Separate comparison inputs and standardize scoring and warnings - #1

Merged

jonabur merged 5 commits into

mainfrom

config/sib200-jeebench-baselines

Oct 1, 2026

Commits on Oct 1, 2026

Use published JEEBench baseline and document SIB-200 accuracy selection
Jonathan Burdge
committed
0c347f6
Separate eval catalogue, weighting profiles and named comparison sets
Jonathan Burdge
committed
aa56b68
Align scoring choices and warn consistently about comparison exclusions
Jonathan Burdge
committed
73f31be
Align JEEBench random baseline to 10.55 percent
Jonathan Burdge
committed
1e2c829
Offer original and code-and-math weighting profiles
Jonathan Burdge
committed
a19a853