Skip to content

[chore] Log generation output buffer size at the start of a training step #1930

Merged
SumanthRH merged 1 commit into
mainfrom
add-buffer-size-at-start
Jul 19, 2026
Merged

[chore] Log generation output buffer size at the start of a training step #1930
SumanthRH merged 1 commit into
mainfrom
add-buffer-size-at-start

Conversation

@SumanthRH

Copy link
Copy Markdown
Member

^^

Signed-off-by: SumanthRH <sumanthrh@anyscale.com>

@gemini-code-assist gemini-code-assist Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request introduces a new metric async/gen_buffer_qsize_at_wait_start to track the buffer occupancy when a training step starts waiting for the generation buffer in fully_async_trainer.py. There are no review comments provided, so I have no additional feedback to offer.

@SumanthRH
SumanthRH merged commit fd79cee into main Jul 19, 2026
5 checks passed
eicherseiji added a commit to eicherseiji/SkyRL that referenced this pull request Jul 20, 2026
…#1930

Review feedback from SumanthRH. The buffer capacity is a run constant:
compute it once in __init__ as the single source for both the queue
construction and a one-time skyrl_gen_buffer_maxsize publish, kept for
dashboard panels. Drop the per-step W&B keys: NovaSky-AI#1930 already logs
async/gen_buffer_qsize_at_wait_start, and this PR keeps only the
step-start Prometheus gauge for the depth.
eicherseiji added a commit to eicherseiji/SkyRL that referenced this pull request Jul 20, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant