# Task-Aware Spectral Pruning: A Mixture-of-Masks Framework for Efficient LLM Inference

arXiv: https://arxiv.org/abs/2609.29499 · HTML: https://mlsanity.com/paper/2609.29499/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Early exit and edge inference, one of 12 directions in Efficient LLM inference and serving.
- Builds on 23 of its 40 references that are in the map.
- First posted on arXiv 24 Aug 2026.

## Placed in

- Efficient LLM inference and serving / [Early exit and edge inference](https://mlsanity.com/inference/early-exit-and-edge-inference/)

## Builds on (most cited first, 23 of 23)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [PaLM: Scaling Language Modeling with Pathways](https://mlsanity.com/paper/2204.02311/) (2022)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [Learning both Weights and Connections for Efficient Neural Networks](https://mlsanity.com/paper/1506.02626/) (2015)
- [MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark](https://mlsanity.com/paper/2406.01574/) (2024)
- [GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding](https://mlsanity.com/paper/2006.16668/) (2020)
- [SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot](https://mlsanity.com/paper/2301.00774/) (2023)
- [Language Models are Multilingual Chain-of-Thought Reasoners](https://mlsanity.com/paper/2210.03057/) (2022)
- [Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruning](https://mlsanity.com/paper/2310.06694/) (2023)
- [Movement Pruning: Adaptive Sparsity by Fine-Tuning](https://mlsanity.com/paper/2005.07683/) (2020)
- [SliceGPT: Compress Large Language Models by Deleting Rows and Columns](https://mlsanity.com/paper/2401.15024/) (2024)
- [ShortGPT: Layers in Large Language Models are More Redundant Than You Expect](https://mlsanity.com/paper/2403.03853/) (2024)
- [DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference](https://mlsanity.com/paper/2004.12993/) (2020)
- [Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learning](https://mlsanity.com/paper/1810.01075/) (2018)
- [Rethinking generalization requires revisiting old ideas: statistical mechanics approaches and complex learning behavior](https://mlsanity.com/paper/1710.09553/) (2017)
- [IG-Pruning: Input-Guided Block Pruning for Large Language Models](https://mlsanity.com/paper/2511.02213/) (2025)
- [LLMs can Compress LLMs: Adaptive Pruning by Agents](https://mlsanity.com/paper/2601.09694/) (2026)

## Abstract (arXiv)

Static pruning imposes one sparse structure on every prompt, even though reasoning, retrieval, generation, coding, and translation can depend on different parts of a language model. We introduce Task-Aware Spectral Pruning (TASP), a post-training framework that calibrates module-level spectral descriptors against measured task-specific ablation effects, closes grouped-query-attention and SwiGLU dependencies during sparse-mask construction, and routes each user turn to one compiled mask that remains fixed throughout prefill and decoding. A module-disjoint pilot first determines whether the spectral signal is informative before full calibration. Under the stated retrospective operating rule, the pilot passes on the evaluated Llama-3-8B and Llama-3-70B checkpoints but rejects Qwen2.5-1.5B, demonstrating that applicability is model-dependent rather than universal. At a 43% active-FLOP reduction, the Llama-3-70B benchmark harness retains 97.7 +/- 0.2% of the dense BF16 score. In the deployment-matched INT8-weight/BF16-compute runtime on a single A100 80GB, the compiled sparse path retains 97.3 +/- 0.2% relative to dense BF16 and reduces decode latency from 45.2 +/- 0.4 to 31.3 +/- 0.4 ms/token, yielding a 1.44x speedup. Factorized ablations, disjoint-module tests, compiled structured baselines, routing-corruption studies, and an explicit 136-GPU-hour calibration audit further delimit the source and operating regime of these gains