app.py

import gradio as gr
import spaces
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

# 只加载一次模型和分词器
MODEL_NAME = "inclusionAI/Ring-mini-2.0"
device = "cuda" if torch.cuda.is_available() else "cpu"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    torch_dtype=torch.float16 if device == "cuda" else torch.float32,
    trust_remote_code=True
).to(device)

@spaces.GPU
def respond(
    message,
    history: list[dict[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    hf_token: gr.OAuthToken = None,  # 保持参数兼容
):
    """
    使用 transformers 在 GPU 上本地推理 inclusionAI/Ring-mini-2.0
    """
    # 拼接历史和 system prompt，兼容 gradio ChatInterface 的消息格式
    prompt = system_message + "\n"
    # gradio history: [{"role": "user"/"assistant", "content": "..."}, ...]
    last_role = None
    for turn in history:
        if turn.get("role") == "user":
            prompt += f"User: {turn['content']}\n"
            last_role = "user"
        elif turn.get("role") == "assistant":
            prompt += f"Assistant: {turn['content']}\n"
            last_role = "assistant"
    prompt += f"User: {message}\nAssistant:"

    input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to(device)
    output_ids = model.generate(
        input_ids,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id,
    )
    output = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)
    # 流式输出
    response = ""
    for token in output.split():
        response += token + " "
        yield response.strip()


"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
chatbot = gr.ChatInterface(
    respond,
    type="messages",
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
)

with gr.Blocks() as demo:
    gr.Markdown("# HuggingFace Running")
    with gr.Sidebar():
        gr.LoginButton()
    chatbot.render()


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