def __init__(self, transform_fn, pretransformed_input, dtype=None,
shape=NONE_SPECIFIED, name=None):
"""Creates the `DeferredTensor` object.
Args:
transform_fn: Python `callable` taking `pretransformed_input` and
returning a `Tensor` (representing by this object).
pretransformed_input: object with `shape`, `dtype` properties (typically a
`tf.Variable`) passed into `transform_fn` when this object is acted upon
in a `Tensor` context, eg, `tf.convert_to_tensor`, `+`, `tf.math.exp`,
etc.
dtype: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).dtype`.
Default value: `None` (i.e., `pretransformed_input.dtype`).
shape: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).shape`.
Default value: `'None'` (i.e., `pretransformed_input.shape`).
name: Python `str` representing this object's `name`; used only in graph
mode.
Default value: `None` (i.e.,
`transform_fn.__name__ + '_' + pretransformed_input.name`).
Raises:
TypeError: if `transform_fn` is not `callable`.
TypeError: if `pretransformed_input` lacks `dtype` and/or `shape`
properties (and `dtype` and/or `shape` arguments are unspecified).
"""
if not callable(transform_fn):
raise TypeError('Argument `transform_fn` must be a Python `callable`.')
if ((dtype is None and not hasattr(pretransformed_input, 'dtype')) or
(shape is None and not hasattr(pretransformed_input, 'shape'))):
raise TypeError('Argument `pretransformed_input` must have `dtype` and '
'`shape` properties (unless `dtype`, `shape` arguments '
'are explicitly provided.')
has_name = bool(name)
if not has_name:
name = '_'.join([
transform_fn.__name__,
getattr(pretransformed_input, 'name', '')])
name = name_util.strip_invalid_chars(name)
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
super(DeferredTensor, self).__init__(name=name)
self._name = name
self._transform_fn = transform_fn
self._pretransformed_input = pretransformed_input
self._dtype = dtype_util.base_dtype(dtype or pretransformed_input.dtype)
self._shape = tf.TensorShape(
pretransformed_input.shape if shape == 'None' else shape)
# Secret handshake with tf.is_tensor to return True for DT.
#
# Works around an exception in LinearOperator (which in 2.0.0 checks only
# `tf.is_tensor`, not also `linear_operator_util.is_ref`:
# ValueError: Graph parent item 0 is not a Tensor;
# <DeferredTensor: dtype=float32, shape=[2], fn=exp>.
# TODO(b/140157055): Remove this shim after LinOp is patched in 2.0.
self.is_tensor_like = True
def __init__(self,
transform_fn,
pretransformed_input,
dtype=None,
shape=NONE_SPECIFIED,
name=None):
"""Creates the `DeferredTensor` object.
Args:
transform_fn: Python `callable` taking `pretransformed_input` and
returning a `Tensor` (representing by this object).
pretransformed_input: object with `shape`, `dtype` properties (typically a
`tf.Variable`) passed into `transform_fn` when this object is acted upon
in a `Tensor` context, eg, `tf.convert_to_tensor`, `+`, `tf.math.exp`,
etc.
dtype: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).dtype`.
Default value: `None` (i.e., `pretransformed_input.dtype`).
shape: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).shape`.
Default value: `'None'` (i.e., `pretransformed_input.shape`).
name: Python `str` representing this object's `name`; used only in graph
mode.
Default value: `None` (i.e.,
`transform_fn.__name__ + '_' + pretransformed_input.name`).
Raises:
TypeError: if `transform_fn` is not `callable`.
TypeError: if `pretransformed_input` lacks `dtype` and/or `shape`
properties (and `dtype` and/or `shape` arguments are unspecified).
