Artifacts
What each model built on each task, shown on the held-out cases it was graded on.
Oxilite query costs
Each model made Oxilite, a Rust SPARQL store, cheaper to load, smaller on disk and cheaper to query. Hidden test query groups are the ones the reward is computed on (every attempt's graded cost ratios and G); public dev query groups show every ratio the public scorer printed during an attempt, but agents could test on them, so they are not graded.