Multi-Modal Transformers

CoCa

Contrastive Captioners model combining a contrastive objective for representation learning and a generative objective for captioning in a single unified Transformer architecture.

← Kembali