"""
if not callable(transform_fn):
raise TypeError(
'Argument `transform_fn` must be a Python `callable`.')
if ((dtype is None and not hasattr(pretransformed_input, 'dtype')) or
(shape is None and not hasattr(pretransformed_input, 'shape'))):
raise TypeError(
'Argument `pretransformed_input` must have `dtype` and '
'`shape` properties (unless `dtype`, `shape` arguments '
'are explicitly provided.')
has_name = bool(name)
if not has_name:
name = '_'.join([
transform_fn.__name__,
getattr(pretransformed_input, 'name', '')
])
name = name_util.strip_invalid_chars(name)
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
super(DeferredTensor, self).__init__(name=name)
self._name = name
self._transform_fn = transform_fn
self._pretransformed_input = pretransformed_input
self._dtype = dtype_util.base_dtype(dtype
or pretransformed_input.dtype)
self._shape = tf.TensorShape(pretransformed_input.shape if shape ==
'None' else shape)
def __init__(self, also_track=None, name=None):
name = name_util.strip_invalid_chars(name or type(self).__name__)
self._also_track = [] if also_track is None else [also_track]
super(Layer, self).__init__(name=name)
self._extra_loss = None
self._extra_result = None
self._trace = False
def __init__(self, name=None):
"""Creates the ExponentialFamily.
Args:
name: Python `str` used as TF namescope for ops created by member
functions. Default value: `None` (i.e., the subclass name).
"""
if not name:
name = type(self).__name__
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
super(ExponentialFamily, self).__init__(name=name)
def __init__(self,
graph_parents=None,
is_constant_jacobian=False,
validate_args=False,
dtype=None,
forward_min_event_ndims=None,
inverse_min_event_ndims=None,
parameters=None,
name=None):
"""Constructs Bijector.
A `Bijector` transforms random variables into new random variables.
Examples:
```python
# Create the Y = g(X) = X transform.
identity = Identity()
# Create the Y = g(X) = exp(X) transform.
exp = Exp()
```
See `Bijector` subclass docstring for more details and specific examples.
Args:
graph_parents: Python list of graph prerequisites of this `Bijector`.
is_constant_jacobian: Python `bool` indicating that the Jacobian matrix is
not a function of the input.
validate_args: Python `bool`, default `False`. Whether to validate input
with asserts. If `validate_args` is `False`, and the inputs are invalid,
correct behavior is not guaranteed.
dtype: `tf.dtype` supported by this `Bijector`. `None` means dtype is not
enforced.
forward_min_event_ndims: Python `integer` indicating the minimum number of
dimensions `forward` operates on.
inverse_min_event_ndims: Python `integer` indicating the minimum number of
dimensions `inverse` operates on. Will be set to
`forward_min_event_ndims` by default, if no value is provided.
parameters: Python `dict` of parameters used to instantiate this
`Bijector`.
name: The name to give Ops created by the initializer.
Raises:
ValueError: If neither `forward_min_event_ndims` and
`inverse_min_event_ndims` are specified, or if either of them is
negative.
ValueError: If a member of `graph_parents` is not a `Tensor`.
"""
if not name:
name = type(self).__name__
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
super(Bijector, self).__init__(name=name)
self._name = name
self._parameters = self._no_dependency(parameters)
self._graph_parents = self._no_dependency(graph_parents or [])
self._is_constant_jacobian = is_constant_jacobian
self._validate_args = validate_args
self._dtype = dtype
self._initial_parameter_control_dependencies = tuple(
d for d in self._parameter_control_dependencies(is_init=True)
if d is not None)
if self._initial_parameter_control_dependencies:
self._initial_parameter_control_dependencies = (
tf.group(*self._initial_parameter_control_dependencies),)
if forward_min_event_ndims is None and inverse_min_event_ndims is None:
raise ValueError('Must specify at least one of `forward_min_event_ndims` '
'and `inverse_min_event_ndims`.')
