global semantic

one token (1) vector-to-sequence decoder (1) character localization (1) state space layer (1) compositional congruence (1) context-aware image restoration (1) abstractive summarization (1) scene text recognition (1) multimodal transformer (1) neural topic model (1)