Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?

arXiv cs.CL Papers

Summary

This paper investigates singular components in LoRA and proposes SCLoRA, a method that uses spectral clipping to improve task adaptation while reducing catastrophic forgetting of pre-trained knowledge.

arXiv:2608.12332v1 Announce Type: new Abstract: In recent years, low-rank adaptation (LoRA) has emerged as a significant paradigm that freezes pre-trained weights and introduces small, learnable adapters instead of fine-tuning the full set of parameters. In this work, we uncover several key insights regarding the singular components of network parameters based on Singular Value Decomposition (SVD). Firstly, the principal singular components with large singular values in pre-trained network parameters can be effectively reused during fine-tuning, whereas the minor components with smaller singular values are more task-specific and require substantial adaptation. Secondly, we first establish the theoretical connection that the uncontrolled growth of singular values in LoRA adapters leads to the forgetting of pre-trained knowledge -- a well-known issue referred to as catastrophic forgetting. Building on these observations, we propose SCLoRA, which injects parameterized singular components with spectral clipping into the pre-trained model in a way that is aware of the spectral distribution of the pre-trained model. SCLoRA effectively adapts to new tasks by focusing updates on components that require adaptation, while simultaneously alleviating catastrophic forgetting. We conduct extensive experiments and demonstrate that SCLoRA not only improves downstream performance but also effectively retains pre-trained knowledge.
Original Article
View Cached Full Text

Cached at: 08/14/26, 09:24 AM

# Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?
Source: [https://arxiv.org/abs/2608.12332](https://arxiv.org/abs/2608.12332)
[View PDF](https://arxiv.org/pdf/2608.12332)

> Abstract:In recent years, low\-rank adaptation \(LoRA\) has emerged as a significant paradigm that freezes pre\-trained weights and introduces small, learnable adapters instead of fine\-tuning the full set of parameters\. In this work, we uncover several key insights regarding the singular components of network parameters based on Singular Value Decomposition \(SVD\)\. Firstly, the principal singular components with large singular values in pre\-trained network parameters can be effectively reused during fine\-tuning, whereas the minor components with smaller singular values are more task\-specific and require substantial adaptation\. Secondly, we first establish the theoretical connection that the uncontrolled growth of singular values in LoRA adapters leads to the forgetting of pre\-trained knowledge \-\- a well\-known issue referred to as catastrophic forgetting\. Building on these observations, we propose SCLoRA, which injects parameterized singular components with spectral clipping into the pre\-trained model in a way that is aware of the spectral distribution of the pre\-trained model\. SCLoRA effectively adapts to new tasks by focusing updates on components that require adaptation, while simultaneously alleviating catastrophic forgetting\. We conduct extensive experiments and demonstrate that SCLoRA not only improves downstream performance but also effectively retains pre\-trained knowledge\.

## Submission history

From: Hyowon Wi \[[view email](https://arxiv.org/show-email/e9e722fd/2608.12332)\] **\[v1\]**Tue, 2 Jun 2026 12:49:42 UTC \(941 KB\)

Similar Articles

Beyond LoRA: Is Sparsity-Induced Adaptation Better?

arXiv cs.LG

This paper proposes sparsity-induced adaptations to LoRA, including Cheap LoRA (cLA) and a chained circulant variant (c³LA), and provides theoretical generalization bounds along with empirical evaluations showing up to 10% training time reduction and 15% peak GPU memory savings while maintaining competitive performance.

Hybrid-LoRA: Bridging Full Fine-Tuning and Low-Rank Adaptation for Post-Training

arXiv cs.LG

Hybrid-LoRA proposes a framework that selectively applies full fine-tuning to a small subset of modules while using LoRA for the rest, achieving performance near full fine-tuning with significantly lower computational cost. Experiments show improvements of up to 5.65% over existing parameter-efficient baselines.