Code for the image generation experiments in Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow.
Code for the iRL experiemnts: https://github.com/qxcv/vdb-irl/
@inproceedings{
VDBPeng18,
title={Variational Discriminator Bottleneck: Improving Imitation Learning,
Inverse RL, and GANs by Constraining Information Flow},
author = {Peng, Xue Bin and Kanazawa, Angjoo and Toyer, Sam and Abbeel, Pieter
and Levine, Sergey},
booktitle={ICLR},
year={2019}
}
Our code is built on the GAN implmentation of Which Training Methods for GANs do actually Converge? [Mescheder et al. ICML 2018]. This repo adds the VGAN and instance noise implementations, along with FID computation.
First download your data and put it into the ./data
folder.
To train a new model, first create a config script similar to the ones provided in the ./configs
folder. You can then train you model using
python train.py PATH_TO_CONFIG
You can monitor the training with tensorboard:
tensorboard --logdir output/<MODEL_NAME>/monitoring/
To generate samples, use
python test.py PATH_TO_CONIFG
You can also create latent space interpolations using
python interpolate.py PATH_TO_CONFIG
A pre-trained model for CelebA-HQ can be found here