(4/n) Support 2D Parallelism - Loading optimizer states correctly #19872
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
awaelchli
added
feature
github-actions
bot
added
fabric
awaelchli
force-pushed
the
feature/tp-full-optim-load
branch
from
99bc77b
to
18c8544
Compare
awaelchli
added
feature
awaelchli
force-pushed
the
feature/tp-full-optim-load
branch
from
18c8544
to
dbee485
Compare
for more information, see https://pre-commit.ci
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #19872 +/- ##
=========================================
- Coverage 84% 59% -25%
=========================================
Files 425 420 -5
Lines 35023 34930 -93
=========================================
- Hits 29371 20529 -8842
- Misses 5652 14401 +8749 |
justusschock
approved these changes
May 17, 2024
lantiga
approved these changes
May 17, 2024
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR adds the missing loading logic for optimizer states. When loading a regular optimizer state dict into a distributed optimizer, we need to convert the parameter keys from ids to fully qualified names. This conversion function is currently under the FSDP1 API, but is not FSDP1 specific. In a conversation on the PyTorch Slack it was discussed that PyTorch should probably support that directly in the
set_optimizer_state_dictfunction in the future, so we will be able to simplify this logic hopefully.📚 Documentation preview 📚: https://pytorch-lightning--19872.org.readthedocs.build/en/19872/
cc @Borda @awaelchli @carmocca @justusschock