Dear Team,
I have the following Training and validation loss.
However, the grader is telling me that the output is not acceptable
Failed test case: slope of val_loss curve above threshold.
Expected:
0.0005 or lower,
but got:
0.0015098783082531807.
To me this has a similar result than the third picture shown in this plot:
How is the threshold calculated from the grader?
Kind regards and thanks,
Mauricio