Can a model read a learner's reasoning? The SENSUS benchmark, open training data and the interactive Duel.
AI & ML interests
Post-training data and evaluation to advance human understanding
Post-training data and evaluation to advance human understanding