Mosaic: Composite projection pruning for resource-efficient LLMs
Year: 2025
Type: article
Abstract: Extensive compute and memory requirements limit the deployment of large language models (LLMs) on any hardware. Compression methods, such as pruning, can reduce model size, which in turn reduces resource requirements. State-of-the-art pruning is based on coarse-grained methods. They are time-consuming and inherently remove critical model parameters, adversely impacting the quality of the pruned model. This paper introduces projection pruning, a novel fine-grained method for pruning LLMs. In addition, LLM projection pruning is enhanced by a new approach we refer to as composite projection pruning — the synergistic combination of unstructured pruning that retains accuracy and structured pruning that reduces model size. We develop Mosaic , a novel system to create and deploy pruned LLMs using composite projection pruning. Mosaic is evaluated using a range of performance and quality metrics on multiple hardware platforms, LLMs, and datasets. Mosaic is 7.19 × faster in producing models than existing approaches. Mosaic models achieve up to 84.2% lower perplexity and 31.4% higher accuracy than models obtained from coarse-grained pruning. Up to 67% faster inference and 68% lower GPU memory use is noted for Mosaic models. • LLMs are difficult to optimize without hardware and software accelerators. • Pruning makes LLMs smaller and faster for Edge devices. • Existing pruning methods rely on accelerators and reduce LLM quality. • Mosaic uses composite pruning to make LLMs resource-efficient without accelerators. • Mosaic has better accuracy and perplexity than existing pruning methods.
Authors: Bailey J. Eccles, Leon Wong, Blesson Varghese
Institutions: University of St Andrews, Rakuten (Japan)
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,000
OpenAlex ID: W4413216838