GEB evaluates multimodal embeddings by decoding answers from the embedding alone, revealing information bottlenecks and model rankings that differ from retrieval-style benchmarks.
Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics , pages=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Generative Embedding Benchmark: How Much Information Survives in a Dense Embedding?
GEB evaluates multimodal embeddings by decoding answers from the embedding alone, revealing information bottlenecks and model rankings that differ from retrieval-style benchmarks.