tfm.vision.layers.ReversibleLayer

Creates a reversible layer.

tfm.vision.layers.ReversibleLayer(
    f: tf_keras.layers.Layer,
    g: tf_keras.layers.Layer,
    manual_grads: bool = True,
    **kwargs
)

Computes y1 = x1 + f(x2), y2 = x2 + g(y1), where f and g can be arbitrary layers that are stateless, which in this case are ResidualInner layers.

f A tf_keras.layers.Layer instance of f inner block referred to in paper. Each reversible layer consists of two inner functions. For example, in RevNet the reversible residual consists of two f/g inner (bottleneck) residual functions. Where the input to the reversible layer is x, the input gets partitioned in the channel dimension and the forward pass follows (eq8): x = [x1; x2], z1 = x1 + f(x2), y2 = x2 + g(z1), y1 = stop_gradient(z1).
g A tf_keras.layers.Layer instance of g inner block referred to in paper. Detailed explanation same as above as f arg.
manual_grads A bool [Testing Only] of whether to manually take gradients as in Algorithm 1 or defer to autograd.
**kwargs Additional keyword arguments to be passed.

Methods

call

View source

call(
    inputs: tf.Tensor, training: Optional[bool] = None
) -> tf.Tensor

This is where the layer's logic lives.

The call() method may not create state (except in its first invocation, wrapping the creation of variables or other resources in tf.init_scope()). It is recommended to create state, including tf.Variable instances and nested Layer instances, in __init__(), or in the build() method that is called automatically before call() executes for the first time.

Args
inputs Input tensor, or dict/list/tuple of input tensors. The first positional inputs argument is subject to special rules:

  • inputs must be explicitly passed. A layer cannot have zero arguments, and inputs cannot be provided via the default value of a keyword argument.
  • NumPy array or Python scalar values in inputs get cast as tensors.
  • TF-Keras mask metadata is only collected from inputs.
  • Layers are built (build(input_shape) method) using shape info from inputs only.
  • input_spec compatibility is only checked against inputs.
  • Mixed precision input casting is only applied to inputs. If a layer has tensor arguments in *args or **kwargs, their casting behavior in mixed precision should be handled manually.
  • The SavedModel input specification is generated using inputs only.
  • Integration with various ecosystem packages like TFMOT, TFLite, TF.js, etc is only supported for inputs and not for tensors in positional and keyword arguments.
*args Additional positional arguments. May contain tensors, although this is not recommended, for the reasons above.
**kwargs Additional keyword arguments. May contain tensors, although this is not recommended, for the reasons above. The following optional keyword arguments are reserved:
  • training: Boolean scalar tensor of Python boolean indicating whether the call is meant for training or inference.
  • mask: Boolean input mask. If the layer's call() method takes a mask argument, its default value will be set to the mask generated for inputs by the previous layer (if input did come from a layer that generated a corresponding mask, i.e. if it came from a TF-Keras layer with masking support).

    • Returns
    A tensor or list/tuple of tensors.