Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note2 min read

SLORR: Simple and Efficient In-Training Low-Rank Regularization

Source

arXiv

arxiv.org/abs/2607.08754v1

Read the source

Source

Plain-Language Brief

The paper introduces SLORR, a simple and efficient method for in-training low-rank regularization of neural networks. It addresses the limitations of existing methods by being stateless and preserving the original architecture, using GPU-friendly approximations. The authors evaluate SLORR on various models and tasks, showing that it improves compressibility with minimal training overhead.

What the Document Says

  • SLORR is a new framework for in-training low-rank regularization.
  • It has two main variants based on the Hoyer sparsity metric and nuclear norm.
  • SLORR directly regularizes original weight matrices using GPU-friendly approximations.
  • It improves compressibility while introducing less than 8% training overhead in ImageNet-1K experiments.
  • It preserves performance better than unregularized models with minimal additional training overhead in LLM pretraining at 135M and 560M scales.

Key Concepts Mentioned

  • SLORR
  • Low-rank regularization
  • Neural networks
  • ImageNet-1K
  • ResNet
  • ViT
  • LLM

Claims Made by the Document

  • SLORR improves compressibility.
  • It introduces minimal training overhead.

Evidence Used

  • Results from ImageNet-1K and LLM pretraining experiments provided in the paper.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Low-rank factorization is widely used to compress neural networks, but modern models are often not naturally amenable to aggressive factorization without significant accuracy loss. Existing training-time low-rank regularizers can improve compressibility, but they often require SVDs of large weight matrices, modify the model architecture (introducing additional trainable parameters), or rely on stateful cached quantities. To address these limitations, we introduce SLORR, a simple, stateless, and architecture-preserving framework for in-training low-rank regularization, instantiated with two main variants based on the Hoyer sparsity metric and the nuclear norm. SLORR directly regularizes the original weight matrices using GPU-friendly approximations for the forward and backward passes of the regularizers, for which we provide approximation guarantees. We first evaluate SLORR on ImageNet-1K across short-horizon continued training of ResNet-50, ViT-B/16, and ViT-L/16, and pretraining of ResNet-18, where SLORR induces compressibility while introducing less than 8% training overhead. We further evaluate SLORR-Hoyer in LLM pretraining at 135M and 560M scales: SLORR-trained compressed models preserve performance substantially better than unregularized models while adding less than 1% average training overhead.