- hidden layers activation is sigmoid
- output layer activation is softmax
- loss function is cross-entropy
- support save and load model weights
- setting hidden layers by paramater "hidden_sizes", examples: hidden_sizes=(100,),there is a hidden layer of 100 neurons; hidden_sizes=(100,50),there are two hidden layers, 100 and 200 neurons.
- Download Data: