opt <-> optimizer parameter

[learniotai]

The function

def train_batch(x, y, model, opt, loss_fn):
    prediction = model(x)[0]
    batch_loss = loss_fn(prediction, y)
    batch_loss.backward()
    optimizer.step()
    optimizer.zero_grad()

has parameter opt and then uses optimizer which isn't passed in so it must be using a global value.

[19valentin99]

Just to mention: it is useful to mention where in the book you encountered this error (at least the Chapter)

[Camaltra]

Seen on the Chap4 (CNN_working_details.ipynb)