Fix metric logging for DPO #583
Merged
+35
−18
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
copybara-service
bot
force-pushed
the
test_820374667
branch
from
d5ec4f3 to
7240a68
Compare
copybara-service
bot
force-pushed
the
test_820374667
branch
from
7240a68 to
6d55e8c
Compare
copybara-service
bot
force-pushed
the
test_820374667
branch
from
6d55e8c to
94aff5a
Compare
Metric logging for DPO is incorrect right now. We should use MetricBuffer to log metrics. The way I figured this out is that when I ran DPOTrainer, the eval accuracy went to 100% after the very first train step, which did not make sense at all. PiperOrigin-RevId: 820400295
copybara-service
bot
force-pushed
the
test_820374667
branch
from
94aff5a to
ac07cfd
Compare
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Outdated suggestions cannot be applied.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.
Fix metric logging for DPO
Metric logging for DPO is incorrect right now. We should use MetricBuffer to log
metrics.
The way I figured this out is that when I ran DPOTrainer, the eval accuracy went to 100% after the very first train step, which did not make sense at all.