promptdojo_

Mission: baseline model showdown — step 5 of 7

This showdown says the rule doesn't beat the majority baseline — but look closely: each contender was scored on a DIFFERENT slice of the test set. A scoreboard is only a comparison when every rung faces the same rows. Fix both scoring calls to use the full test set.

The break is on lines 12, 13 — but read the whole snippet first.

full-screen editor opens — close anytime to keep reading.