elif inverse_min_event_ndims is None:
inverse_min_event_ndims = forward_min_event_ndims
elif forward_min_event_ndims is None:
forward_min_event_ndims = inverse_min_event_ndims
if not isinstance(forward_min_event_ndims, int):
raise TypeError('Expected forward_min_event_ndims to be of '
'type int, got {}'.format(
type(forward_min_event_ndims).__name__))
if not isinstance(inverse_min_event_ndims, int):
raise TypeError('Expected inverse_min_event_ndims to be of '
'type int, got {}'.format(
type(inverse_min_event_ndims).__name__))
if forward_min_event_ndims < 0:
raise ValueError('forward_min_event_ndims must be a non-negative '
'integer.')
if inverse_min_event_ndims < 0:
raise ValueError('inverse_min_event_ndims must be a non-negative '
'integer.')
self._forward_min_event_ndims = forward_min_event_ndims
self._inverse_min_event_ndims = inverse_min_event_ndims
for i, t in enumerate(self._graph_parents):
if t is None or not tf.is_tensor(t):
raise ValueError('Graph parent item %d is not a Tensor; %s.' % (i, t))
# Setup caching after everything else is done.
self._cache = self._setup_cache()
def __init__(self, also_track=None, validate_args=False, name=None):
name = name_util.strip_invalid_chars(name or type(self).__name__)
self._also_track = [] if also_track is None else [also_track]
super(Layer, self).__init__(name=name)
self._trace = False
self._validate_args = validate_args
def __init__(self, name=None):
name = name_util.strip_invalid_chars(name or type(self).__name__)
super(Layer, self).__init__(name=name)
self._extra_loss = None
self._extra_result = None
def __init__(self,
pretransformed_input,
transform_fn,
dtype=None,
shape=NONE_SPECIFIED,
also_track=None,
name=None):
"""Creates the `DeferredTensor` object.
Args:
pretransformed_input: object with `shape`, `dtype` properties (typically a
`tf.Variable`) passed into `transform_fn` when this object is acted upon
in a `Tensor` context, eg, `tf.convert_to_tensor`, `+`, `tf.math.exp`,
etc.
transform_fn: Python `callable` or `tfp.bijectors.Bijector`-like instance.
When `callable`, should take `pretransformed_input` and
return a `Tensor` (representing by this object).
dtype: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).dtype`.
Default value: `None` (i.e.,
`getattr(transform_fn, 'dtype', None) or pretransformed_input.dtype`).
shape: Equivalent to what would otherwise be
`transform_fn(pretransformed_input).shape`.
Default value: `'None'` (i.e.,
`getattr(transform_fn, 'forward_event_shape', lambda x: x)(
pretransformed_input.shape)`).
also_track: Optional instance or structure of instances of `tf.Variable`
and/or `tf.Module`, containing any additional trainable variables that
the `transform_fn` may access beyond the given
`pretransformed_input`. This ensures that such variables
will be correctly tracked in `self.trainable_variables`.
Default value: `None`.
name: Python `str` representing this object's `name`; used only in graph
mode.
Default value: `None` (i.e.,
`(getattr(transform_fn, 'name', None) or
transform_fn.__name__ + '_' + pretransformed_input.name)`).
Raises:
TypeError: if `transform_fn` is not `callable`.
TypeError: if `pretransformed_input` lacks `dtype` and/or `shape`
properties (and `dtype` and/or `shape` arguments are unspecified).
