Towards Effective and Efficient Non-autoregressive Decoding Using Block-based Attention Mask

Open in new window

The document is on a web server that cannot be reached. Please try again later.