search

greenelab / tybalt

Training and evaluating a variational autoencoder for pan-cancer gene expression data
BSD 3-Clause "New" or "Revised" License
162 stars 61 forks source link

Save recon_loss and KL_loss separately to visualize tradeoff #13

Closed gwaybio closed 6 years ago

gwaybio commented 7 years ago

Visualizing how each loss function contributes to the overall loss and how each is learned through each training epoch will be important. It will be important especially if warmup is implemented in #12

gwaybio commented 7 years ago

Possibly of interest will be generating a custom metric for model compilation