Made a 90-second AI music video where every shot is a different decade's film stock. The prompt was longer than the script.
Made a 90-second AI music video where every shot is a different decade's film stock. The prompt was longer than the script.
Saving this immediately. The rubric-before-answer trick fixed my eval drift too — +6% on my set.
Counterpoint: I've seen rubric-first make models overconfident on edge cases. Did you check calibration?