Skip to content

fix(grpo): update grpo_sync seq_logprob_error_masking call site #2599

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.

Sign up for GitHub

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

Closed
ZhiyuLi-Nvidia wants to merge 1 commit into from
Closed

fix(grpo): update grpo_sync seq_logprob_error_masking call site #2599

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 26 additions & 8 deletions nemo_rl/algorithms/grpo_sync.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -769,17 +769,37 @@ def grpo_train_sync(
}
)

(
max_seq_mult_prob_error,
num_masked_seqs,
masked_correct_pct,
) = compute_and_apply_seq_logprob_error_masking(
seq_error_result = compute_and_apply_seq_logprob_error_masking(
train_data=masking_data,
rewards=rewards,
seq_logprob_error_threshold=master_config.grpo[
"seq_logprob_error_threshold"
],
)
seq_logprob_error_metrics = {
"max_seq_mult_prob_error": seq_error_result[
"max_seq_mult_prob_error"
],
"mean_seq_mult_prob_error": seq_error_result[
"mean_seq_mult_prob_error"
],
"min_seq_mult_prob_error": seq_error_result[
"min_seq_mult_prob_error"
],
"max_seq_mult_prob_error_after_mask": seq_error_result[
"max_seq_mult_prob_error_after_mask"
],
"mean_seq_mult_prob_error_after_mask": seq_error_result[
"mean_seq_mult_prob_error_after_mask"
],
"min_seq_mult_prob_error_after_mask": seq_error_result[
"min_seq_mult_prob_error_after_mask"
],
"num_masked_seqs_by_logprob_error": seq_error_result[
"num_masked_seqs"
],
"masked_correct_pct": seq_error_result["masked_correct_pct"],
}
# masking may have mutated sample_mask in place —
# capture the post-masking value for delta-write.
sample_mask = masking_data["sample_mask"]
Expand Down Expand Up @@ -1011,9 +1031,7 @@ def grpo_train_sync(
metrics["generation_logger_metrics"] = generation_logger_metrics
total_valid_tokens += metrics["global_valid_toks"]

metrics["max_seq_mult_prob_error"] = max_seq_mult_prob_error
metrics["num_masked_seqs_by_logprob_error"] = num_masked_seqs
metrics["masked_correct_pct"] = masked_correct_pct
metrics.update(seq_logprob_error_metrics)

consumed_samples += master_config.grpo["num_prompts_per_step"]
timeout.mark_iteration()
Expand Down
Loading