LIMOPro: Reasoning Refinement for Efficient and Effective Test-time Scaling

TL;DR

LIMOPro introduces PIR, a framework that refines reasoning chains by pruning low-importance steps, resulting in more concise, accurate, and computationally efficient large language model inferences.

Contribution

It presents PIR, a novel importance evaluation method that selectively prunes reasoning steps to optimize test-time scaling without sacrificing core reasoning quality.

Findings

Models fine-tuned on PIR data show 0.9% to 6.6% accuracy improvements.

Reasoning chains are reduced by 3% to 41% in token usage.

Enhanced efficiency and accuracy across multiple reasoning benchmarks.

Abstract

Large language models (LLMs) have demonstrated remarkable reasoning capabilities through test-time scaling approaches, particularly when fine-tuned with chain-of-thought (CoT) data distilled from more powerful large reasoning models (LRMs). However, these reasoning chains often contain verbose elements that mirror human problem-solving, categorized as progressive reasoning (the essential solution development path) and functional elements (verification processes, alternative solution approaches, and error corrections). While progressive reasoning is crucial, the functional elements significantly increase computational demands during test-time inference. We introduce PIR (Perplexity-based Importance Refinement), a principled framework that quantitatively evaluates the importance of each reasoning step based on its impact on answer prediction confidence. PIR systematically identifies and selectively prunes only low-importance functional steps while preserving progressive reasoning components, creating optimized training data that maintains the integrity of the core solution path while reducing verbosity. Models fine-tuned on PIR-optimized data exhibit superior test-time scaling properties, generating more concise reasoning chains while achieving improved accuracy (+0.9\% to +6.6\%) with significantly reduced token usage (-3\% to -41\%) across challenging reasoning benchmarks (AIME, AMC, and GPQA Diamond). Our approach demonstrates strong generalizability across different model sizes, data sources, and token budgets, offering a practical solution for deploying reasoning-capable LLMs in scenarios where efficient test-time scaling, response time, and computational efficiency are valuable constraints.