Teach the dashboard and the CI checkers the policy namespace - #2649
Open
fzyzcjy wants to merge 1 commit into
Open
Teach the dashboard and the CI checkers the policy namespace#2649fzyzcjy wants to merge 1 commit into
fzyzcjy wants to merge 1 commit into
Conversation
fzyzcjy
requested review from
Shi-Dong,
Zhichenzzz,
guapisolo,
jybsuper,
maocheng23 and
yueming-yuan
as code owners
August 18, 2026 11:45
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
Claude Code Review
This repository is configured for manual code reviews. Comment @claude review for a one-time review, or @claude review always to subscribe this PR to a review on every future push.
Tip: disable this comment in your organization's Code Review settings.
This was referenced Aug 25, 2026
This was referenced Aug 27, 2026
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-25
branch
from
August 28, 2026 11:42
4892069 to
80dac03
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-24
branch
2 times, most recently
from
August 28, 2026 12:38
5d96759 to
5b408d6
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-25
branch
2 times, most recently
from
August 28, 2026 14:33
239634a to
7a76cb7
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-24
branch
from
August 28, 2026 15:04
02bc9d3 to
c991a46
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-25
branch
from
August 28, 2026 15:04
7a76cb7 to
662d788
Compare
- Record which step key every metric was logged against, and answer it from `/api/meta`: a multi policy run writes `alpha/train/loss` against `alpha/train/step`, which the key-prefix rule the chart axis used cannot guess, so every namespaced curve was drawn against the wrong axis. - Read the perf points of the bubble strip by suffix, since the strip keyed off the exact `perf/step_time` and a namespaced run left it empty. - Strip the namespace before the in-process checkers read a metric: `check_kl` and the rollout log-prob check index `train/ppo_kl` and `rollout/log_probs`, and the CI history gate captures a fixed key list, so a namespaced run silently checked nothing at all. - Cover both halves: the recorded step keys and the bubble lookup, and that the checkers see the names they are written against whichever policy produced the step.
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-25
branch
from
August 28, 2026 18:20
662d788 to
8ebe30e
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op28-24
branch
from
August 28, 2026 18:20
c991a46 to
5c79908
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.
Part of #1837