Closed ensan-hcl closed 2 years ago
I at first considered that changing n_hidden is enough, but perhaps intended behavior here is logging channel_size
. I think result
should be like this.
result = {'lr': learning_rate, 'num_kernel': channel_size,
'val_acc': val_accuracy_list,
'train_acc': train_accuracy_list,
'train_loss': train_loss_list}
Thank you for fixing the bug and PR.
As you pointed out, hidden_size is not the target variable for the parameter search. Therefore, we have merged your PR.
Please note that now we have rearranged notebooks. The new param. search notebook is placed at 02_dnn_simple_pytorch/parameter_search.ipynb. We will merge them after modifing notebook description and python codes.
Best
When running
12_parameter_search.ipynb
, it fails since there is no variablehidden_size
. It seemed that correct name isn_hidden
. so I fixed that point.