CNN training stopped after 8 iterations
19 views (last 30 days)
new_user on 19 Dec 2021
a cnn model in layers structure is showing no error during network analysis but it's stopping at during the training. Showing Training loss is Nan, and stoping after 7-8 iterations only. Can anyone suggest what could be wrong??
Srivardhan Gadila on 27 Dec 2021
If the predictors or the responses contains NaNs, then they are propagated through the network during training. In these cases, the training usually fails to converge. Also, starting from the release R2021b, the trainNetwork automatically stops training when loss is NaN.
You can refer to the following documentation for more information: trainNetwork automatically stops training when loss is NaN, Improve Training Accuracy.