In this week, we were taught to create DNN for time series dataset.
I am confused using learning rate schedule.
We used this to plot the Loss for epoch VS Learning rate to determine the min point of learning rate and retrain the model back to obtain better training result.
so in the next cell laurence used 4e-6 instead of 4e-4 anyone have any idea why? since from the graph the renage of min LR are at xe-4.
1 Like
This should be the Tensorflow Developer specializatio not AI for Everyone. See in this image this point corresponds to 4xe-6, which is close to lowest point on the graph and still stable.
The 4e-4 is on right of the graph is after the sharp increase!
1 Like
alright understood.
Thank you.
Are there any guidelines on which tag I should attach based on the course I took on DeepLearning?
1 Like
Yes, choose the specialisation and relevant course when you create the post!
