#!/usr/bin/env python3
# pyre-strict
from typing import Any, Callable, Dict, List, Optional, Tuple, Union
import torch
import torch.nn.functional as F
from captum._utils.common import (
_format_additional_forward_args,
_format_output,
_format_tensor_into_tuples,
)
from captum._utils.gradient import compute_layer_gradients_and_eval
from captum._utils.typing import TargetType
from captum.attr._utils.attribution import GradientAttribution, LayerAttribution
from captum.log import log_usage
from torch import Tensor
from torch.nn import Module
[docs]
class LayerGradCam(LayerAttribution, GradientAttribution):
r"""
Computes GradCAM attribution for chosen layer. GradCAM is designed for
convolutional neural networks, and is usually applied to the last
convolutional layer.
GradCAM computes the gradients of the target output with respect to
the given layer, averages for each output channel (dimension 2 of
output), and multiplies the average gradient for each channel by the
layer activations. The results are summed over all channels.
Note that in the original GradCAM algorithm described in the paper,
ReLU is applied to the output, returning only non-negative attributions.
For providing more flexibility to the user, we choose to not perform the
ReLU internally by default and return the sign information. To match the
original GradCAM algorithm, it is necessary to pass the parameter
relu_attributions=True to apply ReLU on the final
attributions or alternatively only visualize the positive attributions.
Note: this procedure sums over the second dimension (# of channels),
so the output of GradCAM attributions will have a second
dimension of 1, but all other dimensions will match that of the layer
output.
GradCAM attributions are generally upsampled and can be viewed as a
mask to the input, since a convolutional layer output generally
matches the input image spatially. This upsampling can be performed
using LayerAttribution.interpolate, as shown in the example below.
More details regarding the GradCAM method can be found in the
original paper here:
https://arxiv.org/abs/1610.02391
"""
def __init__(
self,
# pyre-fixme[24]: Generic type `Callable` expects 2 type parameters.
forward_func: Callable,
layer: Module,
device_ids: Union[None, List[int]] = None,
) -> None:
r"""
Args:
forward_func (Callable): The forward function of the model or any
modification of it
layer (torch.nn.Module): Layer for which attributions are computed.
Output size of attribute matches this layer's output
dimensions, except for dimension 2, which will be 1,
since GradCAM sums over channels.
device_ids (list[int]): Device ID list, necessary only if forward_func
applies a DataParallel model. This allows reconstruction of
intermediate outputs from batched results across devices.
If forward_func is given as the DataParallel model itself,
then it is not necessary to provide this argument.
"""
LayerAttribution.__init__(self, forward_func, layer, device_ids)
GradientAttribution.__init__(self, forward_func)
[docs]
@log_usage()
def attribute(
self,
inputs: Union[Tensor, Tuple[Tensor, ...]],
target: TargetType = None,
additional_forward_args: Optional[object] = None,
attribute_to_layer_input: bool = False,
relu_attributions: bool = False,
attr_dim_summation: bool = True,
grad_kwargs: Optional[Dict[str, Any]] = None,
) -> Union[Tensor, Tuple[Tensor, ...]]:
r"""
Args:
inputs (Tensor or tuple[Tensor, ...]): 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
attr_dim_summation (bool, optional): Indicates whether to
sum attributions along dimension 1 (usually channel).
The default (True) means to sum along dimension 1.
Default: True
grad_kwargs (Dict[str, Any], optional): Additional keyword
arguments for torch.autograd.grad.
Default: None
Returns:
*Tensor* or *tuple[Tensor, ...]* of **attributions**:
- **attributions** (*Tensor* or *tuple[Tensor, ...]*):
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 if
the layer inputs / outputs contain multiple tensors,
otherwise a single tensor is returned.
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_tensor_into_tuples(inputs)
additional_forward_args = _format_additional_forward_args(
additional_forward_args
)
# Returns gradient of output with respect to
# hidden layer and hidden layer evaluated at each input.
layer_gradients, layer_evals = 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,
grad_kwargs=grad_kwargs,
)
summed_grads = tuple(
(
torch.mean(
# pyre-fixme[6]: For 1st argument expected `Tensor` but got
# `Tuple[Tensor, ...]`.
layer_grad,
# pyre-fixme[16]: `tuple` has no attribute `shape`.
dim=tuple(x for x in range(2, len(layer_grad.shape))),
keepdim=True,
)
if len(layer_grad.shape) > 2
else layer_grad
)
for layer_grad in layer_gradients
)
if attr_dim_summation:
scaled_acts = tuple(
# pyre-fixme[58]: `*` is not supported for operand types
# `Union[tuple[torch._tensor.Tensor], torch._tensor.Tensor]` and
# `Tuple[Tensor, ...]`.
# pyre-fixme[6]: For 1st argument expected `Tensor` but got
# `Tuple[Tensor, ...]`.
torch.sum(summed_grad * layer_eval, dim=1, keepdim=True)
for summed_grad, layer_eval in zip(summed_grads, layer_evals)
)
else:
scaled_acts = tuple(
# pyre-fixme[58]: `*` is not supported for operand types
# `Union[tuple[torch._tensor.Tensor], torch._tensor.Tensor]` and
# `Tuple[Tensor, ...]`.
summed_grad * layer_eval
for summed_grad, layer_eval in zip(summed_grads, layer_evals)
)
if relu_attributions:
# pyre-fixme[6]: For 1st argument expected `Tensor` but got
# `Union[tuple[Tensor], Tensor]`.
scaled_acts = tuple(F.relu(scaled_act) for scaled_act in scaled_acts)
# pyre-fixme[6]: For 2nd argument expected `Tuple[Tensor, ...]` but got
# `Tuple[Union[tuple[Tensor], Tensor], ...]`.
return _format_output(len(scaled_acts) > 1, scaled_acts)
