In attention mechanisms, the ability to focus on relevant parts of the input sequence while ignoring others is often referred to as _____.

In attention mechanisms, the ability to focus on relevant parts of the input sequence while ignoring others is often referred to as selective attention. This means the model learns to weigh each token differently, giving higher scores to the important ones. For example, in a sentence, the word “dog” might get a higher attention weight than “the” when predicting the next word. The attention scores are computed by comparing query, key, and value vectors, which lets the model decide which tokens matter most. This selective focus is what gives transformers their power to handle long sequences efficiently.