Sparse Attention

Efficient Attention

Paradigm encompassing all attention variants aimed at reducing computational complexity while preserving the modeling capabilities of Transformers.

← Geri