GAIPS Practice Test

Question 1 of 5

Single choice

Which mechanism lets a transformer model relate each token in its input to every other token when it computes representations?

Choose one answer

A. Self-attention, which scores each token against the other tokens in the same sequence
B. Gradient clipping applied to the loss during each optimizer step to stabilize training runs
C. Dropout regularization that randomly zeroes activations during training to reduce overfitting
D. Byte-pair merging rules that build the subword vocabulary before any model layers are applied

Choose an answer or reveal it to continue.

Keyboard shortcuts:A to D or 1 to 4answerEntercheck the answer←→previous or nextFflag for review