Fix saving bug for untied weights with keras 3.2 #1568
Merged
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
mattdangerw
force-pushed
the
fix-saving-bug
branch
from
622a1bd to
0935c08
Compare
Tricky bug, keras.layers.Embedding has start overriding save_own_variables/load_own_variables for quantization. We can no longer rely on absolute indexing for the reverse embedding, e.g. store["1"] = self.reverse_embeddings I also couldn't figure out a way to continue to support loading tied weights untied. This is because the embedding will now unconditionally error if weight counts are mismatched here: https://github.com/keras-team/keras/blob/v3.2.0/keras/layers/core/embedding.py#L232-L264 I don't think we have a direct need for this functionality today, so I just removed it, but could be worth considering how to bring back in the future.
mattdangerw
force-pushed
the
fix-saving-bug
branch
from
0935c08 to
36ef41e
Compare
fchollet
approved these changes
Apr 10, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
Tricky bug, keras.layers.Embedding has started overriding save_own_variables/load_own_variables for quantization. We can no longer rely on absolute indexing for the reverse embedding, e.g.
self.reverse_embeddings.assign(store["1"])I also couldn't figure out a way to continue to support loading tied weights untied. This is because the embedding will now unconditionally error if weight counts are mismatched here: https://github.com/keras-team/keras/blob/v3.2.0/keras/layers/core/embedding.py#L232-L264
I don't think we have a direct need for this functionality today, so I just removed it, but could be worth considering how to bring back in the future.