[optim] Allow torch.float64 scalars for forloop + foreach implementations #115841
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/115841
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 8260136 with merge base 0978482 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
janeyx99
changed the title
albanD
reviewed
Dec 15, 2023
| def _get_scalar_dtype(is_fused=None): | ||
| if is_fused: | ||
| return torch.float32 | ||
| return torch.float64 if torch.get_default_dtype() == torch.float64 else torch.float32 |
There was a problem hiding this comment.
I don't remember the story for default dtype vs default Tensor type. Do we need to check both here?
There was a problem hiding this comment.
I don't think we do, get_default_dtype seems to work as expected:
…implementations" Should allow for uses cases mentioned in #110940 This would allow scalars to also be float64s in the foreach implementation. The fused implementation would still create a float32 step on Adam and AdamW. This PR also does NOT worry about performance and is mainly for enablement. Next steps: - Relax the constraint on fused adam(w) and allow torch.float64 scalars there - Allow _performant_ mixed dtypes in foreach (a bigger project in itself). This PR will conflict with my other PRs, I will figure out a landing order [ghstack-poisoned]
…implementations" Should allow for uses cases mentioned in #110940 This would allow scalars to also be float64s in the foreach implementation. The fused implementation would still create a float32 step on Adam and AdamW. This PR also does NOT worry about performance and is mainly for enablement. Next steps: - Relax the constraint on fused adam(w) and allow torch.float64 scalars there - Allow _performant_ mixed dtypes in foreach (a bigger project in itself). This PR will conflict with my other PRs, I will figure out a landing order [ghstack-poisoned]
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Should allow for uses cases mentioned in #110940
This would allow scalars to also be float64s in the foreach implementation. The fused implementation would still create a float32 step on Adam and AdamW. This PR also does NOT worry about performance and is mainly for enablement.
Next steps:
This PR will conflict with my other PRs, I will figure out a landing order
Stack from ghstack (oldest at bottom):