Limitations of Normalization in Attention Mechanism

Understand

This paper investigates the limitations of the normalization in attention mechanisms.

• We begin with a theoretical framework that enables the identification of the model's selective ability and the geometric separation involved in token selection.
• Our analysis includes explicit bounds on distances and separation criteria for token vectors under softmax scaling.
• Through experiments with pre-trained GPT-2 model, we empirically validate our theoretical results and analyze key behaviors of the attention mechanism.