Just me plying around with some unsupervised algorithms
Repo structure:
- unsuper/
- data/ - files for loading data
- helper/ - helper files
- encoder_decoder.py - some different encoder and decoder architechtures
- expm.py - matrix exponential in pytorch
- losses.py - ELBO loss for variational autoencoders
- spatial_transformer.py - ST layers for different transformers
- utility.py - different helper functions
- models/
- vae.py - classic variational autoencoder models
- vitae_ui.py - variational inferred transformational auto encoders with semantic inference conditioned on transformational latents
- vitae_ci.py - variational inferred transformational auto encoders with unconditional inference on transformational latents
- trainer.py - file that does all the optimization work
- main.py - for running experiments