ConceptAI & ML
Self-attentive feed-forward units
An alternative mechanism that performs vector comparisons first and uses a differently-shaped weight matrix to interpret vector comparisons for optimal weighting.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Feed-forward layers1Self-attention1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.