Weights start small and random (biases often at zero). A balloon photo goes forward, the output is compared with the known icing label, the error (loss) is sent back to nudge weights and biases, and one full pass through the training set is an epoch. Training stops when the output is good enough. What working cycle is that?

Short Explanation

Weights start small and random, a photo goes forward, the output is compared with the known label, the loss is sent back to nudge weights and biases, and one full pass is an epoch. That cycle is initialize, forward, loss, adjust, epoch. An epoch is not the independent test.

Full Explanation

Training dynamics are initialize, forward, loss, adjust, and epoch. That is tester understanding, not a job setup, not the holdout test, and not Chapter 6 drift.

×