Saved in:
Bibliographic Details
Main Authors: Williams, Miles, Aletras, Nikolaos
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2309.08708
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912291227697152
author Williams, Miles
Aletras, Nikolaos
author_facet Williams, Miles
Aletras, Nikolaos
contents The extensive memory footprint of language model (LM) fine-tuning poses a challenge for both researchers and practitioners. LMs use an embedding matrix to represent extensive vocabularies, forming a substantial proportion of the model parameters. While previous work towards memory-efficient fine-tuning has focused on minimizing the number of trainable parameters, reducing the memory footprint of the embedding matrix has yet to be explored. We first demonstrate that a significant proportion of the vocabulary remains unused during fine-tuning. We then propose a simple yet effective approach that leverages this finding to minimize memory usage. We show that our approach provides substantial reductions in memory usage across a wide range of models and tasks. Notably, our approach does not impact downstream task performance, while allowing more efficient use of computational resources.
format Preprint
id arxiv_https___arxiv_org_abs_2309_08708
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Vocabulary-level Memory Efficiency for Language Model Fine-tuning
Williams, Miles
Aletras, Nikolaos
Computation and Language
The extensive memory footprint of language model (LM) fine-tuning poses a challenge for both researchers and practitioners. LMs use an embedding matrix to represent extensive vocabularies, forming a substantial proportion of the model parameters. While previous work towards memory-efficient fine-tuning has focused on minimizing the number of trainable parameters, reducing the memory footprint of the embedding matrix has yet to be explored. We first demonstrate that a significant proportion of the vocabulary remains unused during fine-tuning. We then propose a simple yet effective approach that leverages this finding to minimize memory usage. We show that our approach provides substantial reductions in memory usage across a wide range of models and tasks. Notably, our approach does not impact downstream task performance, while allowing more efficient use of computational resources.
title Vocabulary-level Memory Efficiency for Language Model Fine-tuning
topic Computation and Language
url https://arxiv.org/abs/2309.08708