# Hierarchical Sparse Plus Low Rank Compression of LLM

arXiv: https://arxiv.org/abs/2601.07839 · HTML: https://mlsanity.com/paper/2601.07839/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 32 of its 75 references that are in the map.
- First posted on arXiv 19 Dec 2025.

## Builds on (most cited first, 30 of 32)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://mlsanity.com/paper/2205.11487/) (2022)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding](https://mlsanity.com/paper/1510.00149/) (2015)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [Linformer: Self-Attention with Linear Complexity](https://mlsanity.com/paper/2006.04768/) (2020)
- [Solving Quantitative Reasoning Problems with Language Models](https://mlsanity.com/paper/2206.14858/) (2022)
- [Pruning Filters for Efficient ConvNets](https://mlsanity.com/paper/1608.08710/) (2016)
- [SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot](https://mlsanity.com/paper/2301.00774/) (2023)
- [KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization](https://mlsanity.com/paper/2401.18079/) (2024)
- [Movement Pruning: Adaptive Sparsity by Fine-Tuning](https://mlsanity.com/paper/2005.07683/) (2020)
- [Parameter Efficient Training of Deep Convolutional Neural Networks by Dynamic Sparse Reparameterization](https://mlsanity.com/paper/1902.05967/) (2019)
- [BitNet: Scaling 1-bit Transformers for Large Language Models](https://mlsanity.com/paper/2310.11453/) (2023)
- [SLTrain: a sparse plus low-rank approach for parameter and memory efficient pretraining](https://mlsanity.com/paper/2406.02214/) (2024)
- [Quantization Meets Reasoning: Exploring LLM Low-Bit Quantization Degradation for Mathematical Reasoning](https://mlsanity.com/paper/2501.03035/) (2025)
- [Sparse Plus Low Rank Matrix Decomposition: A Discrete Optimization Approach](https://mlsanity.com/paper/2109.12701/) (2021)
- [Pruning Adversarially Robust Neural Networks without Adversarial Examples](https://mlsanity.com/paper/2210.04311/) (2022)
- [Blockwise Parallel Transformer for Large Context Models](https://mlsanity.com/paper/2305.19370/) (2023)
- [On the Adversarial Robustness of Quantized Neural Networks](https://mlsanity.com/paper/2105.00227/) (2021)
- [Adversarial Pruning: A Survey and Benchmark of Pruning Methods for Adversarial Robustness](https://mlsanity.com/paper/2409.01249/) (2024)
- [Memory-Efficient Vision Transformers: An Activation-Aware Mixed-Rank Compression Strategy](https://mlsanity.com/paper/2402.06004/) (2024)

## Abstract (arXiv)

Modern large language models (LLMs) place extraordinary pressure on memory and compute budgets, making principled compression indispensable for both deployment and continued training. We present Hierarchical Sparse Plus Low-Rank (HSS) compression, a two-stage scheme that (i) removes the largest-magnitude weights into a sparse matrix S and (ii) applies a recursive Hierarchically Sparse Separable (HSS) low-rank factorisation to the dense residual matrix. A recursive rank-reducing strategy and a reverse Cuthill-Mckee (RCM) permutation are introduced to align high weights towards the diagonal with the block-diagonal hierarchy, maximising off-diagonal compressibility (because they are touched only once). HSS is hardware-friendly: its matrix-vector multiply reduces to one sparse and a sequence of thin-matrix multiplications and can be trained end-to-end with standard optimisers. Experiments on LLaMA-7B show that targeting only the self-attention projections (1.6 B parameters of Q, K, and V matrices out of a total 7B parameters) suffices to yield large memory savings while retaining comparable state-of-the-art perplexity scores on test samples of the WikiText dataset. For example, with a 30% sparsity budget and an outer rank of 512, sHSS-RCM achieves a perplexity of 1.64, outperforming dense baselines and classical sparse-plus-SVD variants, while also achieving significant memory savings.