I used label encoder to change strings to integers but when I supply the input to network it shows loss as very large negative number. I used sigmoid activation for output and relu for hidden nodes.
Not able to train my network
Hey @ankur_shuklauR9, you need to use linear/relu as activation for output layer as well. Sigmoid is used in binary classification.
Still not working, I have 2 layers with 32 nodes each. Input shape is (77,) and I am using relu activation for all layers.
Hey @ankur_shuklauR9, you are passing wrong loss function, it should be ‘mse’ and not 'binary_crossentropy
Still not working 
Hey @ankur_shuklauR9, that is due to the reason that mse loss would be quite large and weights backpropagated are at more verge of gradient exploding. Try out different loss function like ‘mean_cross_percentage_entropy’ etc.
You need to try out various functions to get to the final solution.
I hope I’ve cleared your doubt. I ask you to please rate your experience here
Your feedback is very important. It helps us improve our platform and hence provide you
the learning experience you deserve.
On the off chance, you still have some questions or not find the answers satisfactory, you may reopen
the doubt.