构建计算图 API

╰+攻爆jí腚メ 2022-06-11 08:38 251阅读 0赞

> 官方文档：  
> [https://www.tensorflow.org/versions/r0.12/api\_docs/python/framework.html\#building-graphs][https_www.tensorflow.org_versions_r0.12_api_docs_python_framework.html_building-graphs]  
> 极客学院：  
> [http://wiki.jikexueyuan.com/project/tensorflow-zh/api\_docs/python/framework.html\#Graph][http_wiki.jikexueyuan.com_project_tensorflow-zh_api_docs_python_framework.html_Graph]

--------------------

> 用于构造计算图的类和函数

计算图的核心数据结构

## 类：tf.Graph ##

一个TensorFlow 的计算，被表示为一个数据流图

一个计算图包含了：

1. 一系列的操作（Operation）对象，代表一个计算单元；
    2. 还有张量（Tensor）对象，代表了在操作对象之间流动的数据单元

系统中总是有一个默认的计算图会被注册，可以通过调用[`tf.get_default_graph()`][tf.get_default_graph]来访问它。要添加一个操作结点到默认的计算图中，只需要调用定义一个新操作的函数之一即可：

c = tf.constant(4.0)
    assert c.graph is tf.get_default_graph()

另一种典型的用法涉及到 [`Graph.as_default()`][Graph.as_default] 上下文管理器，它在管理器的生命周期内重载（overrides）了当前默认的计算图：

g = tf.Graph()
    with g.as_default():
      # Define operations and tensors in `g`.
      c = tf.constant(30.0)
      assert c.graph is g

重要提示：这个类对计算图的构造是非线程安全的。所有的操作对象应该从一个单独线程中创建，或者必须提供外部的同步协议。除非另有规定，否则所有的函数是非线程安全的。

#### `tf.Graph.__init__()` ####

创建一个新的、空的计算图

#### `tf.Graph.as_default()` ####

返回一个将该计算图作为默认图的上下文管理器。

如果你想在同一个进程内创建多个计算图，那么你应该使用这个函数。为了方便，如果你没有明确地创建一个图，所有的操作都会被添加到一个全局默认的图中。使用这个函数时带着with关键字，来指定所有在该作用域内创建的操作对象添加到这个图中。

默认的图是当前线程的属性。如果你创建了一个新的线程，并且希望在该新线程内使用默认的图，你必须明确地添加 with g.as\_default() ：在那个线程的方法内。

下面的代码示例是等价的：

# 1. Using Graph.as_default():
    g = tf.Graph()
    with g.as_default():
      c = tf.constant(5.0)
      assert c.graph is g
    
    # 2. Constructing and making default:
    with tf.Graph().as_default() as g:
      c = tf.constant(5.0)
      assert c.graph is g

##### 返回一个使用这个计算图作为默认图的上下文管理器。 #####

--------------------

#### `tf.Graph.as_graph_def(from_version=None, add_shapes=False)` ####

返回一个序列化的`GraphDef`表示这个图。

这个序列化的`GraphDef` 可以被导入到其他的图中（使用 [`import_graph_def()`][import_graph_def] ）或者和 [C++ Session API][C_ Session API] 一起使用。

这个函数是线程安全的。

参数：

*  **from\_version** ：可选。如果这个被设置了，将返回一个只包含被添加到这个图且version属性含有这个给定值的结点的**GraphDef**。
 *  **from\_version**: Optional. If this is set, returns a `GraphDef` containing only the nodes that were added to this graph since its `version` property had the given value.
 *  **add\_shapes**: If true, adds an “\_output\_shapes” list attr to each node with the inferred shapes of each of its outputs.

A [`GraphDef`][GraphDef] protocol buffer.

（注：**protocol** **buffer**(以下简称PB)是google 的一种数据交换的格式，它独立于语言，[独立][Link 1]于平台；作为一种效率和兼容性都很优秀的二进制数据传输格式，可以用于诸如网络传输、配置文件、数据存储等诸多领域。）

抛出的异常：

*  **ValueError**: 如果这个 `graph_def` 太大了。

--------------------

#### `tf.Graph.finalize()` ####

完成此图，将它设为只读。

在调用 `g.finalize()` 之后，操作对象就不能再被添加到 `g`中了。这个函数被用于确保在多个线程之间共享计算图时没有操作被添加到图中，例如当使用一个 [`QueueRunner`][QueueRunner] 时。

--------------------

#### `tf.Graph.finalized` ####

如果这个图已经 finalized 了，则该值为True

--------------------

#### `tf.Graph.control_dependencies(control_inputs)` ####

返回一个指定了控制依赖的上下文管理器。

和 `with` 关键字一起用于指定上下文中所有的操作结构的 `control_inputs` 属性的控制依赖。例如：

with g.control_dependencies([a, b, c]):
      # `d` 和 `e` 将只会在 `a`, `b` 和 `c` 运行完毕后才运行。
      d = ...
      e = ...

