Fine Tuning without Catastrophic Forgetting via Selective Low Rank Adaptation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bafghi, Reza Akbarian, Bagwell, Carden, Ravichandran, Avinash, Shrivastava, Ashish, Raissi, Maziar
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912205288505344
author Bafghi, Reza Akbarian
Bagwell, Carden
Ravichandran, Avinash
Shrivastava, Ashish
Raissi, Maziar
author_facet Bafghi, Reza Akbarian
Bagwell, Carden
Ravichandran, Avinash
Shrivastava, Ashish
Raissi, Maziar
contents Adapting deep learning models to new domains often requires computationally intensive retraining and risks catastrophic forgetting. While fine-tuning enables domain-specific adaptation, it can reduce robustness to distribution shifts, impacting out-of-distribution (OOD) performance. Pre-trained zero-shot models like CLIP offer strong generalization but may suffer degraded robustness after fine-tuning. Building on Task Adaptive Parameter Sharing (TAPS), we propose a simple yet effective extension as a parameter-efficient fine-tuning (PEFT) method, using an indicator function to selectively activate Low-Rank Adaptation (LoRA) blocks. Our approach minimizes knowledge loss, retains its generalization strengths under domain shifts, and significantly reduces computational costs compared to traditional fine-tuning. We demonstrate that effective fine-tuning can be achieved with as few as 5\% of active blocks, substantially improving efficiency. Evaluations on pre-trained models such as CLIP and DINO-ViT demonstrate our method's broad applicability and effectiveness in maintaining performance and knowledge retention.
format Preprint
id arxiv_https___arxiv_org_abs_2501_15377
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fine Tuning without Catastrophic Forgetting via Selective Low Rank Adaptation
Bafghi, Reza Akbarian
Bagwell, Carden
Ravichandran, Avinash
Shrivastava, Ashish
Raissi, Maziar
Computer Vision and Pattern Recognition
Adapting deep learning models to new domains often requires computationally intensive retraining and risks catastrophic forgetting. While fine-tuning enables domain-specific adaptation, it can reduce robustness to distribution shifts, impacting out-of-distribution (OOD) performance. Pre-trained zero-shot models like CLIP offer strong generalization but may suffer degraded robustness after fine-tuning. Building on Task Adaptive Parameter Sharing (TAPS), we propose a simple yet effective extension as a parameter-efficient fine-tuning (PEFT) method, using an indicator function to selectively activate Low-Rank Adaptation (LoRA) blocks. Our approach minimizes knowledge loss, retains its generalization strengths under domain shifts, and significantly reduces computational costs compared to traditional fine-tuning. We demonstrate that effective fine-tuning can be achieved with as few as 5\% of active blocks, substantially improving efficiency. Evaluations on pre-trained models such as CLIP and DINO-ViT demonstrate our method's broad applicability and effectiveness in maintaining performance and knowledge retention.
title Fine Tuning without Catastrophic Forgetting via Selective Low Rank Adaptation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2501.15377