Update the _maybe_compute_stride_kjt logic to calculate stride based off of inverse_indices for VBE KJTs. #2925
Conversation
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
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D73824764 |
jd7-tr
force-pushed
the
export-D73824764
branch
from
5b3880b to
9e3d5e6
Compare
|
This pull request was exported from Phabricator. Differential Revision: D73824764 |
…ic (pytorch#2903) Summary: Pull Request resolved: pytorch#2903 # Context * Currently torchrec IR serializer can't handle variable batch use case. * `torch.export` only captures the keys, values, lengths, weights, offsets of a KJT, however, some variable-batch related parameters like `stride_per_rank` or `inverse_indices` would be ignored. * This test case (expected failure right now) covers the vb-KJT scenario for verifying that the serialize_deserialize_ebc use case works fine with KJTs with variable batch size. # Ref Differential Revision: D73051959
jd7-tr
added a commit
to jd7-tr/torchrec
that referenced
this pull request
May 1, 2025
…off of inverse_indices for VBE KJTs. (pytorch#2925) Summary: See D73051959 for context. Update the `_maybe_compute_stride_kjt` logic to calculate stride based off of `inverse_indices` for VBE KJTs. Currently, stride of VBE KJT with `stride_per_key_per_rank` is calculated as the max "stride per key". This is different from the batch size of the EBC output KeyedTensor which is based off of `inverse_indices`. This causes issues in IR module serialization: [debug doc](https://docs.google.com/document/d/1yQhI484cgVloSqIBPAeTQhzfb3ltjvMRiLaQDceHGOU/edit?tab=t.0#heading=h.c66chahhl8df). Differential Revision: D73824764
…off of inverse_indices for VBE KJTs. (pytorch#2925) Summary: Pull Request resolved: pytorch#2925 See D73051959 for context. Update the `_maybe_compute_stride_kjt` logic to calculate stride based off of `inverse_indices` for VBE KJTs. Currently, stride of VBE KJT with `stride_per_key_per_rank` is calculated as the max "stride per key". This is different from the batch size of the EBC output KeyedTensor which is based off of `inverse_indices`. This causes issues in IR module serialization: [debug doc](https://docs.google.com/document/d/1yQhI484cgVloSqIBPAeTQhzfb3ltjvMRiLaQDceHGOU/edit?tab=t.0#heading=h.c66chahhl8df). Differential Revision: D73824764
|
This pull request was exported from Phabricator. Differential Revision: D73824764 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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:
See D73051959 for context.
Update the
_maybe_compute_stride_kjtlogic to calculate stride based off ofinverse_indicesfor VBE KJTs.Currently, stride of VBE KJT with
stride_per_key_per_rankis calculated as the max "stride per key". This is different from the batch size of the EBC output KeyedTensor which is based off ofinverse_indices. This causes issues in IR module serialization: [debug doc].(https://docs.google.com/document/d/1yQhI484cgVloSqIBPAeTQhzfb3ltjvMRiLaQDceHGOU/edit?tab=t.0#heading=h.c66chahhl8df).Differential Revision: D73824764