发表机构
KAIST School of Computing; NYU Courant Institute School of Mathematics, Computing, and Data Science（韩国科学技术院计算学院; 纽约大学库朗数学、计算与数据科学研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
HyperThink通过轻量超网络将长思维轨迹摊销为单次参数更新，消除测试时推理开销，以更少令牌保持推理性能，并显著提升低延迟区域的准确率-延迟权衡。
AI 中文摘要
长形式思维轨迹能够显著提升大型语言模型（LLMs）在多步推理任务中的性能，但会引入高昂的推理时开销，其中延迟主要由顺序解码主导。我们提出HyperThink，一种文本到参数的方法，将这种推理计算摊销为一次基于查询条件的参数更新：一个轻量级超网络读取问题，并预测对基础LLM参数子集的更新，同时一个向量量化解码器将这些更新约束到一组有限的可复用模式中，以提升鲁棒性和迁移性。HyperThink在基础模型自身的输出上进行端到端训练，在测试时消除了长思维轨迹：经过一次超网络前向传播后，适配模型无需中间轨迹即可生成简洁的逐步解决方案和最终答案，使用远少于原有的令牌数量，同时保持强大的推理性能。实验上，HyperThink在数学和通用推理任务上改善了准确率-延迟权衡的低延迟区域，其最强增益出现在接近非思维（near-non-thinking）的机制中。
英文摘要
Long-form thinking traces can substantially improve the multi-step reasoning performance of large language models (LLMs), but they introduce high inference-time overhead, with latency dominated by sequential decoding. We propose HyperThink, a text-to-parameter approach that amortizes this reasoning computation into a single query-conditioned parameter update: a lightweight hypernetwork reads the question and predicts updates to a small subset of the base LLM's parameters, while a vector-quantized decoder constrains them to a finite set of reusable patterns to improve robustness and transfer. Trained end-to-end on outputs from the base model itself, HyperThink eliminates long thinking traces at test time: after one hypernetwork forward pass, the adapted model generates a concise step-by-step solution and final answer without an intermediate trace, using far fewer tokens while retaining strong reasoning performance. Empirically, HyperThink improves the low-latency region of the accuracy-latency trade-off on mathematical and general reasoning tasks, with its strongest gains in the near-non-thinking regime.
CommentsCOLM 2026
