app.py

import os
import time
import spaces
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, BitsAndBytesConfig
import gradio as gr
from threading import Thread

# モデルの定義
MODELS = {
    "Borea-Phi-3.5-mini-Jp": "AXCXEPT/Borea-Phi-3.5-mini-Instruct-Jp",
    "EZO-Common-9B": "HODACHI/EZO-Common-9B-gemma-2-it",
    "Phi-3.5-mini": "microsoft/Phi-3.5-mini-instruct",
}

HF_TOKEN = os.environ.get("HF_TOKEN", None)

# タイトルとプレースホルダーを日本語に変更
TITLE = "<h1><center>Borea/EZO デモアプリ</center></h1>"

PLACEHOLDER = """
<center>
<p>こんにちは、私はAIアシスタントです。何でも質問してください。</p>
</center>
"""

CSS = """
.duplicate-button {
    margin: auto !important;
    color: white !important;
    background: black !important;
    border-radius: 100vh !important;
}
h3 {
    text-align: center;
}
"""

device = "cuda" if torch.cuda.is_available() else "cpu"

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4"
)

model = None
tokenizer = None

def load_model(model_name):
    global model, tokenizer
    model_path = MODELS[model_name]
    tokenizer = AutoTokenizer.from_pretrained(model_path)
    model = AutoModelForCausalLM.from_pretrained(
        model_path,
        torch_dtype=torch.bfloat16,
        device_map="auto",
        quantization_config=quantization_config
    )

@spaces.GPU()
def stream_chat(
    message: str, 
    history: list,
    system_prompt: str,
    temperature: float = 0.8, 
    max_new_tokens: int = 1024, 
    top_p: float = 1.0, 
    top_k: int = 20, 
    repetition_penalty: float = 1.2,
    model_name: str = "Phi-3.5-mini"
):
    global model, tokenizer

    if model is None or tokenizer is None or model.name_or_path != MODELS[model_name]:
        load_model(model_name)

    print(f'message: {message}')
    print(f'history: {history}')

    conversation = [
        {"role": "system", "content": system_prompt}
    ]
    for prompt, answer in history:
        conversation.extend([
            {"role": "user", "content": prompt}, 
            {"role": "assistant", "content": answer},
        ])

    conversation.append({"role": "user", "content": message})

    input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt").to(model.device)
    
    streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
    
    generate_kwargs = dict(
        input_ids=input_ids, 
        max_new_tokens=max_new_tokens,
        do_sample=False if temperature == 0 else True,
        top_p=top_p,
        top_k=top_k,
        temperature=temperature,
        repetition_penalty=repetition_penalty,
        eos_token_id=tokenizer.eos_token_id,
        streamer=streamer,
    )

    with torch.no_grad():
        thread = Thread(target=model.generate, kwargs=generate_kwargs)
        thread.start()
        
    buffer = ""
    for new_text in streamer:
        buffer += new_text
        yield buffer

chatbot = gr.Chatbot(height=600, placeholder=PLACEHOLDER)

with gr.Blocks(css=CSS, theme='ParityError/Interstellar') as demo:
    gr.HTML(TITLE)
    gr.ChatInterface(
        fn=stream_chat,
        chatbot=chatbot,
        fill_height=True,
        additional_inputs=[
            gr.Textbox(
                value="あなたは親切なアシスタントです。",
                label="システムプロンプト",
            ),
            gr.Slider(
                minimum=0,
                maximum=1,
                step=0.1,
                value=0.8,
                label="温度 (Temperature)",
            ),
            gr.Slider(
                minimum=128,
                maximum=8192,
                step=1,
                value=1024,
                label="最大新規トークン数",
            ),
            gr.Slider(
                minimum=0.0,
                maximum=1.0,
                step=0.1,
                value=1.0,
                label="top_p",
            ),
            gr.Slider(
                minimum=1,
                maximum=20,
                step=1,
                value=20,
                label="top_k",
            ),
            gr.Slider(
                minimum=1.0,
                maximum=2.0,
                step=0.1,
                value=1.2,
                label="繰り返しペナルティ",
            ),
            gr.Dropdown(
                choices=list(MODELS.keys()),
                value="Borea-Phi-3.5-mini-Jp",
                label="モデル選択",
            ),
        ],
        examples=[
            ["語彙の勉強を手伝ってください。空欄を埋めるための文章を書いてください。私は正しい選択肢を選びます。"],
            ["子供のアート作品でできる5つの創造的なことを教えてください。捨てたくはないのですが、散らかってしまいます。"],
            ["ローマ帝国についてのランダムな面白い事実を教えてください。"],
            ["ウェブサイトの固定ヘッダーのCSSとJavaScriptのコードスニペットを見せてください。"],
        ],
        cache_examples=False,
    )

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