Fix bad initial values shape assumptions in save_mem_new_scan
rewrite
#1501
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Outdated suggestions cannot be applied.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.
This PR closes #1499. The underlying issue was that
save_mem_new_scan
always assumed that the output storage array graphs, which take the format.set_subtensor(at.empty[:tap_spread], initial_tap_values)
, would have non-scalarinitial_tap_values
; however, this isn't always the case.In general, the logic in
save_mem_new_scan
is a bit convoluted and unclear about the assumptions it makes. These changes don't completely fix that, but they do provide some refactoring that improves the situation a little.