多次调用的 `control_dependencies()` 可以被嵌套， 在这种情况下，一个新的操作的控制依赖将是所有激活的上下文中的 `control_inputs` 的结合体。

with g.control_dependencies([a, b]):
      # Ops constructed here run after `a` and `b`.
      with g.control_dependencies([c, d]):
        # Ops constructed here run after `a`, `b`, `c`, and `d`.

你可以传递一个 None 对象清空控制依赖：

with g.control_dependencies([a, b]):
      # 在这里创建的操作将在 `a` 和 `b`运行之后才运行
      with g.control_dependencies(None):
        # 在这里创建的操作将正常运行，不会等待 `a` 或 `b`
        with g.control_dependencies([c, d]):
          # 在这里创建的操作将在 `c` 和 `d`运行之后才运行，不会等待 `a` 或 `b`

注：控制依赖上下文只应用于在上下文之中创建的操作。 仅仅只是在上下文中使用张量或操作，则不会添加控制依赖。下面的示例说明了这一点：

# 错误
    def my_func(pred, tensor):
      t = tf.matmul(tensor, tensor)
      with tf.control_dependencies([pred]):
        # 这个 matmul 操作是在上下文之外创建的，所以不会添加任何控制依赖
        return t
    
    # 正确
    def my_func(pred, tensor):
      with tf.control_dependencies([pred]):
        # 这个 matmul 操作是在上下文中创建的，所以控制依赖会被添加
        return tf.matmul(tensor, tensor)

参数：

*  **control\_inputs**：一个操作和张量对象的列表，它们将在运行在上下文中定义的操作之前运行。当然也可以设置为 None 来清空控制依赖。

一个为所有在上下文中创建的操作指定了控制依赖的上下文管理器。

抛出的异常：

*  **TypeError** ：如果 control\_inputs 不是一个 Operation 或 Tensor 对象的列表。

--------------------

#### `tf.Graph.device(device_name_or_function)` ####

返回一个指定了要使用的默认设备的上下文管理器。

`device_name_or_function` 这个参数可以是一个设备的名称，一个 device function ，或者是 None：

*  如果它是一个设备的名称，则素有在这个上下文中创建的操作将会被分配给那个名称的设备，除非在上下文中被一个嵌套的 `device()` 重写。
 *  If it is a function, it will be treated as a function from Operation objects to device name strings, and invoked each time a new Operation is created. The Operation will be assigned to the device with the returned name.
 *  If it is None, all `device()` invocations from the enclosing context will be ignored.

For information about the valid syntax of device name strings, see the documentation in[`DeviceNameUtils`][DeviceNameUtils].

示例：

with g.device('/gpu:0'):
      # 所有在此上下文中创建的操作将被放置在 GPU 0 中。
      with g.device(None):
        # 所有在此上下文中创建的操作将不会被分配给任何设备。
    
    # Defines a function from `Operation` to device string.
    def matmul_on_gpu(n):
      if n.type == "MatMul":
        return "/gpu:0"
      else:
        return "/cpu:0"
    
    with g.device(matmul_on_gpu):
      # All operations of type "MatMul" constructed in this context
      # will be placed on GPU 0; all other operations will be placed
      # on CPU 0.

注：The device scope may be overridden by op wrappers or other library code. For example, a variable assignment op `v.assign()` must be colocated with the `tf.Variable` `v`, and incompatible device scopes will be ignored.

参数：

*  **device\_name\_or\_function** ：在上下文中使用的设备的名称或方法。

A context manager that specifies the default device to use for newly created ops.

--------------------

#### `tf.Graph.name_scope(name)` ####

返回为操作创建分级名称的上下文管理器。

A graph maintains a stack of name scopes. A `with name_scope(...):` statement pushes a new name onto the stack for the lifetime of the context.

The `name` argument will be interpreted as follows:

*  A string (not ending with ‘/’) will create a new name scope, in which `name` is appended to the prefix of all operations created in the context. If `name` has been used before, it will be made unique by calling `self.unique_name(name)`.
 *  A scope previously captured from a `with g.name_scope(...) as scope:` statement will be treated as an “absolute” name scope, which makes it possible to re-enter existing scopes.
 *  A value of `None` or the empty string will reset the current name scope to the top-level (empty) name scope.

