Semantic Coverage
A diversity metric that measures the breadth of distribution of generated content within semantic space, typically computed by means of sentence embeddings and density estimation.
Timeline (last 90 days)
Concept-guided prompting achieves a semantic coverage rate of 47.0% in the recipe generation task.
The prompting strategy based on previous titles achieves a semantic coverage rate of 35.1% in the recipe generation task.
The meaningless prompting strategy achieves a semantic coverage rate of 13.1% in the recipe generation task.
Semantic coverage uses sentence embeddings to map generated content into a high-dimensional vector space, and measures the breadth of semantic regions via density estimation or cluster coverage.
Semantic coverage differs from simple deduplication rates or lexical diversity (such as distinct-n) in that it can capture pseudo-diversity where 'the wording changes but the meaning is the same'.
All Facts (5)
Concept-guided prompting achieves a semantic coverage rate of 47.0% in the recipe generation task.
50%UnverifiedThe prompting strategy based on previous titles achieves a semantic coverage rate of 35.1% in the recipe generation task.
50%UnverifiedThe meaningless prompting strategy achieves a semantic coverage rate of 13.1% in the recipe generation task.
50%UnverifiedSemantic coverage uses sentence embeddings to map generated content into a high-dimensional vector space, and measures the breadth of semantic regions via density estimation or cluster coverage.
50%UnverifiedSemantic coverage differs from simple deduplication rates or lexical diversity (such as distinct-n) in that it can capture pseudo-diversity where 'the wording changes but the meaning is the same'.
50%