Sparse-BitNet: 1.58-bit LLMs are Naturally Friendly to Semi-Structured Sparsity
This work provides a method for improving the efficiency of LLMs for practitioners and researchers by combining two previously isolated optimization techniques.
This paper explores the combination of 1.58-bit quantization and N:M semi-structured sparsity for Large Language Models (LLMs), finding that 1.58-bit BitNet is inherently more compatible with N:M sparsity than full-precision models. The proposed Sparse-BitNet framework achieves up to 1.30X speedup in training and inference by integrating these techniques.
Semi-structured N:M sparsity and low-bit quantization (e.g., 1.58-bit BitNet) are two promising approaches for improving the efficiency of large language models (LLMs), yet they have largely been studied in isolation. In this work, we investigate their interaction and show that 1.58-bit BitNet is naturally more compatible with N:M sparsity than full-precision models. To study this effect, we propose Sparse-BitNet, a unified framework that jointly applies 1.58-bit quantization and dynamic N:M sparsification while ensuring stable training for the first time. Across multiple model scales and training regimes (sparse pretraining and dense-to-sparse schedules), 1.58-bit BitNet consistently exhibits smaller performance degradation than full-precision baselines at the same sparsity levels and can tolerate higher structured sparsity before accuracy collapse. Moreover, using our custom sparse tensor core, Sparse-BitNet achieves substantial speedups in both training and inference, reaching up to 1.30X. These results highlight that combining extremely low-bit quantization with semi-structured N:M sparsity is a promising direction for efficient LLMs. Code available at https://github.com/AAzdi/Sparse-BitNet