示例：

with tf.Graph().as_default() as g:
      c = tf.constant(5.0, name="c")
      assert c.op.name == "c"
      c_1 = tf.constant(6.0, name="c")
      assert c_1.op.name == "c_1"
    
      # Creates a scope called "nested"
      with g.name_scope("nested") as scope:
        nested_c = tf.constant(10.0, name="c")
        assert nested_c.op.name == "nested/c"
    
        # Creates a nested scope called "inner".
        with g.name_scope("inner"):
          nested_inner_c = tf.constant(20.0, name="c")
          assert nested_inner_c.op.name == "nested/inner/c"
    
        # Create a nested scope called "inner_1".
        with g.name_scope("inner"):
          nested_inner_1_c = tf.constant(30.0, name="c")
          assert nested_inner_1_c.op.name == "nested/inner_1/c"
    
          # Treats `scope` as an absolute name scope, and
          # switches to the "nested/" scope.
          with g.name_scope(scope):
            nested_d = tf.constant(40.0, name="d")
            assert nested_d.op.name == "nested/d"
    
            with g.name_scope(""):
              e = tf.constant(50.0, name="e")
              assert e.op.name == "e"

The name of the scope itself can be captured by `with g.name_scope(...) as scope:`, which stores the name of the scope in the variable `scope`. This value can be used to name an operation that represents the overall result of executing the ops in a scope. For example:

inputs = tf.constant(...)
    with g.name_scope('my_layer') as scope:
      weights = tf.Variable(..., name="weights")
      biases = tf.Variable(..., name="biases")
      affine = tf.matmul(inputs, weights) + biases
      output = tf.nn.relu(affine, name=scope)

注：This constructor validates the given `name`. Valid scope names match one of the following regular expressions:

[A-Za-z0-9.][A-Za-z0-9_.\\-/]* (for scopes at the root)
    [A-Za-z0-9_.\\-/]* (for other scopes)

参数：

*  **name**: A name for the scope.

返回值：

*  A context manager that installs `name` as a new name scope.

抛出的异常：

*  **ValueError**: If `name` is not a valid scope name.

A `Graph` instance supports an arbitrary number of “collections” that are identified by name. For convenience when building a large graph, collections can store groups of related objects: for example, the `tf.Variable` uses a collection (named [`tf.GraphKeys.GLOBAL_VARIABLES`][tf.GraphKeys.GLOBAL_VARIABLES]) for all variables that are created during the construction of a graph. The caller may define additional collections by specifying a new name.

--------------------

#### `tf.Graph.add_to_collection(name, value)` ####

使用给定的名称将值存储到 `collection` 集合中。

注意： collections 不是 sets，所以它可以多次添加一个值到 collections 中。This function makes sure that duplicates in `names` are ignored, but it will not check for pre-existing membership of `value` in any of the collections in `names`.

`names` can be any iterable, but if `names` is a string, it is treated as a single collection name.

##### Args: #####

*  **names**: The keys for the collections to add to. The `GraphKeys` class contains many standard names for collections.
 *  **value**: The value to add to the collections.

--------------------

#### `tf.Graph.get_collection(name, scope=None)` ####

Returns a list of values in the collection with the given `name`.

This is different from `get_collection_ref()` which always returns the actual collection list if it exists in that it returns a new list each time it is called.

##### Args: #####

*  **name**: The key for the collection. For example, the `GraphKeys` class contains many standard names for collections.
 *  **scope**: (Optional.) If supplied, the resulting list is filtered to include only items whose `name`attribute matches using `re.match`. Items without a `name` attribute are never returned if a scope is supplied and the choice or `re.match` means that a `scope` without special tokens filters by prefix.

##### Returns: #####

The list of values in the collection with the given `name`, or an empty list if no value has been added to that collection. The list contains the values in the order under which they were collected.

--------------------

#### `tf.Graph.get_collection_ref(name)` ####

Returns a list of values in the collection with the given `name`.

If the collection exists, this returns the list itself, which can be modified in place to change the collection. If the collection does not exist, it is created as an empty list and the list is returned.

This is different from `get_collection()` which always returns a copy of the collection list if it exists and never creates an empty collection.

##### Args: #####

*  **name**: The key for the collection. For example, the `GraphKeys` class contains many standard names for collections.

##### Returns: #####

The list of values in the collection with the given `name`, or an empty list if no value has been added to that collection.

