Doing More With Less: Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling

TL;DR

This paper demonstrates that unstructured pruning can enhance test-time scaling performance of reasoning LLMs, challenging the belief that pruning always degrades model effectiveness.

Contribution

It provides extensive empirical evidence that unstructured pruning can outperform full models in reasoning tasks and explores layer-wise sparsity strategies.

Findings

Unstructured pruning improves TTS performance across multiple benchmarks.

Unstructured pruning can outperform unpruned full-weight LLMs.

Layer-wise sparsity strategies significantly impact pruning effectiveness.

Abstract

While current Large Language Models (LLMs) exhibit remarkable reasoning capabilities through test-time compute scaling (TTS), their massive parameter counts and high inference costs have motivated the development of pruning methods that can reduce model size without sacrificing performance. However, specific to reasoning LLMs, prior work has shown that structured pruning (methods which removes entire set of layer blocks), significantly degrades TTS reasoning performance. In this work, we revisit this assumption and instead investigate whether unstructured pruning (methods that carefully remove only certain redundant/detrimental weights) exhibits similar limitations. Surprisingly, our extensive experiments across four reasoning benchmarks on two reasoning LLMs: s1.1-7B and Qwen3-8B, consistently show that unstructured pruning augments TTS performance compared to structured pruning, and at times can even outperform the unpruned full-weight LLMs. Furthermore, we also empirically study the impact of different layer-wise sparsity allocation strategies, which are an important parametric choice for instantiating unstructured pruning methods. These findings challenge the conventional notion that pruning always reduces TTS performance and in fact, suggest that carefully undertaken pruning can improve TTS effectiveness even further.