Knowledge and Language Team

UDOP

2024年2月

UDOP adopts an encoder-decoder Transformer architecture based on T5 for document AI tasks like document image classification, document parsing and document visual question answering. You can use the model for document image classification, document parsing and document visual question answering…