How to print loss with symbol network?


#1

How to print loss with symbol network?


#2

You can use the ‘eval_metric’ parameter of the fit function to print any of the loss metrics listed here. For example, this will print both the Cross Entropy loss and Accuracy:

mod.fit(train_iter, 
              eval_data=val_iter,
              optimizer='sgd',
              optimizer_params={'learning_rate':0.1},
              eval_metric=['ce','acc'],
              batch_end_callback = mx.callback.Speedometer(batch_size, 100), 
              num_epoch=10)

#3

thank you. I am trying it.