NaN training loss

1

Why am I having NaN for training loss, ltrain if I change the variable NUM_OF_CELLS from 8 to 16 ?

image
image980×1050 118 KB

2

Problem solved with some weight regularization