app.py

#!/usr/bin/env python

import os
from collections.abc import Iterator
from threading import Thread

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

DESCRIPTION = """# Swallow-13B instruct"""

if not torch.cuda.is_available():
    DESCRIPTION += "\n<p>Running on CPU 🥶 This demo does not work on CPU.</p>"

if torch.cuda.is_available():
    model_name = "tokyotech-llm/Swallow-13b-instruct-hf"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(
        model_name,
        quantization_config=BitsAndBytesConfig(load_in_8bit=True),
        low_cpu_mem_usage=True,
        device_map="auto",
    )

MAX_INPUT_TOKENS = 2048

PROMPT_DICT = {
    "prompt_input": (
        "以下に、あるタスクを説明する指示があり、それに付随する入力が更なる文脈を提供しています。"
        "リクエストを適切に完了するための回答を記述してください。\n\n"
        "### 指示:\n{instruction}\n\n### 入力:\n{input}\n\n### 応答:"
    ),
    "prompt_no_input": (
        "以下に、あるタスクを説明する指示があります。"
        "リクエストを適切に完了するための回答を記述してください。\n\n"
        "### 指示:\n{instruction}\n\n### 応答:"
    ),
}


def create_prompt(instruction: str, input_text: str | None = None) -> str:
    """Generate a prompt based on the given instruction and an optional input.

    If input is provided, it uses the 'prompt_input' template from PROMPT_DICT.
    If no input is provided, it uses the 'prompt_no_input' template.

    Args:
        instruction (str): The instruction describing the task.
        input_text (str | None): Additional input providing context for the task. Defaults to None.

    Returns:
        str: The generated prompt.

    """
    if input_text:
        # Use the 'prompt_input' template when additional input is provided
        return PROMPT_DICT["prompt_input"].format(instruction=instruction, input=input_text)
    # Use the 'prompt_no_input' template when no additional input is provided
    return PROMPT_DICT["prompt_no_input"].format(instruction=instruction)


@spaces.GPU
@torch.inference_mode()
def run(
    instruction: str,
    input_text: str | None = None,
    max_new_tokens: int = 256,
    temperature: float = 0.99,
    top_p: float = 0.95,
) -> Iterator[str]:
    if input_text == "":
        input_text = None

    prompt = create_prompt(instruction, input_text)
    input_ids = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
    if input_ids.shape[-1] > MAX_INPUT_TOKENS:
        error_message = f"Input exceeds maximum number of tokens ({MAX_INPUT_TOKENS})"
        raise gr.Error(error_message)

    streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
    generate_kwargs = dict(
        {"input_ids": input_ids.to(model.device)},
        streamer=streamer,
        max_new_tokens=max_new_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    outputs = []
    for text in streamer:
        outputs.append(text)
        yield "".join(outputs)


def process_example(instruction: str, input_text: str) -> Iterator[str]:
    yield from run(instruction, input_text)


with gr.Blocks(css_paths="style.css") as demo:
    gr.Markdown(DESCRIPTION)

    with gr.Row():
        with gr.Column():
            instruction = gr.Textbox(label="Instruction", lines=5)
            input_text = gr.Textbox(label="Input (optional)", lines=5)
            run_button = gr.Button()

            with gr.Accordion(label="Advanced Options", open=False):
                max_new_tokens = gr.Slider(label="Max New Tokens", minimum=1, maximum=1024, step=1, value=256)
                temperature = gr.Slider(label="Temperature", minimum=0.0, maximum=2.0, step=0.01, value=0.99)
                top_p = gr.Slider(label="Top P", minimum=0.0, maximum=1.0, step=0.01, value=0.95)

        with gr.Column():
            output = gr.Textbox(label="Output", lines=10)

        run_button.click(
            fn=run,
            inputs=[instruction, input_text, max_new_tokens, temperature, top_p],
            outputs=output,
            api_name="run",
        )

    gr.Examples(
        examples=[
            [
                "以下のトピックに関する詳細な情報を提供してください。",
                "東京工業大学の主なキャンパスについて教えてください。",
            ],
            [
                "以下のトピックに関する詳細な情報を提供してください。",
                "夢オチとは何かについて教えてください。",
            ],
            ["暴れん坊将軍って誰のことですか？", ""],  # noqa: RUF001
        ],
        inputs=[instruction, input_text],
        outputs=output,
        fn=process_example,
        cache_examples=os.getenv("CACHE_EXAMPLES") == "1",
        api_name=False,
    )

if __name__ == "__main__":
    demo.launch()