Transformer Architecture · Chapter 05
Multi-Head Attention
h parallel heads, same cost, richer representation
This chapter is part of the Pro track. Pro beta members receive the complete self-paced architecture sequence.
Related reading · Knowledge Lab