Clone coding of PyTorch-VAE to study VAE
python train.py -c configs/celeba.yaml
python train.py -c configs/mnist.yaml
Original image | 2-D latent space | 5-D latent space | 10-D latent space | 20-D latent space |
---|---|---|---|---|
![]() |
![]() |
![]() |
![]() |
![]() |
- Use WandB instead of TensorBoard
- Can train with MNIST dataset
- Add MNIST reconstruction results
- Visualize of learned data manifold