Sparse attention
Attention in which each position looks at only a chosen subset of other positions, instead of all of them. It cuts the cost of attention on long sequences.
Introduced by
Sparse Transformer, 2019
Used by
- Longformer, 2020
- GPT-3, 2020
- BigBird, 2020