cuny-cl / maxwell Goto Github PK

We use TQDM to log progress during each epoch of training. But we don't use it during the validation phase that computes log-likelihood, and when this takes more than a few seconds it looks like nothing is happening. The obvious thing to do, then, is to use a separate TQDM slider during that phase too. For how it ought to look, we can consult what happens during validation in Yoyodyne. This of course is of low importance but is a nice-to-have.