--------------------

#### `tf.Graph.as_graph_element(obj, allow_tensor=True, allow_operation=True)` ####

Returns the object referred to by `obj`, as an `Operation` or `Tensor`.

This function validates that `obj` represents an element of this graph, and gives an informative error message if it is not.

This function is the canonical way to get/validate an object of one of the allowed types from an external argument reference in the Session API.

This method may be called concurrently from multiple threads.

##### Args: #####

*  **obj**: A `Tensor`, an `Operation`, or the name of a tensor or operation. Can also be any object with an `_as_graph_element()` method that returns a value of one of these types.
 *  **allow\_tensor**: If true, `obj` may refer to a `Tensor`.
 *  **allow\_operation**: If true, `obj` may refer to an `Operation`.

##### Returns: #####

The `Tensor` or `Operation` in the Graph corresponding to `obj`.

##### Raises: #####

*  **TypeError**: If `obj` is not a type we support attempting to convert to types.
 *  **ValueError**: If `obj` is of an appropriate type but invalid. For example, an invalid string.
 *  **KeyError**: If `obj` is not an object in the graph.

--------------------

#### `tf.Graph.get_operation_by_name(name)` ####

Returns the `Operation` with the given `name`.

This method may be called concurrently from multiple threads.

##### Args: #####

*  **name**: The name of the `Operation` to return.

##### Returns: #####

The `Operation` with the given `name`.

##### Raises: #####

*  **TypeError**: If `name` is not a string.
 *  **KeyError**: If `name` does not correspond to an operation in this graph.

--------------------

#### `tf.Graph.get_tensor_by_name(name)` ####

Returns the `Tensor` with the given `name`.

This method may be called concurrently from multiple threads.

##### Args: #####

*  **name**: The name of the `Tensor` to return.

##### Returns: #####

The `Tensor` with the given `name`.

##### Raises: #####

*  **TypeError**: If `name` is not a string.
 *  **KeyError**: If `name` does not correspond to a tensor in this graph.

--------------------

#### `tf.Graph.get_operations()` ####

Return the list of operations in the graph.

You can modify the operations in place, but modifications to the list such as inserts/delete have no effect on the list of operations known to the graph.

This method may be called concurrently from multiple threads.

##### Returns: #####

A list of Operations.

--------------------

#### `tf.Graph.seed` ####

The graph-level random seed of this graph.

--------------------

#### `tf.Graph.unique_name(name, mark_as_used=True)` ####

Return a unique operation name for `name`.

Note: You rarely need to call `unique_name()` directly. Most of the time you just need to create `with g.name_scope()` blocks to generate structured names.

`unique_name` is used to generate structured names, separated by `"/"`, to help identify operations when debugging a graph. Operation names are displayed in error messages reported by the TensorFlow runtime, and in various visualization tools such as TensorBoard.

If `mark_as_used` is set to `True`, which is the default, a new unique name is created and marked as in use. If it’s set to `False`, the unique name is returned without actually being marked as used. This is useful when the caller simply wants to know what the name to be created will be.

##### Args: #####

*  **name**: The name for an operation.
 *  **mark\_as\_used**: Whether to mark this name as being used.

##### Returns: #####

A string to be passed to `create_op()` that will be used to name the operation being created.

--------------------

#### `tf.Graph.version` ####

Returns a version number that increases as ops are added to the graph.

Note that this is unrelated to the [GraphDef version][].

--------------------

#### `tf.Graph.graph_def_versions` ####

The GraphDef version information of this graph.

For details on the meaning of each version, see [`GraphDef`][GraphDef].

##### Returns: #####

A `VersionDef`.

--------------------

#### `tf.Graph.create_op(op_type, inputs, dtypes, input_types=None, name=None, attrs=None, op_def=None, compute_shapes=True, compute_device=True)` ####

Creates an `Operation` in this graph.

This is a low-level interface for creating an `Operation`. Most programs will not call this method directly, and instead use the Python op constructors, such as `tf.constant()`, which add ops to the default graph.

##### Args: #####

*  **op\_type**: The `Operation` type to create. This corresponds to the `OpDef.name` field for the proto that defines the operation.
 *  **inputs**: A list of `Tensor` objects that will be inputs to the `Operation`.
 *  **dtypes**: A list of `DType` objects that will be the types of the tensors that the operation produces.
 *  **input\_types**: (Optional.) A list of `DType`s that will be the types of the tensors that the operation consumes. By default, uses the base `DType` of each input in `inputs`. Operations that expect reference-typed inputs must specify `input_types` explicitly.
 *  **name**: (Optional.) A string name for the operation. If not specified, a name is generated based on `op_type`.
 *  **attrs**: (Optional.) A dictionary where the key is the attribute name (a string) and the value is the respective `attr` attribute of the `NodeDef` proto that will represent the operation (an `AttrValue`proto).
 *  **op\_def**: (Optional.) The `OpDef` proto that describes the `op_type` that the operation will have.
 *  **compute\_shapes**: (Optional.) If True, shape inference will be performed to compute the shapes of the outputs.
 *  **compute\_device**: (Optional.) If True, device functions will be executed to compute the device property of the Operation.

