Repository navigation
fix(grpo_sync): honor force_on_policy_ratio in TQ trainer #3087
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Changes from all commits
Commits
Show all changes
19 commits
Select commit
Hold shift + click to select a range
8831d62
fix(grpo_sync): honor force_on_policy_ratio in TQ trainer
ZhiyuLi-Nvidia efb3081
refactor(grpo): drop misleading 0-fill for skipped seq_logprob metrics
ZhiyuLi-Nvidia d0efc1c
perf(grpo_sync): drop prev_logprobs TQ round-trip when force_on_polic…
ZhiyuLi-Nvidia 141afd0
refactor(grpo_sync): tighten helper + unify select_fields with dispatch
ZhiyuLi-Nvidia 088c9bb
refactor(grpo_sync): thread skip_prev_logprobs through train_from_meta
ZhiyuLi-Nvidia 6cb80f0
refactor(grpo_sync): driver owns the train field list
ZhiyuLi-Nvidia 50b1ccf
refactor(tq_policy): use DP_TRAIN_FIELDS as direct default for train_…
ZhiyuLi-Nvidia 94e5164
docs(grpo_sync): correct the prev_logprobs placeholder comment
ZhiyuLi-Nvidia b839b14
refactor(grpo_sync): pass None to compute_advantage when skipping pre…
ZhiyuLi-Nvidia a82d3ef
fix(grpo_sync): guard log_data prev_logprobs.tolist() against None
ZhiyuLi-Nvidia 435236d
Revert "refactor(grpo_sync): pass None to compute_advantage when skip…
ZhiyuLi-Nvidia 86fbfe6
style(grpo_sync): apply ruff-format to train_fields generator
ZhiyuLi-Nvidia 3333116
fix(grpo): assert-out use_kl_in_reward + force_on_policy_ratio
ZhiyuLi-Nvidia da11484
test(grpo): parametrized coverage for _resolve_logprob_skip_flags + k…
ZhiyuLi-Nvidia def6ee0
test(grpo): use MasterConfig+ClippedPGLossConfig instead of SimpleNam…
ZhiyuLi-Nvidia d21dba7
fix(grpo): defensive getattr for use_kl_in_reward assertion
ZhiyuLi-Nvidia 4d04c3d
fix(grpo): address PR #3087 review comments
ZhiyuLi-Nvidia 12ed729
refactor(grpo_sync): extract _train_fields_for_step + tighten tq comment
ZhiyuLi-Nvidia a522962
docs(tq_policy): restore composition + writers in train_from_meta com…
ZhiyuLi-Nvidia File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
|
ZhiyuLi-Nvidia marked this conversation as resolved.
Resolved
|
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
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
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
Oops, something went wrong.
Oops, something went wrong.
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.
Uh oh!
There was an error while loading. Please reload this page.