Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Xiangyu, Liu, Jing, Wang, Ye, Brand, Matthew, Pu, Wang, Koike-Akino, Toshiaki
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2505.21835
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912399384117248
author Chen, Xiangyu
Liu, Jing
Wang, Ye
Brand, Matthew
Pu
Wang
Koike-Akino, Toshiaki
author_facet Chen, Xiangyu
Liu, Jing
Wang, Ye
Brand, Matthew
Pu
Wang
Koike-Akino, Toshiaki
contents To reduce model size during post-training, compression methods, including knowledge distillation, low-rank approximation, and pruning, are often applied after fine-tuning the model. However, sequential fine-tuning and compression sacrifices performance, while creating a larger than necessary model as an intermediate step. In this work, we aim to reduce this gap, by directly constructing a smaller model while guided by the downstream task. We propose to jointly fine-tune and compress the model by gradually distilling it to a pruned low-rank structure. Experiments demonstrate that joint fine-tuning and compression significantly outperforms other sequential compression methods.
format Preprint
id arxiv_https___arxiv_org_abs_2505_21835
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TuneComp: Joint Fine-tuning and Compression for Large Foundation Models
Chen, Xiangyu
Liu, Jing
Wang, Ye
Brand, Matthew
Pu
Wang
Koike-Akino, Toshiaki
Machine Learning
Artificial Intelligence
To reduce model size during post-training, compression methods, including knowledge distillation, low-rank approximation, and pruning, are often applied after fine-tuning the model. However, sequential fine-tuning and compression sacrifices performance, while creating a larger than necessary model as an intermediate step. In this work, we aim to reduce this gap, by directly constructing a smaller model while guided by the downstream task. We propose to jointly fine-tune and compress the model by gradually distilling it to a pruned low-rank structure. Experiments demonstrate that joint fine-tuning and compression significantly outperforms other sequential compression methods.
title TuneComp: Joint Fine-tuning and Compression for Large Foundation Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2505.21835