def attribute( # type: ignore self, inputs: TensorOrTupleOfTensorsGeneric, baselines: Union[TensorOrTupleOfTensorsGeneric, Callable[..., TensorOrTupleOfTensorsGeneric]], target: TargetType = None, additional_forward_args: Any = None, return_convergence_delta: bool = False, custom_attribution_func: Union[None, Callable[..., Tuple[Tensor, ...]]] = None, ) -> Union[TensorOrTupleOfTensorsGeneric, Tuple[ TensorOrTupleOfTensorsGeneric, Tensor]]: r""" Args: inputs (tensor or tuple of tensors): Input for which attributions are computed. If forward_func takes a single tensor as input, a single input tensor should be provided. If forward_func takes multiple tensors as input, a tuple of the input tensors should be provided. It is assumed that for all given input tensors, dimension 0 corresponds to the number of examples (aka batch size), and if multiple input tensors are provided, the examples must be aligned appropriately. baselines (tensor, tuple of tensors, callable): Baselines define reference samples that are compared with the inputs. In order to assign attribution scores DeepLift computes the differences between the inputs/outputs and corresponding references. Baselines can be provided as: - a single tensor, if inputs is a single tensor, with the first dimension equal to the number of examples in the baselines' distribution. The remaining dimensions must match with input tensor's dimension starting from the second dimension. - a tuple of tensors, if inputs is a tuple of tensors, with the first dimension of any tensor inside the tuple equal to the number of examples in the baseline's distribution. The remaining dimensions must match the dimensions of the corresponding input tensor starting from the second dimension. - callable function, optionally takes `inputs` as an argument and either returns a single tensor or a tuple of those. It is recommended that the number of samples in the baselines' tensors is larger than one. target (int, tuple, tensor or list, optional): Output indices for which gradients are computed (for classification cases, this is usually the target class). If the network returns a scalar value per example, no target index is necessary. For general 2D outputs, targets can be either: - a single integer or a tensor containing a single integer, which is applied to all input examples - a list of integers or a 1D tensor, with length matching the number of examples in inputs (dim 0). Each integer is applied as the target for the corresponding example. For outputs with > 2 dimensions, targets can be either: - A single tuple, which contains #output_dims - 1 elements. This target index is applied to all examples. - A list of tuples with length equal to the number of examples in inputs (dim 0), and each tuple containing #output_dims - 1 elements. Each tuple is applied as the target for the corresponding example. Default: None additional_forward_args (any, optional): If the forward function requires additional arguments other than the inputs for which attributions should not be computed, this argument can be provided. It must be either a single additional argument of a Tensor or arbitrary (non-tuple) type or a tuple containing multiple additional arguments including tensors or any arbitrary python types. These arguments are provided to forward_func in order, following the arguments in inputs. Note that attributions are not computed with respect to these arguments. Default: None return_convergence_delta (bool, optional): Indicates whether to return convergence delta or not. If `return_convergence_delta` is set to True convergence delta will be returned in a tuple following attributions. Default: False custom_attribution_func (callable, optional): A custom function for computing final attribution scores. This function can take at least one and at most three arguments with the following signature: - custom_attribution_func(multipliers) - custom_attribution_func(multipliers, inputs) - custom_attribution_func(multipliers, inputs, baselines) In case this function is not provided we use the default logic defined as: multipliers * (inputs - baselines) It is assumed that all input arguments, `multipliers`, `inputs` and `baselines` are provided in tuples of same length. `custom_attribution_func` returns a tuple of attribution tensors that have the same length as the `inputs`. Default: None Returns: **attributions** or 2-element tuple of **attributions**, **delta**: - **attributions** (*tensor* or tuple of *tensors*): Attribution score computed based on DeepLift rescale rule with respect to each input feature. Attributions will always be the same size as the provided inputs, with each value providing the attribution of the corresponding input index. If a single tensor