Still-Rise / continual_learning_PST

Single-Net Continual Learning with Progressive Segmented Training (PST)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Enviroments:

Pytorch: 1.0.1_post2

Python 3.6

CUDA 9.0+

For details, see requirements.txt

To run CIFAR-100 experiment in 'Experiment' section, which is incrementally learning 5/10/20/50 tasks, perform:

python PST_main.py --classes_per_task 5/10/20/50

Alternatively, you can change the hyper-parameters by:

python PST_main.py --classes_per_task 5 --memory_size 5000 --NA_C0 30

About

Single-Net Continual Learning with Progressive Segmented Training (PST)


Languages

Language:Python 100.0%