PyTorch implementation for "Guiding Energy-based Models via Contrastive Latent Variables" (accepted as a Spotlight presentation in ICLR 2023)
TL;DR: A simple yet effective framework for improving energy-based models (EBMs) via contrastive representation learning.
conda create -n ebm python=3.9
conda activate ebm
conda install pytorch torchvision cudatoolkit=11.3 -c pytorch
conda install torchmetrics -c conda-forge
conda install ignite -c pytorch-nightly
pip install omegaconf
pip install torch-fidelity
pip install kornia==0.6.3
pip install tensorboard
pip install sklearnexport CUDA_VISIBLE_DEVICES=0
python train.py configs/cifar10.yamlYou can modify options using YAML config files or key=value command-line arguments. See utils.parse_config() and OmegaConf for details.
python test_fid.py logs/cifar10/resnet_resnet18/ours/config.yaml use_ema=trueThis command saves 50k generated samples into samples.pth in the log directory. You can use this file for official pytorch FID evaluation. Note that the FID value obtained from our code is similar to that from the official evaluation.
python test_ood.py logs/cifar10/resnet_resnet18/ours/config.yaml use_ema=true \
ood_data.name=svhn ood_data.root=/data \
model.beta=0.1 model.ebm_augmentation=none
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent