Hi,
In lesson 1, while the model learns, it outputs 2 tables which have identical headers (epoch, train_loss,…). The first table has only epoch 0, and the second table goes from epoch 0 to as many epochs that are defined. What is the difference between epoch 0 in the first table and epoch 0 in the second table ?
Two tables, initial frozen epoch/s followed by the unfrozen epochs.
Fine tune with
Learner.freeze
forfreeze_epochs
, then withLearner.unfreeze
forepochs
, using discriminative LR.
More on Unfreezing and Transfer Learning
in the fastai book CH05_pet_breeds
1 Like
Thanks.