Reasoning-Aware Compression: Identifying and Protecting Vulnerable Reasoning Circuits for Energy-Efficient LLM Deployment
On the map: LLM reasoning → Test-time compute and efficient thinking
also in 1 more
Abstract (arXiv)
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, projection) pairs via a perturbation sweep on a held-out calibration split, then selectively restores the most sensitive circuits to FP16. Three findings emerge. First, INT4 quantization can increase energy by extending reasoning chains; a 25% power reduction becomes a net energy increase on GSM8K. Second, vulnerability is task-dependent: attention projections are more critical for mathematical reasoning, and sensitivity patterns differ by architecture in logical inference. Third, selective compression achieves Pareto-optimal points inaccessible to uniform methods: R1-Qwen-7B Top-10% on ProofWriter gains +12 pp over FP16 at -9.7% energy, validated on held-out data across five reasoning benchmarks.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 8 of its 16 references that are in the map. First posted on arXiv 31 Aug 2026.
Builds on
8 of 16 references.
Closest ancestors · 1
Same direction, newest first
Other references · 7
- 2024The Llama 3 Herd of Modelscited by 14,715
- 2021Training Verifiers to Solve Math Word Problemscited by 10,109
- 2025Qwen3 Technical Reportcited by 8,820
- 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
- 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
- 2021Carbon Emissions and Large Neural Network Trainingcited by 567
- 2025To Think or Not to Think: Exploring the Unthinking Vulnerability in Large Reasoning Modelscited by 22
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026ReSET: Accurate Latency-Critical NVFP4 Reasoning via Step-Aware Temperature Scaling
- 2025Not All Bits Are Equal: Scale-Dependent Memory Optimization Strategies for Reasoning Models
- 2026What Survives When You Compress a Recursive Reasoner for the Edge?
- 2026Memory Augmentation Unlocks Efficient Chain-of-Thought Reasoning
- 2025Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning
- 2025TrimR: Verifier-based Training-Free Thinking Compression for Efficient Test-Time Scaling
- 2026Measuring the Inference Energy Efficiency of Open-Weight LLMs across Benchmarks, Reasoning Modes, and Hardware
- 2026KARA: Efficient Reasoning LLM Serving via Sliding-Window KV Cache Compression