from keras import backend as K
get_3rd_layer_output = K.function([model.layers[0].input, K.learning_phase()],
[model.layers[3].output])
# output in test mode = 0
layer_output = get_3rd_layer_output([x, 0])[0]
# output in train mode = 1
layer_output = get_3rd_layer_output([x, 1])[0]
My question is there a way of acheiving the same in chainer?
Your kind help is highly appreciated. Thank you.
Abdullahi Mohammad