##### Raises: #####

*  **TypeError**: if any of the inputs is not a `Tensor`.
 *  **ValueError**: if colocation conflicts with existing device assignment.

##### Returns: #####

An `Operation` object.

--------------------

#### `tf.Graph.gradient_override_map(op_type_map)` ####

EXPERIMENTAL: A context manager for overriding gradient functions.

This context manager can be used to override the gradient function that will be used for ops within the scope of the context.

For example:

@tf.RegisterGradient("CustomSquare")
    def _custom_square_grad(op, grad):
      # ...
    
    with tf.Graph().as_default() as g:
      c = tf.constant(5.0)
      s_1 = tf.square(c)  # Uses the default gradient for tf.square.
      with g.gradient_override_map({
       "Square": "CustomSquare"}):
        s_2 = tf.square(s_2)  # Uses _custom_square_grad to compute the
                              # gradient of s_2.

##### Args: #####

*  **op\_type\_map**: A dictionary mapping op type strings to alternative op type strings.

##### Returns: #####

A context manager that sets the alternative op type to be used for one or more ops created in that context.

##### Raises: #####

*  **TypeError**: If `op_type_map` is not a dictionary mapping strings to strings.

#### Other Methods ####

--------------------

#### `tf.Graph.building_function` ####

Returns True iff this graph represents a function.

--------------------

#### `tf.Graph.clear_collection(name)` ####

Clears all values in a collection.

##### Args: #####

*  **name**: The key for the collection. The `GraphKeys` class contains many standard names for collections.

--------------------

#### `tf.Graph.colocate_with(op, ignore_existing=False)` ####

Returns a context manager that specifies an op to colocate with.

Note: this function is not for public use, only for internal libraries.

For example:

a = tf.Variable([1.0])
    with g.colocate_with(a):
      b = tf.constant(1.0)
      c = tf.add(a, b)

`b` and `c` will always be colocated with `a`, no matter where `a` is eventually placed.

##### Args: #####

*  **op**: The op to colocate all created ops with.
 *  **ignore\_existing**: If true, only applies colocation of this op within the context, rather than applying all colocation properties on the stack.

##### Raises: #####

*  **ValueError**: if op is None.

##### Yields: #####

A context manager that specifies the op with which to colocate newly created ops.

--------------------

#### `tf.Graph.container(container_name)` ####

Returns a context manager that specifies the resource container to use.

Stateful operations, such as variables and queues, can maintain their states on devices so that they can be shared by multiple processes. A resource container is a string name under which these stateful operations are tracked. These resources can be released or cleared with `tf.Session.reset()`.

For example:

with g.container('experiment0'):
      # All stateful Operations constructed in this context will be placed
      # in resource container "experiment0".
      v1 = tf.Variable([1.0])
      v2 = tf.Variable([2.0])
      with g.container("experiment1"):
        # All stateful Operations constructed in this context will be
        # placed in resource container "experiment1".
        v3 = tf.Variable([3.0])
        q1 = tf.FIFOQueue(10, tf.float32)
      # All stateful Operations constructed in this context will be
      # be created in the "experiment0".
      v4 = tf.Variable([4.0])
      q1 = tf.FIFOQueue(20, tf.float32)
      with g.container(""):
        # All stateful Operations constructed in this context will be
        # be placed in the default resource container.
        v5 = tf.Variable([5.0])
        q3 = tf.FIFOQueue(30, tf.float32)
    
    # Resets container "experiment0", after which the state of v1, v2, v4, q1
    # will become undefined (such as uninitialized).
    tf.Session.reset(target, ["experiment0"])

##### Args: #####

*  **container\_name**: container name string.

##### Returns: #####

A context manager for defining resource containers for stateful ops, yields the container name.

--------------------

#### `tf.Graph.get_all_collection_keys()` ####

Returns a list of collections used in this graph.

--------------------

#### `tf.Graph.is_feedable(tensor)` ####

Returns `True` if and only if `tensor` is feedable.

