本文整理汇总了Python中tensorflow.python.ops.state_ops.variable_op_v2方法的典型用法代码示例。如果您正苦于以下问题：Python state_ops.variable_op_v2方法的具体用法？Python state_ops.variable_op_v2怎么用？Python state_ops.variable_op_v2使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类tensorflow.python.ops.state_ops的用法示例。
在下文中一共展示了state_ops.variable_op_v2方法的1个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: define_local
# 需要导入模块: from tensorflow.python.ops import state_ops [as 别名]
# 或者: from tensorflow.python.ops.state_ops import variable_op_v2 [as 别名]
def define_local(self, name, value):
if name in self._locals:
raise Exception("Local already defined: %s" % name)
should_wrap_in_var = False
if self._wrap_locals_in_vars:
if isinstance(value, tf.Tensor):
should_wrap_in_var = True
# HACK(adamb) Unwrapping in here really isn't great, since auto-unwrapping can create unexpected behavior.
if isinstance(value, RetvalBag) and value.len() == 1:
if isinstance(value.get(None), tf.Tensor):
should_wrap_in_var = True
value = value.get(None)
if should_wrap_in_var:
variable = state_ops.variable_op_v2(
value.get_shape(),
value.dtype.base_dtype)
with tf.control_dependencies(None):
value = tf.identity(
tf.cond(
tf.is_variable_initialized(variable),
lambda: variable,
lambda: tf.assign(variable, value)
)
)
print("value", value)
self._locals[name] = value
return value