Merged
Conversation
FurtherAI
force-pushed
the
austin/fix_pr_607_regression_from_pr_619
branch
from
f9d79b2 to
348410b
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Removes the Schulman KL term that PR 619 accidentally reintroduced into
art.loss.loss_fn, restoring the behavior from PR 607.Root Cause
The LoRA correctness work in PR 619 refactored
Lossand brought back the direct KL-divergence term andLoss.klfield that PR 607 had intentionally removed.Changes
src/art/loss.pyloss.klscaling path from the Megatron oracle worker helperLossdoes not grow theklfield back silentlyValidation
uv sync --all-extrasuv run pytest src/art/test/test_kl_advantage.py -quv run python -m py_compile src/art/loss.py tests/integration/megatron_oracle_worker.py src/art/test/test_kl_advantage.py