in jupyter notebook Examples, when applying integrated_gradient, "mask = inter_grad.get_mask(x[0])" might need to be changed to "mask = inter_grad.GetMask(x[0])" cuz from the current implementation, the class method get_mask is GetMask considering it needs to call the super class's method get_mask. So if call integraded_gradient's get mask method as get_mask just like what has been done in the Examples, it will return the exact same images as vanilla's get_mask.
visual_propagation.py, line 15, "outs = [layer.output for layer in model.layers]" might need to be changed to "outs = [layer.output for layer in model.layers][1:]" to ensure the input layer is different from output layer.
in jupyter notebook Examples, when applying integrated_gradient, "mask = inter_grad.get_mask(x[0])" might need to be changed to "mask = inter_grad.GetMask(x[0])" cuz from the current implementation, the class method get_mask is GetMask considering it needs to call the super class's method get_mask. So if call integraded_gradient's get mask method as get_mask just like what has been done in the Examples, it will return the exact same images as vanilla's get_mask.
visual_propagation.py, line 15, "outs = [layer.output for layer in model.layers]" might need to be changed to "outs = [layer.output for layer in model.layers][1:]" to ensure the input layer is different from output layer.