update FSDPTransform to delegate materialization to the MaterializationTransform #995
Conversation
This file contains 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
Co-authored-by: Masaki Kozuki <mkozuki@nvidia.com>
This was referenced Aug 19, 2024
crcrpar
reviewed
Aug 19, 2024
Co-authored-by: Masaki Kozuki <mkozuki@nvidia.com>
t-vi
force-pushed
the
tom/fsdp-composable-step3
branch
from
4ce64b7 to
be3c931
Compare
Co-authored-by: Masaki Kozuki <mkozuki@nvidia.com>
Co-authored-by: Masaki Kozuki <mkozuki@nvidia.com>
t-vi
force-pushed
the
tom/fsdp-composable-step4
branch
from
bdc0f4b to
4ae9d49
Compare
|
I need to investigate the failing test in test_transforms in more detail, the FSDP part seems in good shape. |
Co-authored-by: Luca Antiga <luca@lightning.ai>
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Outdated suggestions cannot be applied.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.
This is the fourth / final step for #987 .
After this, the FSDP transform plays well with Materialization Transform and should also work with previous transforms (however, the bnb quantization changes tensor layout in a way not terribly amenable to sharding, this needs investigation).
This is a draft to remember to update the branch.
cc: @crcrpar
Edit I need to do my rebasing homework. This PR only changes files in distributed/ the others are from a merge with #980 which will go away after rebasing.