is provided as inputs, a single tensor is returned. If a tuple is provided for inputs, a tuple of corresponding sized tensors is returned. - **delta** (*tensor*, returned if return_convergence_delta=True): This is computed using the property that the total sum of forward_func(inputs) - forward_func(baselines) must be very close to the total sum of attributions computed based on approximated SHAP values using Deeplift's rescale rule. Delta is calculated for each example input and baseline pair, meaning that the number of elements in returned delta tensor is equal to the `number of examples in input` * `number of examples in baseline`. The deltas are ordered in the first place by input example, followed by the baseline. Note that the logic described for deltas is guaranteed when the default logic for attribution computations is used, meaning that the `custom_attribution_func=None`, otherwise it is not guaranteed and depends on the specifics of the `custom_attribution_func`. Examples:: >>> # ImageClassifier takes a single input tensor of images Nx3x32x32, >>> # and returns an Nx10 tensor of class probabilities. >>> net = ImageClassifier() >>> dl = DeepLiftShap(net) >>> input = torch.randn(2, 3, 32, 32, requires_grad=True) >>> # Computes shap values using deeplift for class 3. >>> attribution = dl.attribute(input, target=3) """ baselines = _format_callable_baseline(baselines, inputs) assert isinstance( baselines[0], torch.Tensor ) and baselines[0].shape[0] > 1, ( "Baselines distribution has to be provided in form of a torch.Tensor" " with more than one example but found: {}." " If baselines are provided in shape of scalars or with a single" " baseline example, `DeepLift`" " approach can be used instead.".format(baselines[0])) # Keeps track whether original input is a tuple or not before # converting it into a tuple. is_inputs_tuple = _is_tuple(inputs) inputs = _format_input(inputs) # batch sizes inp_bsz = inputs[0].shape[0] base_bsz = baselines[0].shape[0] ( exp_inp, exp_base, exp_tgt, exp_addit_args, ) = self._expand_inputs_baselines_targets(baselines, inputs, target, additional_forward_args) attributions = super().attribute( exp_inp, exp_base, target=exp_tgt, additional_forward_args=exp_addit_args, return_convergence_delta=cast(Literal[True, False], return_convergence_delta), custom_attribution_func=custom_attribution_func, ) if return_convergence_delta: attributions, delta = cast(Tuple[Tuple[Tensor, ...], Tensor], attributions) attributions = tuple( self._compute_mean_across_baselines(inp_bsz, base_bsz, cast(Tensor, attribution)) for attribution in attributions) if return_convergence_delta: return _format_attributions(is_inputs_tuple, attributions), delta else: return _format_attributions(is_inputs_tuple, attributions)
def attribute( self, inputs: Union[Tensor, Tuple[Tensor, ...]], baselines: Optional[ Union[Tensor, int, float, Tuple[Union[Tensor, int, float], ...]] ] = None, target: Optional[ Union[int, Tuple[int, ...], Tensor, List[Tuple[int, ...]]] ] = None, additional_forward_args: Any = None, n_steps: int = 50, method: str = "gausslegendre", internal_batch_size: Optional[int] = None, return_convergence_delta: bool = False, attribute_to_layer_input: bool = False, ) -> Union[ Tensor, Tuple[Tensor, ...], Tuple[Union[Tensor, Tuple[Tensor, ...]], Tensor] ]: r""" This method attributes the output of the model with given target index (in case it is provided, otherwise it assumes that output is a scalar) to layer inputs or outputs of the model, depending on whether `attribute_to_layer_input` is set to True or False, using the approach described above. In addition to that it also returns, if `return_convergence_delta` is set to True, integral approximation delta based on the completeness property of integrated gradients. Args: inputs (tensor or tuple of tensors): Input for which layer integrated gradients are computed. If forward_func takes a single tensor as input, a single input tensor should be provided. If forward_func takes multiple tensors as input, a tuple of the input tensors should be provided. It is assumed that for all given input tensors, dimension 0 corresponds to the number of examples, and if multiple input tensors are provided, the examples must be aligned appropriately. baselines (scalar, tensor, tuple of scalars or tensors, optional): Baselines define the starting point from which integral is computed and can be provided as: - a single tensor, if inputs is a single tensor, with exactly the same dimensions as inputs or the first dimension is one and the remaining dimensions match with inputs. - a single scalar, if inputs is a single tensor, which will be broadcasted for each input value in input tensor. - a tuple of tensors or scalars, the baseline