def _deconv_matching_assert(self, model, output_layer, test_input):
out = model(test_input)
attrib = Deconvolution(model)
neuron_attrib = NeuronDeconvolution(model, output_layer)
for i in range(out.shape[1]):
deconv_vals = attrib.attribute(test_input, target=i)
neuron_deconv_vals = neuron_attrib.attribute(test_input, (i, ))
assertTensorAlmostEqual(self,
deconv_vals,
neuron_deconv_vals,
delta=0.01)
def _deconv_test_assert(self,
model,
test_input,
expected,
additional_input=None):
deconv = Deconvolution(model)
attributions = deconv.attribute(
test_input, target=0, additional_forward_args=additional_input)
for i in range(len(test_input)):
assertTensorAlmostEqual(self,
attributions[i],
expected[i],
delta=0.01)
def _deconv_test_assert(
self,
model: Module,
test_input: TensorOrTupleOfTensorsGeneric,
expected: Tuple[List[List[float]], ...],
additional_input: Any = None,
) -> None:
deconv = Deconvolution(model)
attributions = deconv.attribute(
test_input, target=0, additional_forward_args=additional_input
)
for i in range(len(test_input)):
assertTensorAlmostEqual(self, attributions[i], expected[i], delta=0.01)
def _deconv_matching_assert(
self,
model: Module,
output_layer: Module,
test_input: TensorOrTupleOfTensorsGeneric,
) -> None:
out = model(test_input)
attrib = Deconvolution(model)
self.assertFalse(attrib.multiplies_by_inputs)
neuron_attrib = NeuronDeconvolution(model, output_layer)
for i in range(out.shape[1]):
deconv_vals = attrib.attribute(test_input, target=i)
neuron_deconv_vals = neuron_attrib.attribute(test_input, (i,))
assertTensorAlmostEqual(self, deconv_vals, neuron_deconv_vals, delta=0.01)