--------------------

#### `tf.Graph.is_fetchable(tensor_or_op)` ####

Returns `True` if and only if `tensor_or_op` is fetchable.

--------------------

#### `tf.Graph.prevent_feeding(tensor)` ####

Marks the given `tensor` as unfeedable in this graph.

--------------------

#### `tf.Graph.prevent_fetching(op)` ####

Marks the given `op` as unfetchable in this graph.

--------------------

## class tf.Operation ##

Represents a graph node that performs computation on tensors.

An `Operation` is a node in a TensorFlow `Graph` that takes zero or more `Tensor` objects as input, and produces zero or more `Tensor` objects as output. Objects of type `Operation` are created by calling a Python op constructor (such as [`tf.matmul()`][tf.matmul]) or [`Graph.create_op()`][Graph.create_op].

For example `c = tf.matmul(a, b)` creates an `Operation` of type “MatMul” that takes tensors `a`and `b` as input, and produces `c` as output.

After the graph has been launched in a session, an `Operation` can be executed by passing it to[`Session.run()`][Session.run]. `op.run()` is a shortcut for calling `tf.get_default_session().run(op)`.

--------------------

#### `tf.Operation.name` ####

The full name of this operation.

--------------------

#### `tf.Operation.type` ####

The type of the op (e.g. `"MatMul"`).

--------------------

#### `tf.Operation.inputs` ####

The list of `Tensor` objects representing the data inputs of this op.

--------------------

#### `tf.Operation.control_inputs` ####

The `Operation` objects on which this op has a control dependency.

Before this op is executed, TensorFlow will ensure that the operations in `self.control_inputs`have finished executing. This mechanism can be used to run ops sequentially for performance reasons, or to ensure that the side effects of an op are observed in the correct order.

##### Returns: #####

A list of `Operation` objects.

--------------------

#### `tf.Operation.outputs` ####

The list of `Tensor` objects representing the outputs of this op.

--------------------

#### `tf.Operation.device` ####

The name of the device to which this op has been assigned, if any.

##### Returns: #####

The string name of the device to which this op has been assigned, or an empty string if it has not been assigned to a device.

--------------------

#### `tf.Operation.graph` ####

The `Graph` that contains this operation.

--------------------

#### `tf.Operation.run(feed_dict=None, session=None)` ####

Runs this operation in a `Session`.

Calling this method will execute all preceding operations that produce the inputs needed for this operation.

*N.B.* Before invoking `Operation.run()`, its graph must have been launched in a session, and either a default session must be available, or `session` must be specified explicitly.

##### Args: #####

*  **feed\_dict**: A dictionary that maps `Tensor` objects to feed values. See [`Session.run()`][Session.run] for a description of the valid feed values.
 *  **session**: (Optional.) The `Session` to be used to run to this operation. If none, the default session will be used.

--------------------

#### `tf.Operation.get_attr(name)` ####

Returns the value of the attr of this op with the given `name`.

##### Args: #####

*  **name**: The name of the attr to fetch.

##### Returns: #####

The value of the attr, as a Python object.

##### Raises: #####

*  **ValueError**: If this op does not have an attr with the given `name`.

--------------------

#### `tf.Operation.traceback` ####

Returns the call stack from when this operation was constructed.

#### Other Methods ####

--------------------

#### `tf.Operation.__init__(node_def, g, inputs=None, output_types=None, control_inputs=None, input_types=None, original_op=None, op_def=None)` ####

Creates an `Operation`.

NOTE: This constructor validates the name of the `Operation` (passed as `node_def.name`). Valid `Operation` names match the following regular expression:

[A-Za-z0-9.][A-Za-z0-9_.\\-/]*

##### Args: #####

*  **node\_def**: `node_def_pb2.NodeDef`. `NodeDef` for the `Operation`. Used for attributes of `node_def_pb2.NodeDef`, typically `name`, `op`, and `device`. The `input` attribute is irrelevant here as it will be computed when generating the model.
 *  **g**: `Graph`. The parent graph.
 *  **inputs**: list of `Tensor` objects. The inputs to this `Operation`.
 *  **output\_types**: list of `DType` objects. List of the types of the `Tensors` computed by this operation. The length of this list indicates the number of output endpoints of the `Operation`.
 *  **control\_inputs**: list of operations or tensors from which to have a control dependency.
 *  **input\_types**: List of `DType` objects representing the types of the tensors accepted by the `Operation`. By default uses `[x.dtype.base_dtype for x in inputs]`. Operations that expect reference-typed inputs must specify these explicitly.
 *  **original\_op**: Optional. Used to associate the new `Operation` with an existing `Operation`(for example, a replica with the op that was replicated).
 *  **op\_def**: Optional. The `op_def_pb2.OpDef` proto that describes the op type that this `Operation` represents.

