Skip to content

Latest commit

 

History

History
21 lines (13 loc) · 567 Bytes

README.md

File metadata and controls

21 lines (13 loc) · 567 Bytes
Raw

LOCO

LOCO (Low-rank and Orthogonal Space for Weights Modification)

Code for paper Low-rank and Orthogonal Space for Weights Modification Enhances Learning Efficiency in Neural Networks without Gradient Backpropagation.

Requirements

If the environment configurations are different, the results may vary greatly or even failing to work properly. You may need to adjust the code details according to your needs. Linux: Ubuntu 16.04

  • cuda 11.8

  • Python 3.7.16

  • torch 1.13.1 (pytorch)

  • torchvision 0.14.1

  • numpy 1.21.4

  • scikit-learn 1.0.2

  • scipy 1.7.3