promptdojo_

Loss, forward, backward, step — step 4 of 7

ch 43 · training loops, backprop, optimizers, and schedulers4/7

Complete the two-parameter step: the gradient for b drops the trailing x, and both parameters step downhill.

read, then continue.