Skip to content

Official code for ICML 2024 paper "Learning to Continually Learn with the Bayesian Principle"

Notifications You must be signed in to change notification settings

soochan-lee/SB-MCL

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Learning to Continually Learn with the Bayesian Principle

This repository contains the code for our ICML 2024 paper titled Learning to Continually Learn with the Bayesian Principle.

Requirements

  • Python 3.10
  • Pip packages:
pip install -r requirements.txt

Usage

The basic usage of the training script is as follows:

python train.py -c [config] -o [override options] -l [log directory]

In cfg/, we provide the configuration files for all the experiments in the paper.

After training, we evaluate the models using the following command:

python evaluate.py -l [log directory]

The SB-MCL (MAP) scores can be attained by turning on the map option.

python evaluate.py -l [SB-MCL log directory] -o "map=True"

Datasets

All datasets except MS-Celeb-1M are downloaded automatically by the code. Note that downloading the CASIA dataset may take days.

MS-Celeb-1M

Use BitTorrent to download the dataset from Academic Torrents.

transmission-cli https://academictorrents.com/download/9e67eb7cc23c9417f39778a8e06cca5e26196a97.torrent -w data

About

Official code for ICML 2024 paper "Learning to Continually Learn with the Bayesian Principle"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages