C++ TensorFlow SoftmaxCrossEntropWithLogits returns (cost, gradients), how to access cost?

Question

I am trying to implement a simple neural network in C++ TensorFlow.

I am unable to access loss returned by SoftmaxCrossEntropyWithLogits function ( https://www.tensorflow.org/api_docs/cc/class/tensorflow/ops/softmax-cross-entropy-with-logits ).

Please check the "Returns" segment in above mentioned link. I want to access only "loss" output. I am unable to write this in C++.

Thanks

Answer 1

here it doesn't return anything, its an attribute. You can access it from the class instance.

#define the loss class instance with your logits and labels
softmax_loss_function = SoftmaxCrossEntropyWithLogits(const ::tensorflow::Scope & scope, ::tensorflow::Input features, ::tensorflow::Input labels) 
# access the loss
loss = softmax_loss_function.loss