TY  - RPRT
TI  - Revisiting Compositionality in Dual-Encoder Vision-Language Models: The Role of Inference
AU  - Imanol Miranda
AU  - Ander Salaberria
AU  - Eneko Agirre
AU  - Gorka Azkune
PY  - 2026
UR  - https://arxiv.org/abs/2604.11496
ID  - 2604.11496
ER  -