VPG-EA：变分后验引导提升LLM推理效率，缓解过度思考
Efficient LLM Reasoning via Variational Posterior Guidance with Efficiency Awareness
这篇论文为LLM推理效率问题提供了理论严谨且可落地的解决方案，做推理优化或模型压缩的研究者可以直接参考其变分蒸馏方法，值得细读。
大语言模型在复杂推理中常出现“过度思考”问题，导致推理链过长、效率低下。现有强化学习方法通过设计复杂奖励函数压缩推理链，但高质量样本在探索空间中极为稀疏，形成采样瓶颈。
受认知科学启发，研究者从理论上证明，参考答案引导的后验分布比先验分布具有更高期望效用，可突破高质量样本的采样瓶颈。为此，他们提出VPG-EA框架，将高效推理形式化为变分推断问题，引入效率感知的证据下界作为理论基础。
该框架采用参数共享的双流架构实例化后验分布和先验策略，通过交叉视图评估过滤伪高效路径，再通过变分蒸馏将后验的高效模式单向迁移至先验策略。在DeepSeek-R1-Distill-Qwen-1.
5B和7B规模上的实验显示，VPG-EA在综合效率指标ε³上分别比最强基线提升8.73%和12.
37%。
Efficient LLM Reasoning via Variational Posterior Guidance with Efficiency Awareness
Although large language models rely on chain-of-thought for complex reasoning, the overthinking phenomenon severely degrades inference efficiency. Existing reinforcement learning methods compress reasoning chains by designing elaborate reward functions, which renders high-quality samples extremely sparse in the exploration space and creates a sampling bottleneck for the prior policy. Inspired by cognitive science, we theoretically prove that a posterior distribution guided by reference answers achieves higher expected utility than the prior distribution, thus capable of breaking through the sampling bottleneck of high-quality samples. However, the posterior distribution is unavailable during inference. To this end, we formalize efficient reasoning as a variational inference problem and introduce an efficiency-aware evidence lower bound as the theoretical foundation. Based on this, we propose the VPG-EA framework. It adopts a parameter-shared dual-stream architecture to instantiate both the posterior distribution and the prior policy; after filtering out pseudo-efficient paths via cross-view evaluation, it unidirectionally transfers the posterior's efficient patterns to the prior policy through variational distillation. Experiments on DeepSeek-R1-Distill-Qwen-1.5B and 7B scales demonstrate that VPG-EA improves the comprehensive efficiency metric epsilon cubed by 8.73% and 12.37% over the strongest baselines on each model size, respectively.
This report is categorized under AITOP's "Papers" desk. Key entities and technical topics under continuous tracking include reasoning models, efficiency optimization, variational inference, distillation.
Primary source verification: published directly by arXiv: DeepSeek with first-party authority.
