Bruce W. Lee comments on [Paper] Language Models Don’t Learn the Physical Manifestation of Language

[Bruce W. Lee]

“image caption generation and video simulation currently used in Sora will partially correct some of these errors.” I’m in line with this idea.