corresponding to each tensor in the inputs' tuple can be: - either a tensor with matching dimensions to corresponding tensor in the inputs' tuple or the first dimension is one and the remaining dimensions match with the corresponding input tensor. - or a scalar, corresponding to a tensor in the inputs' tuple. This scalar value is broadcasted for corresponding input tensor. In the cases when `baselines` is not provided, we internally use zero scalar corresponding to each input tensor. Default: None target (int, tuple, tensor or list, optional): Output indices for which gradients are computed (for classification cases, this is usually the target class). If the network returns a scalar value per example, no target index is necessary. For general 2D outputs, targets can be either: - a single integer or a tensor containing a single integer, which is applied to all input examples - a list of integers or a 1D tensor, with length matching the number of examples in inputs (dim 0). Each integer is applied as the target for the corresponding example. For outputs with > 2 dimensions, targets can be either: - A single tuple, which contains #output_dims - 1 elements. This target index is applied to all examples. - A list of tuples with length equal to the number of examples in inputs (dim 0), and each tuple containing #output_dims - 1 elements. Each tuple is applied as the target for the corresponding example. Default: None additional_forward_args (any, optional): If the forward function requires additional arguments other than the inputs for which attributions should not be computed, this argument can be provided. It must be either a single additional argument of a Tensor or arbitrary (non-tuple) type or a tuple containing multiple additional arguments including tensors or any arbitrary python types. These arguments are provided to forward_func in order following the arguments in inputs. For a tensor, the first dimension of the tensor must correspond to the number of examples. It will be repeated for each of `n_steps` along the integrated path. For all other types, the given argument is used for all forward evaluations. Note that attributions are not computed with respect to these arguments. Default: None n_steps (int, optional): The number of steps used by the approximation method. Default: 50. method (string, optional): Method for approximating the integral, one of `riemann_right`, `riemann_left`, `riemann_middle`, `riemann_trapezoid` or `gausslegendre`. Default: `gausslegendre` if no method is provided. internal_batch_size (int, optional): Divides total #steps * #examples data points into chunks of size internal_batch_size, which are computed (forward / backward passes) sequentially. For DataParallel models, each batch is split among the available devices, so evaluations on each available device contain internal_batch_size / num_devices examples. If internal_batch_size is None, then all evaluations are processed in one batch. Default: None return_convergence_delta (bool, optional): Indicates whether to return convergence delta or not. If `return_convergence_delta` is set to True convergence delta will be returned in a tuple following attributions. Default: False attribute_to_layer_input (bool, optional): Indicates whether to compute the attribution with respect to the layer input or output. If `attribute_to_layer_input` is set to True then the attributions will be computed with respect to layer input, otherwise it will be computed with respect to layer output. Note that currently it is assumed that either the input or the output of internal layer, depending on whether we attribute to the input or output, is a single tensor. Support for multiple tensors will be added later. Default: False Returns: **attributions** or 2-element tuple of **attributions**, **delta**: - **attributions** (*tensor* or tuple of *tensors*): Integrated gradients with respect to `layer`'s inputs or outputs. Attributions will always be the same size and dimensionality as the input or output of the given layer, depending on whether we attribute to the inputs or outputs of the layer which is decided by the input flag `attribute_to_layer_input`. - **delta** (*tensor*, returned if return_convergence_delta=True): The difference between the total approximated and true integrated gradients. This is computed using the property that the total sum of forward_func(inputs) - forward_func(baselines) must equal the total sum of the integrated gradient. Delta is calculated per example, meaning that the number of elements in returned delta tensor is equal to the number of of examples in inputs. Examples:: >>> # ImageClassifier takes a single input tensor of images Nx3x32x32, >>> # and returns an Nx10 tensor