@inproceedings{gao-etal-2026-thinking,
    title = "Thinking Economically: A Hierarchical Framework for Adaptive-Complexity Reasoning in {LLM}s",
    author = "Gao, Yubo  and
      Wu, Haotian  and
      Chen, Hong  and
      Huang, Junquan  and
      Yan, Yibo  and
      Li, Jungang  and
      Dongfang, Zihao  and
      Tao, Sicheng  and
      Tan, PS  and
      Zhang, Jie  and
      Hu, Xuming",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1965/",
    pages = "39434--39444",
    ISBN = "979-8-89176-395-1",
    abstract = "Chain-of-Thought (CoT) has significantly enhanced LLM reasoning, yet often incurs substantial computational overhead due to ``overthinking'': generating excessively long rationales without commensurate accuracy gains. Existing efficiency methods typically apply uniform compression, which overlooks a critical observation that reasoning complexity is heterogeneous at two distinct granularities: across different problems and within individual reasoning steps. This motivates our principle of \textbf{Thinking Economically}: intelligently allocating computational resources based on intrinsic task and step demands rather than pursuing uniform brevity. We propose Hierarchical Adaptive Budgeter (HAB), a training framework that operationalizes this principle through coarse-to-fine budgeting. At the inter-step level, HAB predicts the optimal reasoning depth for each problem. At the intra-step level, HAB learns step-specific token budgeting signals from PPL-derived step comparisons and an adaptive Pareto optimization objective that captures the local quality-efficiency trade-off, while a Fisher Information-based pruner further provides fine-grained training-time guidance, thereby encouraging the generator to internalize more economical reasoning patterns. Experiments on GSM8K and MATH500 show that HAB not only surpasses standard CoT in accuracy but also reduces token usage, achieving a stronger performance-efficiency trade-off than the compared baselines."
}