enable merging parameters for diloco #212
Merged
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
facebook-github-bot
added
the
CLA Signed
tushar00jain
force-pushed
the
pr212
branch
18 times, most recently
from
e609b5a to
b93142d
Compare
tushar00jain
force-pushed
the
pr212
branch
5 times, most recently
from
3f8e0b8 to
323fb47
Compare
This was referenced Jun 15, 2025
Merged
tushar00jain
force-pushed
the
pr212
branch
2 times, most recently
from
37da9d9 to
8ce038d
Compare
tushar00jain
force-pushed
the
pr212
branch
8 times, most recently
from
7a016b6 to
355deed
Compare
tushar00jain
force-pushed
the
pr212
branch
3 times, most recently
from
b5eb209 to
6ad8993
Compare
tushar00jain
force-pushed
the
pr212
branch
3 times, most recently
from
8b76f91 to
6524d16
Compare
d4l3k
approved these changes
Jul 12, 2025
| @@ -719,6 +721,7 @@ def test_streaming_diloco_commit_failure( | |||
| "diloco_args": { | |||
| "fragment_sync_delay": fragment_sync_delay, | |||
| "sync_every": 4, | |||
| "fragment_update_alpha": alpha, | |||
There was a problem hiding this comment.
Thoughts on adding some mocked out tests where we actually check the numerical values against some references to avoid regressions?
With a fixed seed and deterministic mode in torch we can add regression tests by comparing against a known value.
d4l3k
reviewed
Jul 12, 2025
torchft/local_sgd.py
Outdated
| Merges the local and global parameters. | ||
| """ | ||
| for name, p in self._model_fragment.named_parameters(): | ||
| torch.lerp( |
There was a problem hiding this comment.
this isn't inplace right? having some numerics tests would be nice to catch issues like this
There was a problem hiding this comment.
yep let me add this and regression test in a separate diff?
Summary: - merge local and global parameters of the model after synchronization - add the "alpha" parameter to integration tests Test Plan: ``` pytest -vs ./torchft/local_sgd_integ_test.py ```
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.
Summary:
Test Plan:
Stack created with Sapling. Best reviewed with ReviewStack.