[NF4][FSDP] return contiguous quantization_factor
#298
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Outdated suggestions cannot be applied.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.
before,
NF4.quantization_factor
is non-contiguous, meaningNF4.stride()
is (16, 0)after,
NF4.quantization_factor
contiguous withNF4.stride()
is (1, 0). it's easy to work with FSDP since we do.view(unit8)
and it requries contiguous tensor