Skip to content
AI News HubLIVE
Original source2 min read

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

Summary

A new arXiv paper, SCLoRA, analyzes network parameters via SVD and finds that large-singular-value principal components in pre-trained weights can be reused, while small minor components are task-specific and need adaptation. It theoretically links uncontrolled singular-value growth in LoRA adapters to catastrophic forgetting, and proposes injecting parameterized singular components with spectral clipping. Experiments show better downstream performance with less forgetting of pre-trained knowledge.

SourcearXiv Computational LinguisticsAuthor: Hyowon Wi, Noseong Park
Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 2 Jun 2026]

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

View a PDF of the paper titled Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?, by Hyowon Wi and 1 other authors

View PDF

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.

Comments: ACL 2026 Main Conference

Subjects:

Computation and Language (cs.CL); Machine Learning (cs.LG)

Cite as: arXiv:2608.12332 [cs.CL]

(or arXiv:2608.12332v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2608.12332

arXiv-issued DOI via DataCite

Submission history

From: Hyowon Wi [view email] [v1] Tue, 2 Jun 2026 12:49:42 UTC (941 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Can Spectral-Clipping Enable Better Learning While Forgetting Less for Low-Rank Adaptation?, by Hyowon Wi and 1 other authors

View PDF

TeX Source

view license

Current browse context:

cs.CL

new | recent | 2026-08

Change to browse by:

cs cs.LG

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • Principal singular components with large singular values in pre-trained weights can be effectively reused during fine-tuning; minor components are more task-specific.
  • The paper establishes a theoretical connection between uncontrolled singular-value growth in LoRA adapters and catastrophic forgetting.
  • SCLoRA injects parameterized singular components with spectral clipping in a way aware of the pre-trained model's spectral distribution.
  • Extensive experiments show improved downstream performance while retaining pre-trained knowledge; accepted to ACL 2026 Main Conference.

Highlights and analysis are generated automatically and may contain errors. Check the original source.