dSprites

'dSprites' is a dataset that contains 2D shapes varying in factors such as shape, scale, orientation, and position, and it is used to evaluate the performance of models in disentangled representation learning.

Papers using dSprites (8)

L-VAE: Variational Auto-Encoder with Learnable Beta for Disentangled Representation2025 · 2 citesIB-GAN: Disentangled Representation Learning with Information Bottleneck Generative Adversarial Networks2025Multiple Invertible and Partial-Equivariant Function for Latent Vector Transformation to Enhance Disentanglement in VAEs2025Disentanglement as Identifiable Pushforward Factorisation2024Uniform Transformation: Refining Latent Representation in Variational Autoencoders2024 · 2 citesDisentangled Representation Learning Using ($\beta$-)VAE and GAN2022DOT-VAE: Disentangling One Factor at a Time2022Variantional autoencoder with decremental information bottleneck for disentanglement2023