Reasoning-Aware Compression: Identifying and Protecting Vulnerable Reasoning Circuits for Energy-Efficient LLM Deployment

Abstract

Large Reasoning Models (LRMs) impose substantial energy costs during deployment, yet current compression methods apply uniform quantization across all components, risking damage to critical reasoning circuits. We present a reasoning-aware compression framework that benchmarks quantization conditions across five reasoning benchmarks, GSM8K, FOLIO, MATH-500, ProofWriter, and MuSiQue, with hardware-level GPU energy measurement; profiles per-module INT4 vulnerability across all 196-224 (layer, proje

Categories

@misc{twagirayezu2026reasoningaware,
  title = {{Reasoning-Aware Compression: Identifying and Protecting Vulnerable Reasoning Circuits for Energy-Efficient LLM Deployment}},
  author = {Leonard Twagirayezu and Prasenjit Mitra},
  year = {2026},
  month = sep,
  eprint = {2609.05512},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.05512}
}