Socratic Models: Composing Zero-Shot Multimodal Reasoning with Language

Benchmark Model Rank Results
video-retrieval-on-msr-vtt-1kaSocratic Models#56video-to-text R@1: 42.8