Uncategorized

[2310.09499] One-Shot Sensitivity-Aware Mixed Sparsity Pruning for Large Language Models



Download a PDF of the paper titled One-Shot Sensitivity-Aware Mixed Sparsity Pruning for Large Language Models, by Hang Shao and 2 other authors

Download PDF

Abstract:Various Large Language Models(LLMs) from the Generative Pretrained Transformer~(GPT) family have achieved outstanding performances in a wide range of text generation tasks. However, the enormous model sizes have hindered their practical use in real-world applications due to high inference latency. Therefore, improving the efficiencies of LLMs through quantization, pruning, and other means has been a key issue in LLM studies. In this work, we propose a method based on Hessian sensitivity-aware mixed sparsity pruning to prune LLMs to at least 50\% sparsity without the need of any retraining. It allocates sparsity adaptively based on sensitivity, allowing us to reduce pruning-induced error while maintaining the overall sparsity level. The advantages of the proposed method exhibit even more when the sparsity is extremely high. Furthermore, our method is compatible with quantization, enabling further compression of LLMs.

Submission history

From: Shao Hang [view email]
[v1]
Sat, 14 Oct 2023 05:43:09 UTC (241 KB)
[v2]
Tue, 30 Jan 2024 09:04:06 UTC (242 KB)



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *