Light-PEFT: Lightening Parameter-Efficient Fine-Tuning via Early Pruning

TL;DR

Light-PEFT introduces a novel early pruning framework that significantly reduces redundant parameters in foundation models and PEFT modules, leading to more efficient fine-tuning without sacrificing performance.

Contribution

The paper proposes the Light-PEFT framework with early pruning methods to improve training efficiency and reduce parameters in large language model fine-tuning.

Findings

Prunes over 40% of foundation model parameters.

Reduces trainable parameters to 25% of traditional PEFT.

Achieves speedup and memory reduction while maintaining performance.

Abstract

Parameter-efficient fine-tuning (PEFT) has emerged as the predominant technique for fine-tuning in the era of large language models. However, existing PEFT methods still have inadequate training efficiency. Firstly, the utilization of large-scale foundation models during the training process is excessively redundant for certain fine-tuning tasks. Secondly, as the model size increases, the growth in trainable parameters of empirically added PEFT modules becomes non-negligible and redundant, leading to inefficiency. To achieve task-specific efficient fine-tuning, we propose the Light-PEFT framework, which includes two methods: Masked Early Pruning of the Foundation Model and Multi-Granularity Early Pruning of PEFT. The Light-PEFT framework allows for the simultaneous estimation of redundant parameters in both the foundation model and PEFT modules during the early stage of training. These parameters can then be pruned for more efficient fine-tuning. We validate our approach on GLUE, SuperGLUE, QA tasks, and various models. With Light-PEFT, parameters of the foundation model can be pruned by up to over 40%, while still controlling trainable parameters to be only 25% of the original PEFT method. Compared to utilizing the PEFT method directly, Light-PEFT achieves training and inference speedup, reduces memory usage, and maintains comparable performance and the plug-and-play feature of PEFT.