https://github.com/gmvandeven/incremental_cifar100
Implementation of BI-R in T&E framework.
https://github.com/gmvandeven/incremental_cifar100
Last synced: 3 months ago
JSON representation
Implementation of BI-R in T&E framework.
- Host: GitHub
- URL: https://github.com/gmvandeven/incremental_cifar100
- Owner: GMvandeVen
- Created: 2020-03-11T12:39:17.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-03-11T15:41:08.000Z (over 5 years ago)
- Last Synced: 2025-01-11T14:22:19.530Z (9 months ago)
- Language: Python
- Size: 1.4 MB
- Stars: 1
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Class-incremental learning with Brain-Inspired Replay in the T&E framework
Besides Brain-Inspired Replay (BI-R), we have currently implemented simple fine-tuning (None), Synaptic Intelligence (SI) and Learning without Forgetting (LwF) as baseline methods.
To run each of these methods, use the following:
- None: ```python train_cifar100_incremental```
- LwF: ```python train_cifar100_incremental --lwf```
- SI: ```python train_cifar100_incremental --si```
- BI-R: ```python train_cifar100_incremental --bir```
- BI-R + SI: ```python train_cifar100_incremental --bir --si```
Log-files for evaluating performance should be produced by running any of the above commands, but we are still working on evaluating these log-files using the l2metrics-environment.