Bridging Semantic and Modality Gaps in Zero-Shot Captioning via Retrieval from Synthetic Data.

Zhiyue Liu, Wenkai Zhou

Browse the full EMNLP paper archive.