of class probabilities. >>> # It contains an attribute conv1, which is an instance of nn.conv2d, >>> # and the output of this layer has dimensions Nx12x32x32. >>> net = ImageClassifier() >>> lig = LayerIntegratedGradients(net, net.conv1) >>> input = torch.randn(2, 3, 32, 32, requires_grad=True) >>> # Computes layer integrated gradients for class 3. >>> # attribution size matches layer output, Nx12x32x32 >>> attribution = lig.attribute(input, target=3) """ inps, baselines = _format_input_baseline(inputs, baselines) _validate_input(inps, baselines, n_steps, method) baselines = _tensorize_baseline(inps, baselines) additional_forward_args = _format_additional_forward_args( additional_forward_args ) if self.device_ids is None: self.device_ids = getattr(self.forward_func, "device_ids", None) inputs_layer, is_layer_tuple = _forward_layer_eval( self.forward_func, inps, self.layer, device_ids=self.device_ids, additional_forward_args=additional_forward_args, attribute_to_layer_input=attribute_to_layer_input, ) baselines_layer, _ = _forward_layer_eval( self.forward_func, baselines, self.layer, device_ids=self.device_ids, additional_forward_args=additional_forward_args, attribute_to_layer_input=attribute_to_layer_input, ) # inputs -> these inputs are scaled def gradient_func( forward_fn: Callable, inputs: Union[Tensor, Tuple[Tensor, ...]], target_ind: Optional[ Union[int, Tuple[int, ...], Tensor, List[Tuple[int, ...]]] ] = None, additional_forward_args: Any = None, ) -> Tuple[Tensor, ...]: if self.device_ids is None: scattered_inputs = (inputs,) else: # scatter method does not have a precise enough return type in its # stub, so suppress the type warning. scattered_inputs = scatter( # type:ignore inputs, target_gpus=self.device_ids ) scattered_inputs_dict = { scattered_input[0].device: scattered_input for scattered_input in scattered_inputs } with torch.autograd.set_grad_enabled(True): def layer_forward_hook(module, hook_inputs, hook_outputs=None): device = _extract_device(module, hook_inputs, hook_outputs) if is_layer_tuple: return scattered_inputs_dict[device] return scattered_inputs_dict[device][0] if attribute_to_layer_input: hook = self.layer.register_forward_pre_hook(layer_forward_hook) else: hook = self.layer.register_forward_hook(layer_forward_hook) output = _run_forward( self.forward_func, additional_forward_args, target_ind, ) hook.remove() assert output[0].numel() == 1, ( "Target not provided when necessary, cannot" " take gradient with respect to multiple outputs." ) # torch.unbind(forward_out) is a list of scalar tensor tuples and # contains batch_size * #steps elements grads = torch.autograd.grad(torch.unbind(output), inputs) return grads self.ig.gradient_func = gradient_func all_inputs = ( (inps + additional_forward_args) if additional_forward_args is not None else inps ) attributions = self.ig.attribute( inputs_layer, baselines=baselines_layer, target=target, additional_forward_args=all_inputs, n_steps=n_steps, method=method, internal_batch_size=internal_batch_size, return_convergence_delta=False, ) if return_convergence_delta: start_point, end_point = baselines, inps # computes approximation error based on the completeness axiom delta = self.compute_convergence_delta( attributions, start_point, end_point, additional_forward_args=additional_forward_args, target=target, ) return _format_attributions(is_layer_tuple, attributions), delta return _format_attributions(is_layer_tuple, attributions)
def attribute( self, inputs: Union[Tensor, Tuple[Tensor, ...]], target: TargetType = None, additional_forward_args: Any = None, attribute_to_layer_input: bool = False, relu_attributions: bool = False, ) -> Union[Tensor, Tuple[Tensor, ...]]: r""" Args: inputs (tensor or tuple of tensors): Input for which attributions are computed. If forward_func takes a single tensor as input, a single input tensor should be provided. If forward_func takes multiple tensors as input, a tuple of the input tensors should be provided. It is assumed that for all given input tensors, dimension 0 corresponds to the number of examples, and if multiple input tensors are provided, the examples must be aligned appropriately. target (int, tuple, tensor or list, optional): Output indices for which gradients are computed (for classification cases, this is usually the target class). If the network returns a scalar value per example, no target index is necessary. For general 2D outputs, targets can be either: - a single integer or a tensor containing a single integer, which is applied to all input examples - a list of integers or a 1D tensor, with length matching the number of examples in inputs (dim 0). Each integer is applied as the target for the corresponding example. For outputs with > 2 