ELYZA Shortcut 1.0 Qwen 32B

🚀 ELYZA-Shortcut-1.0-Qwen-32B

ELYZA-Shortcut-1.0-Qwen-32B 是在推理模型 ELYZA-Thinking-1.0-Qwen-32B 開發過程中衍生出的非推理模型。該模型基於 Qwen/Qwen2.5-32B-Instruct，經過後續訓練，可繞過逐步推理過程，直接生成最終答案（基於通義千問構建）。

🚀 快速開始

本模型可結合 Hugging Face Transformers 庫使用。以下是使用該模型進行推理的示例代碼：

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "elyza/ELYZA-Shortcut-1.0-Qwen-32B"
prompt = "仕事の熱意を取り戻すためのアイデアを5つ挙げてください。"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto",
)
model.eval()
messages = [{"role": "user", "content": prompt}]
input_text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
token_ids = tokenizer.encode(
    input_text, add_special_tokens=False, return_tensors="pt"
)
with torch.no_grad():
    output_ids = model.generate(
        token_ids.to(model.device),
        max_new_tokens=8192,
        do_sample=True,
        temperature=0.6,
        top_p=0.95,
    )
output = tokenizer.decode(
    output_ids.tolist()[0][token_ids.size(1):], skip_special_tokens=True
)
print(output)

對於模型部署，建議使用 vLLM 創建兼容 OpenAI 的服務器，示例命令如下：

vllm serve elyza/ELYZA-Shortcut-1.0-Qwen-32B \
    --tensor-parallel-size 8 \
    --max-model-len 32768

✨ 主要特性

在後續訓練階段，該模型使用問題 - 解決方案對進行有監督微調（SFT）訓練。這些問題 - 解決方案對是通過基於蒙特卡羅樹搜索（MCTS）的算法探索出最優推理路徑後，去除推理步驟得到的。更多詳細信息，請參考 我們的博客文章。

💻 使用示例

基礎用法

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "elyza/ELYZA-Shortcut-1.0-Qwen-32B"
prompt = "仕事の熱意を取り戻すためのアイデアを5つ挙げてください。"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto",
)
model.eval()
messages = [{"role": "user", "content": prompt}]
input_text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
token_ids = tokenizer.encode(
    input_text, add_special_tokens=False, return_tensors="pt"
)
with torch.no_grad():
    output_ids = model.generate(
        token_ids.to(model.device),
        max_new_tokens=8192,
        do_sample=True,
        temperature=0.6,
        top_p=0.95,
    )
output = tokenizer.decode(
    output_ids.tolist()[0][token_ids.size(1):], skip_special_tokens=True
)
print(output)

高級用法

vllm serve elyza/ELYZA-Shortcut-1.0-Qwen-32B \
    --tensor-parallel-size 8 \
    --max-model-len 32768

📚 詳細文檔

引用方式

@misc{elyza2025thinking,
    title={elyza/ELYZA-Thinking-1.0-Qwen-32B},
    url={https://huggingface.co/elyza/ELYZA-Thinking-1.0-Qwen-32B},
    author={Masato Hirakawa and Tomoaki Nakamura and Akira Sasaki and Daisuke Oba and Shoetsu Sato},
    year={2025},
}

參考文獻

@misc{qwen2.5,
    title = {Qwen2.5: A Party of Foundation Models},
    url = {https://qwenlm.github.io/blog/qwen2.5/},
    author = {Qwen Team},
    month = {September},
    year = {2024}
}

@article{qwen2,
      title={Qwen2 Technical Report},
      author={An Yang and Baosong Yang and Binyuan Hui and Bo Zheng and Bowen Yu and Chang Zhou and Chengpeng Li and Chengyuan Li and Dayiheng Liu and Fei Huang and Guanting Dong and Haoran Wei and Huan Lin and Jialong Tang and Jialin Wang and Jian Yang and Jianhong Tu and Jianwei Zhang and Jianxin Ma and Jin Xu and Jingren Zhou and Jinze Bai and Jinzheng He and Junyang Lin and Kai Dang and Keming Lu and Keqin Chen and Kexin Yang and Mei Li and Mingfeng Xue and Na Ni and Pei Zhang and Peng Wang and Ru Peng and Rui Men and Ruize Gao and Runji Lin and Shijie Wang and Shuai Bai and Sinan Tan and Tianhang Zhu and Tianhao Li and Tianyu Liu and Wenbin Ge and Xiaodong Deng and Xiaohuan Zhou and Xingzhang Ren and Xinyu Zhang and Xipin Wei and Xuancheng Ren and Yang Fan and Yang Yao and Yichang Zhang and Yu Wan and Yunfei Chu and Yuqiong Liu and Zeyu Cui and Zhenru Zhang and Zhihao Fan},
      journal={arXiv preprint arXiv:2407.10671},
      year={2024}
}

📄 許可證

本項目採用 apache-2.0 許可證。

📦 模型信息