app.py

import gradio as gr
from huggingface_hub import InferenceClient
import time

client = InferenceClient("Qwen/Qwen2.5-3b-Instruct")

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    progress=gr.Progress()  # 進捗表示用
):
    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

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

    # AI応答時間計測開始
    start_time = time.time()  
    response = client.chat_completion(
        messages,
        max_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
    )
    elapsed_time = time.time() - start_time  # AI応答時間計測終了

    # ユーザーに進捗を表示
    progress(0, f"応答中... {elapsed_time:.2f}秒")  # 初期応答時間表示
    time.sleep(0.5)  # 応答中に少し待機
    total_response_time = elapsed_time + 0.5  # 総応答時間を計算

    return response.choices[0].message.content, f"予測時間: {elapsed_time:.2f}秒 / 総応答時間: {total_response_time:.2f}秒"

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="ユーザーの質問や依頼にのみ答えてください。ポジティブに答えてください", label="システムメッセージ"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="最大新規トークン"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="温度"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (核サンプリング)",
        ),
    ],
    css="""
    .gradio-container {
        background-color: #212121;
    }
    """
)

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