dimensions, targets can be either: - A single tuple, which contains #output_dims - 1 elements. This target index is applied to all examples. - A list of tuples with length equal to the number of examples in inputs (dim 0), and each tuple containing #output_dims - 1 elements. Each tuple is applied as the target for the corresponding example. Default: None additional_forward_args (any, optional): If the forward function requires additional arguments other than the inputs for which attributions should not be computed, this argument can be provided. It must be either a single additional argument of a Tensor or arbitrary (non-tuple) type or a tuple containing multiple additional arguments including tensors or any arbitrary python types. These arguments are provided to forward_func in order following the arguments in inputs. Note that attributions are not computed with respect to these arguments. Default: None attribute_to_layer_input (bool, optional): Indicates whether to compute the attributions with respect to the layer input or output. If `attribute_to_layer_input` is set to True then the attributions will be computed with respect to the layer input, otherwise it will be computed with respect to layer output. Note that currently it is assumed that either the input or the outputs of internal layers, depending on whether we attribute to the input or output, are single tensors. Support for multiple tensors will be added later. Default: False relu_attributions (bool, optional): Indicates whether to apply a ReLU operation on the final attribution, returning only non-negative attributions. Setting this flag to True matches the original GradCAM algorithm, otherwise, by default, both positive and negative attributions are returned. Default: False Returns: *tensor* or tuple of *tensors* of **attributions**: - **attributions** (*tensor* or tuple of *tensors*): Attributions based on GradCAM method. Attributions will be the same size as the output of the given layer, except for dimension 2, which will be 1 due to summing over channels. Attributions are returned in a tuple based on whether the layer inputs / outputs are contained in a tuple from a forward hook. For standard modules, inputs of a single tensor are usually wrapped in a tuple, while outputs of a single tensor are not. Examples:: # >>> # ImageClassifier takes a single input tensor of images Nx3x32x32, # >>> # and returns an Nx10 tensor of class probabilities. # >>> # It contains a layer conv4, which is an instance of nn.conv2d, # >>> # and the output of this layer has dimensions Nx50x8x8. # >>> # It is the last convolution layer, which is the recommended # >>> # use case for GradCAM. # >>> net = ImageClassifier() # >>> layer_gc = LayerGradCam(net, net.conv4) # >>> input = torch.randn(2, 3, 32, 32, requires_grad=True) # >>> # Computes layer GradCAM for class 3. # >>> # attribution size matches layer output except for dimension # >>> # 1, so dimensions of attr would be Nx1x8x8. # >>> attr = layer_gc.attribute(input, 3) # >>> # GradCAM attributions are often upsampled and viewed as a # >>> # mask to the input, since the convolutional layer output # >>> # spatially matches the original input image. # >>> # This can be done with LayerAttribution's interpolate method. # >>> upsampled_attr = LayerAttribution.interpolate(attr, (32, 32)) """ inputs = _format_input(inputs) additional_forward_args = _format_additional_forward_args( additional_forward_args) gradient_mask = apply_gradient_requirements(inputs) # Returns gradient of output with respect to # hidden layer and hidden layer evaluated at each input. layer_gradients, layer_evals, is_layer_tuple = compute_layer_gradients_and_eval( self.forward_func, self.layer, inputs, target, additional_forward_args, device_ids=self.device_ids, attribute_to_layer_input=attribute_to_layer_input, ) undo_gradient_requirements(inputs, gradient_mask) # Gradient Calculation end # what I add: shape from PyG to General PyTorch layer_gradients = tuple( layer_grad.transpose(0, 1).unsqueeze(0) for layer_grad in layer_gradients) layer_evals = tuple( layer_eval.transpose(0, 1).unsqueeze(0) for layer_eval in layer_evals) # end summed_grads = tuple( torch.mean( layer_grad, dim=tuple(x for x in range(2, len(layer_grad.shape))), keepdim=True, ) for layer_grad in layer_gradients) scaled_acts = tuple( torch.sum(summed_grad * layer_eval, dim=1, keepdim=True) for summed_grad, layer_eval in zip(summed_grads, layer_evals)) if relu_attributions: scaled_acts = tuple( F.relu(scaled_act) for scaled_act in scaled_acts) # what I add: shape from General PyTorch to PyG scaled_acts = tuple( scaled_act.squeeze(0).transpose(0, 1) for scaled_act in scaled_acts) # end return _format_attributions(is_layer_tuple, scaled_acts)