Search Results for author:
Found 1 papers, 0 papers with code
A Simple and Effective Positional Encoding for Transformers
Our analysis shows that the gain actually comes from moving positional information to attention layer from the input.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
