Defeasible Visual Entailment: Benchmark, Evaluator, and Reward-Driven Optimization

Open in new window