A unified framework of perturbation and gradient-based attribution methods for Deep Neural Networks interpretability. DeepExplain also includes support for Shapley Values sampling. (ICLR 2018)
Hi,
While trying to run multiple_input_keras.ipynb and https://github.com/marcoancona/DeepExplain/blob/master/examples/mnist_tensorflow.ipynb, I can install and create the models just fine but when it comes time to use DeepExplain as demonstrated in the example folders, I receive this traceback:
line 3343, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "", line 15, in
attributions_gradin = de.explain('grad*input', target_tensor, input_tensor, xs)
line 94, in _set_check_baseline
if self.baseline is None:
AttributeError: 'GradientXInput' object has no attribute 'baseline'
I'm unsure what is happening, for reference my tf version is '1.14.0'. I'm sure it's a silly error but any clarification would be great.
Hi, While trying to run multiple_input_keras.ipynb and https://github.com/marcoancona/DeepExplain/blob/master/examples/mnist_tensorflow.ipynb, I can install and create the models just fine but when it comes time to use DeepExplain as demonstrated in the example folders, I receive this traceback: line 3343, in run_code exec(code_obj, self.user_global_ns, self.user_ns) File "", line 15, in
attributions_gradin = de.explain('grad*input', target_tensor, input_tensor, xs)
line 94, in _set_check_baseline
if self.baseline is None:
AttributeError: 'GradientXInput' object has no attribute 'baseline'
I'm unsure what is happening, for reference my tf version is '1.14.0'. I'm sure it's a silly error but any clarification would be great.
Thanks.