Spaces:
Running
Running
Commit History
-
Add the cross-base-model view: twenty published methods on three base models
Verified signatureOne new section renders the twenty-method roster scored on substrate P across three base models as a slope chart. The vertical axis carries rank within a base model or gain over that model's no-intervention row, never raw K-Score, following the benchmark's within-model reading rule. Collapsed cells hold no score and sit in a proportionally sized band rather than being plotted as zeros.5be569b -
Give the submission section a short anchor
Verified signaturefc5c16d -
Put the baseline and the null-intervention control at the top of the matrix
Verified signaturebf927c3 -
Link each method name to its source paper
Verified signature26064b8 -
Publish the K-Bench leaderboard
Verified signature50bb224 -
initial commit
Verified signaturea27fd70
