Track entropy at training time #178
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
It may be useful to see how entropy changes over time, since once entropy collapses the model is unlikely to learn much more.
Not sure whether this is the best way to do it; notably, when doing it this way we aren't able to track entropy on the val set, only at training.
This is a subset of my changes from the
increase-entropy
branch, where I tried to explicitly reward increased entropy. However that branch, while successfully maintaining higher entropy, didn't seem to help training, so I only copied over the reporting changes here.