Scaled Y-Axis for Module Training
complete
S
Schmitty
Currently, the graph showing the loss of a module training scales from 0 to the max value. But in most instances, the loss will only be in a range of 2 to 3. As a result, the slow reduction of loss can barely be seen as most of the graph just becomes a flat line. I suggest changing the range of the graph dynamically.
E
Etienne
complete