##### Raises: #####

*  **TypeError**: if control inputs are not Operations or Tensors, or if `node_def` is not a `NodeDef`, or if `g` is not a `Graph`, or if `inputs` are not tensors, or if `inputs` and `input_types` are incompatible.
 *  **ValueError**: if the `node_def` name is not valid.

--------------------

#### `tf.Operation.__str__()` ####

--------------------

#### `tf.Operation.colocation_groups()` ####

Returns the list of colocation groups of the op.

--------------------

#### `tf.Operation.node_def` ####

Returns a serialized `NodeDef` representation of this operation.

##### Returns: #####

A [`NodeDef`][NodeDef] protocol buffer.

--------------------

#### `tf.Operation.op_def` ####

Returns the `OpDef` proto that represents the type of this op.

##### Returns: #####

An [`OpDef`][OpDef] protocol buffer.

--------------------

#### `tf.Operation.values()` ####

DEPRECATED: Use outputs.

--------------------

## class tf.Tensor ##

代表一个操作的输出之一。

一个张量（ `Tensor`）是一个操作（`Operation`）的输出之一的符号句柄。 它不持有该输出的值， 而是提供了一个在 TensorFlow 的会话\[`Session`\] 中计算那些值的方式。

这个类有两个主要的目的：

1.  一个张量（ `Tensor` ）能够作为输入传递到另一个操作（ `Operation`）中。 This builds a dataflow connection between operations, which enables TensorFlow to execute an entire `Graph` that represents a large, multi-step computation.
2.  After the graph has been launched in a session, the value of the `Tensor` can be computed by passing it to [`Session.run()`][Session.run]. `t.eval()` is a shortcut for calling`tf.get_default_session().run(t)`.

In the following example, `c`, `d`, and `e` are symbolic `Tensor` objects, whereas `result` is a numpy array that stores a concrete value:

# 构建一个数据流图
    c = tf.constant([[1.0, 2.0], [3.0, 4.0]])
    d = tf.constant([[1.0, 1.0], [0.0, 1.0]])
    e = tf.matmul(c, d)
    
    # 构建一个 `Session` 来执行这个图
    sess = tf.Session()
    
    # Execute the graph and store the value that `e` represents in `result`.
    result = sess.run(e)

--------------------

#### `tf.Tensor.dtype` ####

The `DType` of elements in this tensor.

--------------------

#### `tf.Tensor.name` ####

The string name of this tensor.

--------------------

#### `tf.Tensor.value_index` ####

The index of this tensor in the outputs of its `Operation`.

--------------------

#### `tf.Tensor.graph` ####

The `Graph` that contains this tensor.

--------------------

#### `tf.Tensor.op` ####

The `Operation` that produces this tensor as an output.

--------------------

#### `tf.Tensor.consumers()` ####

Returns a list of `Operation`s that consume this tensor.

##### Returns: #####

A list of `Operation`s.

--------------------

#### `tf.Tensor.eval(feed_dict=None, session=None)` ####

Evaluates this tensor in a `Session`.

Calling this method will execute all preceding operations that produce the inputs needed for the operation that produces this tensor.

*N.B.* Before invoking `Tensor.eval()`, its graph must have been launched in a session, and either a default session must be available, or `session` must be specified explicitly.

##### Args: #####

*  **feed\_dict**: A dictionary that maps `Tensor` objects to feed values. See [`Session.run()`][Session.run] for a description of the valid feed values.
 *  **session**: (Optional.) The `Session` to be used to evaluate this tensor. If none, the default session will be used.

##### Returns: #####

A numpy array corresponding to the value of this tensor.

--------------------

#### `tf.Tensor.get_shape()` ####

Returns the `TensorShape` that represents the shape of this tensor.

The shape is computed using shape inference functions that are registered in the Op for each `Operation`. See [`TensorShape`][TensorShape] for more details of what a shape represents.

The inferred shape of a tensor is used to provide shape information without having to launch the graph in a session. This can be used for debugging, and providing early error messages. For example:

c = tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
    
    print(c.get_shape())
    ==> TensorShape([Dimension(2), Dimension(3)])
    
    d = tf.constant([[1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0]])
    
    print(d.get_shape())
    ==> TensorShape([Dimension(4), Dimension(2)])
    
    # Raises a ValueError, because `c` and `d` do not have compatible
    # inner dimensions.
    e = tf.matmul(c, d)
    
    f = tf.matmul(c, d, transpose_a=True, transpose_b=True)
    
    print(f.get_shape())
    ==> TensorShape([Dimension(3), Dimension(4)])

In some cases, the inferred shape may have unknown dimensions. If the caller has additional information about the values of these dimensions, `Tensor.set_shape()` can be used to augment the inferred shape.

