Walk me through the mechanics of self-attention in Transformer models and how its computational complexity scales with sequence length.
