Boren Hu

visual question answering (1) self-supervised learning (1) question answering (1) text generation (1) passage retrieval (1) multimodal large language model (1) layer skipping (1) retrieval-augmented generation (1) inference speed (1) domain-specific knowledge (1) evidence extraction (1) bert inference (1) logit len (1) self-aligned learning (1) dynamic early exiting (1) cross-layer contrastive learning (1)