refactor: remove redundant Tensor allocation in GaussianConditional.update() #343
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
…pdate()
In GaussianConditional.update(), a temporary Tensor is allocated:
quantized_cdf = torch.Tensor(len(pmf_length), max_length + 2)
quantized_cdf = self._pmf_to_cdf(pmf, tail_mass, pmf_length, max_length)
The first line is a dead store: the variable is immediately overwritten by
the result of _pmf_to_cdf(). This removes the unnecessary allocation.
- No functional changes
- Slightly reduces heap traffic and avoids creating an uninitialized Tensor
- Keeps dtype/device fully defined by _pmf_to_cdf()
fracape
approved these changes
Oct 22, 2025
|
Thanks for merging. |
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Outdated suggestions cannot be applied.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.
Suggestion
This PR removes a redundant one-line Tensor allocation in
GaussianConditional.update()withinentropy_models.py.Testing
Ran unit tests that exercise GaussianConditional.update() and dependent
compression/decompression paths; no regressions observed.
Verified that self._quantized_cdf, self._offset, and self._cdf_length
match previous values bit-for-bit for a fixed scale_table.
Why this trivial change matters
TorchDynamo or FX tracing can sometimes interpret redundant allocations as live ops, bloating the captured graph.
Removing this line avoids all of the above while keeping
_pmf_to_cdf()solely responsible for dtype and device consistency.No behavioral change is expected; this is a safe cleanup.