Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Efficient LLM Reasoning via Variational Posterior Guidance with Efficiency Awareness

About

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.

Zizhao Chen, Yuying Li, Siting Lin, Lianxi Wang• 2026

Related benchmarks

TaskDatasetResultRank
General ReasoningGPQA-Diamond & MMLU-Pro
Accuracy43.83
35
General ReasoningGPQA Diamond
Accuracy38.88
19
Mathematical ReasoningGSM8K
Accuracy (ACC)92
14
Mathematical ReasoningMATH 500
Accuracy (ACC)93
14
Mathematical ReasoningAIME24
Accuracy (ACC)56.67
14
General ReasoningMMLU-Pro
ACC48.77
14
Mathematical ReasoningAIME 25
ACC36.67
14
Mathematical ReasoningMATH 500
Accuracy89.6
5
Mathematical ReasoningAIME24
ACC46.7
5
Mathematical ReasoningGSM8K
Accuracy89.16
3
Showing 10 of 10 rows

Other info

Follow for update