Fix SimulateQuantizedEinsum to use einsum_str instead of module name … #495
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
…for pattern-specific quantization - Changed line 192 in gemma/peft/_quantization.py from einsum_str=self.wrapped.name to einsum_str=einsum_str - This enables pattern-specific quantization axis selection logic in get_axis_to_reduce_from_einsum_str() - Previously, module names like 'attention_proj' were passed, always returning None and forcing fallback to generic per-channel scaling - Now actual einsum equations like 'BTD,NDH->BTNH' are passed, enabling optimal pattern-specific scaling - Improves quantization accuracy for all einsum operations in quantization-aware training workflows
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
…for pattern-specific quantization