5fa1a76
1
2
3
Attention mechanisms Most transformer models use full attention in the sense that the attention matrix is square.
