Add torch.tensor_split op support for PyTorch frontend#2715
Open
eyupcanakman wants to merge 1 commit into
Open
Add torch.tensor_split op support for PyTorch frontend#2715eyupcanakman wants to merge 1 commit into
eyupcanakman wants to merge 1 commit into
Conversation
tensor_split was not implemented in the TorchScript frontend. Add a converter that lowers to mb.split, for an integer number of sections (first dim_size % n pieces get one extra element) and a list of split indices. Non-monotonic indices raise NotImplementedError since they map to overlapping slices that mb.split cannot represent. Fixes apple#2226.
Collaborator
|
Thanks @eyupcanakman. This change looks good. CI: https://gitlab.com/coremltools1/coremltools/-/pipelines/2554644433 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Converting a TorchScript model that calls
torch.tensor_splitfails withPyTorch convert function for op 'tensor_split' not implemented.Adds a converter that computes the split sizes and lowers to
mb.split, for both an integer section count (firstdim_size % npieces get one extra element) and an explicit list of indices. Non-monotonic indices raiseNotImplementedError, since they map to overlapping slicesmb.splitcannot represent.torch.exportalready decomposestensor_splitinto slice ops, so this only affects the TorchScript frontend.Test plan
pytest test_torch_ops.py::TestTensorSplitcovers integer sections, index lists with negative indexing, and the non-monotonic refusal, across all frontend and backend combinations.Fixes #2226.