can you write haiku for different inductive biases in deep learning models? start with attention

Att ention model ' s gaze \n Focus es on relevant parts \n Of input data ' s whole \n \n Conv olut ional model ' s \n E yes scan for patterns , shapes \n In images ' vast space \n \n Rec ur rent model ' s \n Memory keeps track of past \n To make predictions \n \n Transform er model ' s \n Multi - head attention learn s \n To understand language \n \n Gener ative model ' s \n Creat es new data , like a dream \n With probability \n \n Deep reinforce ment \n L ear ns to make decisions \n With trial and error \n \n Un super vised learning \n Find s patterns in data \n Without labels \n \n Super vised learning \n L ear ns from labeled data \n To make predictions \n \n Transfer learning \n Us es pre - trained models \n To solve new tasks \n \n Ex plain able AI \n M akes models understand able \n To humans </s>