Updates Interpretation class to be memory efficient #3558
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Outdated suggestions cannot be applied.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.
Interpretation
is now memory efficient and should be able to process any size dataset, provided the hardware could train the same model.Interpretation
now callsget_preds
for each item to generate inps, targs, etc.GatherPredsCallback
has two new arguments,with_preds
andwith_targs
to control whether it returns predictions or targets. These are used byInterpretation
for memory efficient creation.Also modifies
GatherPredsCallback
to usetorch.save
, which handles both tensors and tuples of tensors.