This error is probably without any consequence, but why not correct it?
In the very last cell of the notebook lesson4-mnist_sgd.ipynb,
loss_fn = nn.NLLLoss()
is never used, and instead a previous
loss
object is used.
This error is probably without any consequence, but why not correct it?
In the very last cell of the notebook lesson4-mnist_sgd.ipynb,
loss_fn = nn.NLLLoss()
is never used, and instead a previous
loss
object is used.
And losses.append(loss)
should be replaced with losses.append(l)
, as in previous parts of the notebook.
Alternatively, keep loss_fn
and make the appropriate changes inside the loop.