Skip to content

Latest commit

 

History

History
37 lines (31 loc) · 825 Bytes

README.md

File metadata and controls

37 lines (31 loc) · 825 Bytes
Raw

Deep Variational Information Bottleneck


Overview

Pytorch implementation of Deep Variational Information Bottleneck(paper, original code)

ELBO monte_carlo

Dependencies

python 3.6.4
pytorch 0.3.1.post2
tensorboardX(optional)
tensorflow(optional)

Usage

  1. train
python main.py --mode train --beta 1e-3 --tensorboard True --env_name [NAME]
  1. test
python main.py --mode test --env_name [NAME] --load_ckpt best_acc.tar

References

  1. Deep Learning and the Information Bottleneck Principle, Tishby et al.
  2. Deep Variational Information Bottleneck, Alemi et al.
  3. Tensorflow Demo : https://github.com/alexalemi/vib_demo