        for human, assistant in history:
            instruction += (
                "<|im_start|>user\n"
                + human
                + "\n<|im_end|>\n<|im_start|>assistant\n"
                + assistant
            )
        instruction += (
            "\n<|im_start|>user\n" + message + "\n<|im_end|>\n<|im_start|>assistant\n"
        )
    elif CHAT_TEMPLATE == "Mistral Instruct":
        stop_tokens = ["</s>", "[INST]", "[INST] ", "<s>", "[/INST]", "[/INST] "]
        instruction = "<s>[INST] " + system_prompt
        for human, assistant in history:
            instruction += human + " [/INST] " + assistant + "</s>[INST]"
        instruction += " " + message + " [/INST]"
    elif CHAT_TEMPLATE == "Bielik":
        stop_tokens = ["</s>"]
        prompt_builder = ["<s>[INST] "]
        if system_prompt:
            prompt_builder.append(f"<<SYS>>\n{system_prompt}\n<</SYS>>\n\n")
        for human, assistant in history:
            prompt_builder.append(f"{human} [/INST] {assistant}</s>[INST] ")
        prompt_builder.append(f"{message} [/INST]")
        instruction = "".join(prompt_builder)
    else:
        raise Exception(
            "Incorrect chat template, select 'ChatML' or 'Mistral Instruct'"
        )
    print(instruction)

    for output_text in generate(
        instruction,
        stop_tokens,
        temperature,
        max_new_tokens,
        top_k,
        repetition_penalty,
        top_p,
    ):
        yield output_text


# Create Gradio interface
def update_examples():
    exs = [["Kim jesteś?"], ["Ile to jest 9+2-1?"], ["Napisz mi coś miłego."]]
    random.shuffle(exs)
    return gr.Dataset(samples=exs)


with gr.Blocks() as demo:
    chatbot = gr.Chatbot(label="Chatbot", likeable=True, render=False)
    chat = gr.ChatInterface(
        predict,
        chatbot=chatbot,
        title=EMOJI + " " + MODEL_NAME + " - online chat demo",
        description=DESCRIPTION,
        examples=[["Kim jesteś?"], ["Ile to jest 9+2-1?"], ["Napisz mi coś miłego."]],
        additional_inputs_accordion=gr.Accordion(
            label="⚙️ Parameters", open=False, render=False
        ),
        additional_inputs=[
            gr.Textbox("", label="System prompt", render=False),
            gr.Slider(0, 1, 0.6, label="Temperature", render=False),
            gr.Slider(128, 4096, 1024, label="Max new tokens", render=False),
            gr.Slider(1, 80, 40, step=1, label="Top K sampling", render=False),
            gr.Slider(0, 2, 1.1, label="Repetition penalty", render=False),
            gr.Slider(0, 1, 0.95, label="Top P sampling", render=False),
        ],
        theme=gr.themes.Soft(primary_hue=COLOR),
    )
    demo.load(update_examples, None, chat.examples_handler.dataset)

demo.queue(max_size=20).launch()