Ensure Batchifier returns compatible arrays for hybrid engines like OnnxRuntime #3447
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Outdated suggestions cannot be applied.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.
Addresses issue raised from https://deepjavalibrary.slack.com/archives/C01AURG857U/p1724804411722599.
Essentially batchifying / unbatchifying OrtNDArrays produce PtNDArrays, and this causes downstream issues, particularly in OrtSymbolBlock:114:
Because of the batchify call during Translator processInput() call, the initial OrtNDArrays end up as PtNDArrays, and then the ret.attach call incorrectly sets the ret NDList's main (and alternate manager) both to PtNDManager. This inhibits a lot of post-processing code from working as the returned NDArrays have the wrong manager set.
Maybe there is a more elegant way of fixing this within OrtSymbolBlock, but I couldn't see any easy way of passing in the Predict NDManager in.