All vision-language models should have hyperbolic embeddings. Vision and language are incredibly hierarchical in nature! See below our latest work on hyperbolic vision-language models that exploit visual compositions through entailment:
All vision-language models should have hyperbolic embeddings. Vision and language are incredibly hierarchical in nature! See below our latest work on hyperbolic vision-language models that exploit visual compositions through entailment:
@PascalMettes Non-Euclidean geometries are hardly a new concept. What the transformer truly needs is a design overhaul, not an endless series of cosmetic fixes. ai-cosmos.hashnode.dev/beyond-gradien…
