experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update documentation for trainable SyncOnRead variables.

Browse Source 
PiperOrigin-RevId: 252010869
This commit is contained in:
Peter Buchlovsky 2019-06-07 01:48:27 -07:00 committed by TensorFlower Gardener
parent bfc8733ffb
commit c8bcbeff7c
5 changed files with 27 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/contrib/framework/python/ops/variables.py
View File

@ -249,8 +249,6 @@ def variable(name,
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses when to synchronize. `AUTO` and the current `DistributionStrategy` chooses when to synchronize.
If `synchronization` is set to `ON_READ`, `trainable` must not be set to
`True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -327,8 +325,6 @@ def model_variable(name,
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses when to synchronize. `AUTO` and the current `DistributionStrategy` chooses when to synchronize.
If `synchronization` is set to `ON_READ`, `trainable` must not be set to
`True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.

3
tensorflow/python/eager/def_function.py
View File

@ -97,8 +97,7 @@ class UnliftedInitializerVariable(resource_variable_ops.UninitializedVariable):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.

15
tensorflow/python/ops/resource_variable_ops.py
View File

@ -388,7 +388,8 @@ class ResourceVariable(variables.VariableV1):
trainable: If `True`, the default, also adds the variable to the graph trainable: If `True`, the default, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
Defaults to `True` unless `synchronization` is set to `ON_READ`. Defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
validate_shape: Ignored. Provided for compatibility with tf.Variable. validate_shape: Ignored. Provided for compatibility with tf.Variable.
@ -421,8 +422,7 @@ class ResourceVariable(variables.VariableV1):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -495,7 +495,8 @@ class ResourceVariable(variables.VariableV1):
trainable: If `True`, the default, also adds the variable to the graph trainable: If `True`, the default, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
Defaults to `True` unless `synchronization` is set to `ON_READ`. Defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
caching_device: Optional device string or function describing where the caching_device: Optional device string or function describing where the
@ -520,8 +521,7 @@ class ResourceVariable(variables.VariableV1):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -1793,8 +1793,7 @@ class UninitializedVariable(ResourceVariable):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.

18
tensorflow/python/ops/variable_scope.py
View File

@ -337,7 +337,8 @@ class _VariableStore(object):
forced to be False. forced to be False.
trainable: If `True` also add the variable to the graph collection trainable: If `True` also add the variable to the graph collection
`GraphKeys.TRAINABLE_VARIABLES` (see `tf.Variable`). `trainable` `GraphKeys.TRAINABLE_VARIABLES` (see `tf.Variable`). `trainable`
defaults to `True` unless `synchronization` is set to `ON_READ`. defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys to add the `Variable` to. collections: List of graph collections keys to add the `Variable` to.
Defaults to `[GraphKeys.GLOBAL_VARIABLES]` (see `tf.Variable`). Defaults to `[GraphKeys.GLOBAL_VARIABLES]` (see `tf.Variable`).
caching_device: Optional device string or function describing where the caching_device: Optional device string or function describing where the
@ -1588,8 +1589,7 @@ Args:
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -1733,8 +1733,6 @@ def _get_partitioned_variable(name,
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses when to synchronize. `AUTO` and the current `DistributionStrategy` chooses when to synchronize.
If `synchronization` is set to `ON_READ`, `trainable` must not be set to
`True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -2596,8 +2594,8 @@ def variable_creator_scope_v1(variable_creator):
trainable: If `True`, the default, also adds the variable to the graph trainable: If `True`, the default, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
`trainable` defaults to `True` unless `synchronization` is `trainable` defaults to `True`, unless `synchronization` is
set to `ON_READ`. set to `ON_READ`, in which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
validate_shape: If `False`, allows the variable to be initialized with a validate_shape: If `False`, allows the variable to be initialized with a
@ -2620,8 +2618,7 @@ def variable_creator_scope_v1(variable_creator):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -2690,8 +2687,7 @@ def variable_creator_scope(variable_creator):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.

25
tensorflow/python/ops/variables.py
View File

@ -404,8 +404,8 @@ class Variable(six.with_metaclass(VariableMetaclass,
that case, `dtype` must be specified. (Note that initializer functions that case, `dtype` must be specified. (Note that initializer functions
from init_ops.py must first be bound to a shape before being used here.) from init_ops.py must first be bound to a shape before being used here.)
trainable: If `True`, GradientTapes automatically watch uses trainable: If `True`, GradientTapes automatically watch uses
of this variable. Defaults to `True` unless `synchronization` is of this variable. Defaults to `True`, unless `synchronization` is
set to `ON_READ`. set to `ON_READ`, in which case it defaults to `False`.
validate_shape: If `False`, allows the variable to be initialized with a validate_shape: If `False`, allows the variable to be initialized with a
value of unknown shape. If `True`, the default, the shape of value of unknown shape. If `True`, the default, the shape of
`initial_value` must be known. `initial_value` must be known.
@ -436,8 +436,7 @@ class Variable(six.with_metaclass(VariableMetaclass,
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -1449,7 +1448,8 @@ class VariableV1(Variable):
trainable: If `True`, also adds the variable to the graph trainable: If `True`, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
Defaults to `True` unless `synchronization` is set to `ON_READ`. Defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
validate_shape: If `False`, allows the variable to be initialized with a validate_shape: If `False`, allows the variable to be initialized with a
@ -1485,8 +1485,7 @@ class VariableV1(Variable):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -1545,7 +1544,8 @@ class RefVariable(VariableV1):
trainable: If `True`, also adds the variable to the graph trainable: If `True`, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
Defaults to `True` unless `synchronization` is set to `ON_READ`. Defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
validate_shape: If `False`, allows the variable to be initialized with a validate_shape: If `False`, allows the variable to be initialized with a
@ -1580,8 +1580,7 @@ class RefVariable(VariableV1):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.
@ -1653,7 +1652,8 @@ class RefVariable(VariableV1):
trainable: If `True`, also adds the variable to the graph trainable: If `True`, also adds the variable to the graph
collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as collection `GraphKeys.TRAINABLE_VARIABLES`. This collection is used as
the default list of variables to use by the `Optimizer` classes. the default list of variables to use by the `Optimizer` classes.
Defaults to `True` unless `synchronization` is set to `ON_READ`. Defaults to `True`, unless `synchronization` is set to `ON_READ`, in
which case it defaults to `False`.
collections: List of graph collections keys. The new variable is added to collections: List of graph collections keys. The new variable is added to
these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`. these collections. Defaults to `[GraphKeys.GLOBAL_VARIABLES]`.
validate_shape: If `False`, allows the variable to be initialized with a validate_shape: If `False`, allows the variable to be initialized with a
@ -1682,8 +1682,7 @@ class RefVariable(VariableV1):
aggregated. Accepted values are constants defined in the class aggregated. Accepted values are constants defined in the class
`tf.VariableSynchronization`. By default the synchronization is set to `tf.VariableSynchronization`. By default the synchronization is set to
`AUTO` and the current `DistributionStrategy` chooses `AUTO` and the current `DistributionStrategy` chooses
when to synchronize. If `synchronization` is set to `ON_READ`, when to synchronize.
`trainable` must not be set to `True`.
aggregation: Indicates how a distributed variable will be aggregated. aggregation: Indicates how a distributed variable will be aggregated.
Accepted values are constants defined in the class Accepted values are constants defined in the class
`tf.VariableAggregation`. `tf.VariableAggregation`.