Zhiliang Peng

foundation model (2) transformer architecture (2) transfer learning (2) representation learning (2) multimodal learning (2) visual transformer (2) self-supervised learning (1) knowledge distillation (1) instance segmentation (1) model compression (1) convolutional neural network (1) vision language model (1) masked autoencoder (1) attention map (1) model initialization (1) transformer encoder-decoder (1) multimodal pretraining (1) masked image modeling (1) feature pyramid network (1) weakly supervised object localization (1)