Python variable_scope.AUTO_REUSE属性代码示例

# 需要导入模块: from tensorflow.python.ops import variable_scope [as 别名]
# 或者: from tensorflow.python.ops.variable_scope import AUTO_REUSE [as 别名]
def register_block(self, layer_key, fisher_block, reuse=VARIABLE_SCOPE):
        if reuse is VARIABLE_SCOPE:
            reuse = variable_scope.get_variable_scope().reuse

        if reuse is True or (reuse is variable_scope.AUTO_REUSE and
                                     layer_key in self.fisher_blocks):
            result = self.fisher_blocks[layer_key]
            if type(result) != type(fisher_block):  # pylint: disable=unidiomatic-typecheck
                raise ValueError(
                    "Attempted to register FisherBlock of type %s when existing "
                    "FisherBlock has type %s." % (type(fisher_block), type(result)))
            return result
        if reuse is False and layer_key in self.fisher_blocks:
            raise ValueError("FisherBlock for %s is already in LayerCollection." %
                             (layer_key,))

        # Insert fisher_block into self.fisher_blocks.
        if layer_key in self.fisher_blocks:
            raise ValueError("Duplicate registration: {}".format(layer_key))
        # Raise an error if any variable in layer_key has been registered in any
        # other blocks.
        variable_to_block = {
            var: (params, block)
            for (params, block) in self.fisher_blocks.items()
            for var in ensure_sequence(params)
        }
        for variable in ensure_sequence(layer_key):
            if variable in variable_to_block:
                prev_key, prev_block = variable_to_block[variable]
                raise ValueError(
                    "Attempted to register layer_key {} with block {}, but variable {}"
                    " was already registered in key {} with block {}.".format(
                        layer_key, fisher_block, variable, prev_key, prev_block))
        self.fisher_blocks[layer_key] = fisher_block
        return fisher_block 

# 需要导入模块: from tensorflow.python.ops import variable_scope [as 别名]
# 或者: from tensorflow.python.ops.variable_scope import AUTO_REUSE [as 别名]
def _get_or_create_stop_var():
  with tf.compat.v1.variable_scope(
      name_or_scope='signal_early_stopping',
      values=[],
      reuse=tf.compat.v1.AUTO_REUSE):
    return tf.compat.v1.get_variable(
        name='STOP',
        shape=[],
        dtype=tf.dtypes.bool,
        initializer=tf.compat.v1.initializers.constant(False),
        collections=[tf.compat.v1.GraphKeys.GLOBAL_VARIABLES],
        trainable=False) 

# 需要导入模块: from tensorflow.python.ops import variable_scope [as 别名]
# 或者: from tensorflow.python.ops.variable_scope import AUTO_REUSE [as 别名]
def _get_or_create_stop_var_with_aggregation(self):
    with variable_scope.variable_scope(
        name_or_scope='signal_early_stopping',
        values=[],
        reuse=variable_scope.AUTO_REUSE):
      return variable_scope.get_variable(
          name='STOP',
          shape=[],
          dtype=tf.dtypes.int32,
          initializer=init_ops.constant_initializer(0),
          collections=[ops.GraphKeys.GLOBAL_VARIABLES],
          synchronization=variable_scope.VariableSynchronization.ON_WRITE,
          aggregation=variable_scope.VariableAggregation.SUM,
          trainable=False) 

# 需要导入模块: from tensorflow.python.ops import variable_scope [as 别名]
# 或者: from tensorflow.python.ops.variable_scope import AUTO_REUSE [as 别名]
def _create_or_get_iterations_per_loop():
  """Creates or gets the iterations_per_loop variable.

  In TPUEstimator, the user provided computation, the model_fn, is wrapped
  inside a tf.while_loop for peak performance. The iterations of the loop are
  specified by this variable, which adjusts its value on the CPU after each TPU
  program execution and before the next TPU execution.

  The purpose of using a variable, rather then a constant, is to allow
  TPUEstimator adapt the TPU training iterations according to the final steps
  specified by users. For example, if the user sets the iterations_per_loop as 4
  in TPUConfig and steps as 10 in TPUEstimator.train(), the iterations_per_loop
  variable will have the following value before each TPU training.

      - 1-th TPU execution: iterations_per_loop = 4
      - 2-th TPU execution: iterations_per_loop = 4
      - 3-th TPU execution: iterations_per_loop = 2

  As model_fn increases the global step once per train_op invocation, the global
  step is 10 after all TPU executions, matching the steps=10 inputs passed in by
  users.

  Returns:
    A TF non-trainable resource variable.

  Raises:
    RuntimeError: If multi iterations_per_loop variables were found.
  """
  graph = ops.get_default_graph()
  collection_name = '{}_{}'.format(_TPU_ESTIMATOR, _ITERATIONS_PER_LOOP_VAR)
  iter_vars = graph.get_collection(collection_name)
  if len(iter_vars) == 1:
    return iter_vars[0]
  elif len(iter_vars) > 1:
    raise RuntimeError('Multiple iterations_per_loop_var in collection.')

  with ops.colocate_with(training_util.get_global_step()):
    with variable_scope.variable_scope(
        _TPU_ESTIMATOR, reuse=variable_scope.AUTO_REUSE):
      return variable_scope.get_variable(
          _ITERATIONS_PER_LOOP_VAR,
          initializer=init_ops.zeros_initializer(),
          shape=[],
          dtype=dtypes.int32,
          trainable=False,
          collections=[collection_name, ops.GraphKeys.LOCAL_VARIABLES],
          use_resource=True)