I want to build a feed forward-neural.network and train it (to overfitt ) on a small portion of the input features. For this I used dropout regularization as it follows the logic on training on small portion and then test model on the whole features (Turning off during training and turn on test).
But since I am interested on overfitting, I don't think dropout is a good solution for me. So how can I turn off some input nodes during training in the same way as used in Dropout regularization, but this time I don't want randomly turn them off, but chose which featured will be ignored during training?
You can do:
for param in model.parameters():
param.requires_grad = False
to turn off layers in the model
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.