@inproceedings{li-etal-2026-long,
    title = "Too Long, Do Re-weighting for Efficient {LLM} Reasoning Compression",
    author = "Li, Zhong-Zhi  and
      Liang, Xiao  and
      Tang, Zihao  and
      Ji, Lei  and
      Wang, Peijie  and
      Xu, Haotian  and
      W, Xing  and
      Huang, Haizhen  and
      Deng, Weiwei  and
      Gong, Yeyun  and
      Guo, Zhijiang  and
      Liu, Xiao  and
      Yin, Fei  and
      Liu, Cheng-Lin",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.1856/",
    pages = "39943--39962",
    ISBN = "979-8-89176-390-6",
    abstract = "Large Language Models (LLMs) have recently achieved remarkable progress on complex reasoning tasks by leveraging extended Chain-of-Thought (CoT) techniques. These reasoning processes can be roughly categorized into System-1 (fast and intuitive) and System-2 (slow and deliberate) paradigms. However, excessive reliance on lengthy System-2-style reasoning during inference can produce extremely long outputs, thereby reducing efficiency. In this work, we propose Thinking Length Data Re-weighting (TLDR), that does not rely on sophisticated data annotations or interpolation between multiple models. We continuously balance the weights between the model{'}s System-1 and System-2 data to eliminate redundant reasoning processes while preserving the model{'}s reasoning capability. We validate our method across multiple base models, including Deepseek-R1-Distilled Qwen models, as well as on a diverse benchmarks with varying difficulty levels. Our method significantly reduces the number of output tokens by nearly 40{\%} while maintaining the accuracy of the reasoning."
}