Stage 2+3a of the MMIDAS pipeline. Optionally trains a UDAGAN VAE-GAN augmenter (02a), trains the core cpl-mixVAE model with iterative pruning (02b), and evaluates all checkpoints to select the optimal model order (03a). Outputs evaluation figures and evaluation_results.json for human review before running MMIDAS_Analyze.
This is a companion discussion topic for the original entry at github.com/broadinstitute/warp/MMMIDAS_Train