Paper Detail

TailLoR: Protecting Principal Components in Parameter-Efficient Continual Learning

Marius Dragoi, Ioana Pintilie, Alexandra Dragomir, Antonio Barbalau, Florin Brad

arxiv Score 10.0

Published 2026-06-04 · First seen 2026-06-05

Research Track A

Abstract

Parameter-efficient finetuning methods based on spectral decomposition have enabled progress in Continual Learning. In this paper we introduce TailLoR, which utilizes the singular bases U and V of the pre-trained weights as a fixed reference frame to learn a low-rank update applied to the singular value matrix. A soft spectral penalty discourages updates aligned with dominant singular directions, reducing interference while routing fine-grained adaptation into the highly flexible, long-tail spectral coordinates.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@article{dragoi2026taillor,
  title = {TailLoR: Protecting Principal Components in Parameter-Efficient Continual Learning},
  author = {Marius Dragoi and Ioana Pintilie and Alexandra Dragomir and Antonio Barbalau and Florin Brad},
  year = {2026},
  abstract = {Parameter-efficient finetuning methods based on spectral decomposition have enabled progress in Continual Learning. In this paper we introduce TailLoR, which utilizes the singular bases U and V of the pre-trained weights as a fixed reference frame to learn a low-rank update applied to the singular value matrix. A soft spectral penalty discourages updates aligned with dominant singular directions, reducing interference while routing fine-grained adaptation into the highly flexible, long-tail spec},
  url = {https://arxiv.org/abs/2606.06494},
  keywords = {cs.LG, Singular value decomposition, Computer science, Frame (networking), Decomposition, Mathematical optimization},
  eprint = {2606.06494},
  archiveprefix = {arXiv},
}

Metadata

{}