Pruning Large Language Models by Identifying and Preserving Functional Networks
This work addresses the need for more efficient LLMs in real-world applications by improving pruning methods to reduce GPU memory and accelerate inference, though it appears incremental as it builds on existing structured pruning techniques.
The paper tackles the problem of structured pruning in large language models (LLMs) by identifying and preserving functional networks to avoid disrupting their macro architecture, resulting in efficient model pruning with successful identification of key neurons.
Structured pruning is one of the representative techniques for compressing large language models (LLMs) to reduce GPU memory consumption and accelerate inference speed. It offers significant practical value in improving the efficiency of LLMs in real-world applications. Current structured pruning methods typically rely on assessment of the importance of the structure units and pruning the units with less importance. Most of them overlooks the interaction and collaboration among artificial neurons that are crucial for the functionalities of LLMs, leading to a disruption in the macro functional architecture of LLMs and consequently a pruning performance degradation. Inspired by the inherent similarities between artificial neural networks and functional neural networks in the human brain, we alleviate this challenge and propose to prune LLMs by identifying and preserving functional networks within LLMs in this study. To achieve this, we treat an LLM as a digital brain and decompose the LLM into functional networks, analogous to identifying functional brain networks in neuroimaging data. Afterwards, an LLM is pruned by preserving the key neurons within these functional networks. Experimental results demonstrate that the proposed method can successfully identify and locate functional networks and key neurons in LLMs, enabling efficient model pruning. Our code is available at https://github.com/WhatAboutMyStar/LLM_ACTIVATION.