"""
pretransformed_input = tensor_util.convert_nonref_to_tensor(
pretransformed_input, name='pretransformed_input')
if dtype is None:
dtype = (getattr(transform_fn, 'dtype', None)
or dtype_util.base_dtype(pretransformed_input.dtype))
try:
dtype = None if dtype is None else tf.as_dtype(dtype)
except TypeError:
raise TypeError('Argument `dtype` must be convertible to a '
'`tf.dtypes.DType`; saw "{}" of type "{}".'.format(
repr(dtype), type(dtype)))
if shape == NONE_SPECIFIED:
shape = getattr(transform_fn, 'forward_event_shape', _identity)
shape = shape(pretransformed_input.shape)
try:
shape = tf.TensorShape(shape)
except TypeError:
raise TypeError('Argument `shape` must be convertible to a '
'`tf.TensorShape`; saw "{}".'.format(shape))
name = name or getattr(transform_fn, 'name', None)
if not name:
name = '_'.join([
transform_fn.__name__,
getattr(pretransformed_input, 'name', '')
])
name = name_util.strip_invalid_chars(name)
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
if hasattr(transform_fn, 'forward'):
fwd_name = '"{}"'.format(transform_fn.name)
else:
fwd_name = transform_fn.__name__
if not callable(transform_fn):
raise TypeError('Argument `transform_fn` must be `callable`.')
super(DeferredTensor, self).__init__(name=name)
self._pretransformed_input = pretransformed_input
self._transform_fn = transform_fn
self._dtype = dtype
self._shape = shape
self._also_track = also_track
self._name = name
self._fwd_name = fwd_name
# Secret handshake with tf.is_tensor to return True for DT.
#
# Works around an exception in LinearOperator (which in 2.0.0 checks only
# `tf.is_tensor`, not also `linear_operator_util.is_ref`:
# ValueError: Graph parent item 0 is not a Tensor;
# <DeferredTensor: dtype=float32, shape=[2], fn=exp>.
# TODO(b/140157055): Remove this shim after LinOp is patched in 2.0.
self.is_tensor_like = True
def __init__(self,
dtype,
reparameterization_type,
validate_args,
allow_nan_stats,
parameters=None,
graph_parents=None,
name=None):
"""Constructs the `Distribution`.
**This is a private method for subclass use.**
Args:
dtype: The type of the event samples. `None` implies no type-enforcement.
reparameterization_type: Instance of `ReparameterizationType`.
If `tfd.FULLY_REPARAMETERIZED`, then samples from the distribution are
fully reparameterized, and straight-through gradients are supported.
If `tfd.NOT_REPARAMETERIZED`, then samples from the distribution are not
fully reparameterized, and straight-through gradients are either
partially unsupported or are not supported at all.
validate_args: Python `bool`, default `False`. When `True` distribution
parameters are checked for validity despite possibly degrading runtime
performance. When `False` invalid inputs may silently render incorrect
outputs.
allow_nan_stats: Python `bool`, default `True`. When `True`, statistics
(e.g., mean, mode, variance) use the value "`NaN`" to indicate the
result is undefined. When `False`, an exception is raised if one or
more of the statistic's batch members are undefined.
parameters: Python `dict` of parameters used to instantiate this
`Distribution`.
graph_parents: Python `list` of graph prerequisites of this
`Distribution`.
name: Python `str` name prefixed to Ops created by this class. Default:
subclass name.
Raises:
ValueError: if any member of graph_parents is `None` or not a `Tensor`.
"""
if not name:
name = type(self).__name__
name = name_util.camel_to_lower_snake(name)
name = name_util.get_name_scope_name(name)
name = name_util.strip_invalid_chars(name)
super(Distribution, self).__init__(name=name)
self._name = name
graph_parents = [] if graph_parents is None else graph_parents
for i, t in enumerate(graph_parents):
if t is None or not tf.is_tensor(t):
raise ValueError('Graph parent item %d is not a Tensor; %s.' % (i, t))
self._dtype = dtype
self._reparameterization_type = reparameterization_type
self._allow_nan_stats = allow_nan_stats
self._validate_args = validate_args
self._parameters = self._no_dependency(parameters)
self._parameters_sanitized = False
self._graph_parents = graph_parents
self._initial_parameter_control_dependencies = tuple(
d for d in self._parameter_control_dependencies(is_init=True)
if d is not None)
if self._initial_parameter_control_dependencies:
self._initial_parameter_control_dependencies = (
tf.group(*self._initial_parameter_control_dependencies),)
