exp(glm45air): rubric monitoring/shaping configs (rlm + swe judges, research/glm-5.2)#2997
Draft
mikasenghaas wants to merge 2 commits into
Draft
exp(glm45air): rubric monitoring/shaping configs (rlm + swe judges, research/glm-5.2)#2997mikasenghaas wants to merge 2 commits into
mikasenghaas wants to merge 2 commits into
Conversation
Based on configs/glm45air/swe.toml, attaching rlm + swe rubric judges (GLM-5.2 on default Prime inference) to the scaleswe train env: - swe_rubric_monitor.toml: weight=0 (metric-only monitoring; emits rlm/*, swe/*) - swe_rubric_shaped.toml: weight=0.1 (shapes reward alongside the task reward) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…nitors - judge model -> research/glm-5.2 (default Prime inference) - add rlm + swe rubric judges (weight 0) on the swebench-verified eval env in both configs Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Rubric-judge ablations on
configs/glm45air/swe.toml(GLM-4.5-Air on scaleswe, eval on SWE-Bench-Verified), attaching therlm(harness-mechanics) +swe(code-usage) rubric judges fromresearch-environments, judged byresearch/glm-5.2.configs/glm45air/swe_rubric_monitor.tomlrlm/*+swe/*metrics, no reward effectconfigs/glm45air/swe_rubric_shaped.tomlIdentical except the train judge weight (0 vs 0.1). Eval judges are weight-0 in both (eval isn't trained).
research/glm-5.2returns HTTP 404 on the default Prime inference endpoint (as do all GLM-5.2 id variants;/modelsis 403). These configs won't run until that model is accessible (access grant, or a base_url/api-key override for the research server).