Compositional Generalization
A core concept in AI and cognitive science referring to a model's ability to recombine learned primitive operations to solve composite tasks not encountered during training. It is regarded as a critical threshold toward genuine intelligent reasoning.
Timeline (last 90 days)
The academic community typically uses compositional generalization to measure genuine reasoning ability.
The concept of compositional generalization originates from the generative grammar theory of linguist Noam Chomsky, referring to the generation of infinite combinations through finite rules.
Research shows that large-scale multimodal models trained on massive data have to some extent developed emergent compositional generalization capabilities, but still fail on certain extreme combinations.
All Facts (3)
The academic community typically uses compositional generalization to measure genuine reasoning ability.
50%UnverifiedThe concept of compositional generalization originates from the generative grammar theory of linguist Noam Chomsky, referring to the generation of infinite combinations through finite rules.
50%UnverifiedResearch shows that large-scale multimodal models trained on massive data have to some extent developed emergent compositional generalization capabilities, but still fail on certain extreme combinations.
50%