Error for unsupported precision types with ModelParallelStrategy #19902
+46
−13
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
github-actions
bot
added
fabric
for more information, see https://pre-commit.ci
awaelchli
requested review from
carmocca,
justusschock,
Borda and
williamFalcon
as code owners
carmocca
approved these changes
May 23, 2024
lantiga
approved these changes
May 23, 2024
for more information, see https://pre-commit.ci
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #19902 +/- ##
=========================================
- Coverage 84% 59% -25%
=========================================
Files 426 421 -5
Lines 35247 35156 -91
=========================================
- Hits 29621 20752 -8869
- Misses 5626 14404 +8778 |
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Outdated suggestions cannot be applied.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?
Adds error handling if user passes
precision="16-mixed". This would require a sharded gradscaler to be configured, but we opted for not supporting this at the moment.precision="bf16-mixed"(bfloat) is still supported, as it doesn't require a grad scaler. Other precisions such as bitsandbytes are also not allowed (they don't compose with this strategy) by design.cc @Borda @carmocca @justusschock @awaelchli