##### Returns: #####

A `TensorShape` representing the shape of this tensor.

--------------------

#### `tf.Tensor.set_shape(shape)` ####

Updates the shape of this tensor.

This method can be called multiple times, and will merge the given `shape` with the current shape of this tensor. It can be used to provide additional information about the shape of this tensor that cannot be inferred from the graph alone. For example, this can be used to provide additional information about the shapes of images:

_, image_data = tf.TFRecordReader(...).read(...)
    image = tf.image.decode_png(image_data, channels=3)
    
    # The height and width dimensions of `image` are data dependent, and
    # cannot be computed without executing the op.
    print(image.get_shape())
    ==> TensorShape([Dimension(None), Dimension(None), Dimension(3)])
    
    # We know that each image in this dataset is 28 x 28 pixels.
    image.set_shape([28, 28, 3])
    print(image.get_shape())
    ==> TensorShape([Dimension(28), Dimension(28), Dimension(3)])

##### Args: #####

*  **shape**: A `TensorShape` representing the shape of this tensor.

##### Raises: #####

*  **ValueError**: If `shape` is not compatible with the current shape of this tensor.

#### Other Methods ####

--------------------

#### `tf.Tensor.__abs__(x, name=None)` ####

Computes the absolute value of a tensor.

Given a tensor of real numbers `x`, this operation returns a tensor containing the absolute value of each element in `x`. For example, if x is an input element and y is an output element, this operation computes y=|x|.

See [`tf.complex_abs()`][tf.complex_abs] to compute the absolute value of a complex number.

##### Args: #####

*  **x**: A `Tensor` or `SparseTensor` of type `float32`, `float64`, `int32`, or `int64`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` or `SparseTensor` the same size and type as `x` with absolute values.

--------------------

#### `tf.Tensor.__add__(x, y)` ####

Returns x + y element-wise.

*NOTE*: `Add` supports broadcasting. `AddN` does not. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `half`, `float32`, `float64`, `uint8`, `int8`, `int16`, `int32`, `int64`, `complex64`, `complex128`, `string`.
 *  **y**: A `Tensor`. Must have the same type as `x`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__and__(x, y)` ####

Returns the truth value of x AND y element-wise.

*NOTE*: `LogicalAnd` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor` of type `bool`.
 *  **y**: A `Tensor` of type `bool`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__bool__()` ####

Dummy method to prevent a tensor from being used as a Python `bool`.

This overload raises a `TypeError` when the user inadvertently treats a `Tensor` as a boolean (e.g. in an `if` statement). For example:

if tf.constant(True):  # Will raise.
      # ...
    
    if tf.constant(5) < tf.constant(7):  # Will raise.
      # ...

This disallows ambiguities between testing the Python value vs testing the dynamic condition of the `Tensor`.

##### Raises: #####

`TypeError`.

--------------------

#### `tf.Tensor.__div__(x, y)` ####

Returns x / y element-wise.

*NOTE*: `Div` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `half`, `float32`, `float64`, `uint8`, `int8`, `uint16`, `int16`, `int32`, `int64`, `complex64`, `complex128`.
 *  **y**: A `Tensor`. Must have the same type as `x`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__eq__(other)` ####

--------------------

#### `tf.Tensor.__floordiv__(x, y)` ####

Divides `x / y` elementwise, rounding toward the most negative integer.

The same as `tf.div(x,y)` for integers, but uses `tf.floor(tf.div(x,y))` for floating point arguments so that the result is always an integer (though possibly an integer represented as floating point). This op is generated by `x // y` floor division in Python 3 and in Python 2.7 with `from __future__ import division`.

Note that for efficiency, `floordiv` uses C semantics for negative numbers (unlike Python and Numpy).

`x` and `y` must have the same type, and the result will have the same type as well.

##### Args: #####

*  **x**: `Tensor` numerator of real numeric type.
 *  **y**: `Tensor` denominator of real numeric type.
 *  **name**: A name for the operation (optional).

##### Returns: #####

`x / y` rounded down (except possibly towards zero for negative integers).

##### Raises: #####

*  **TypeError**: If the inputs are complex.

--------------------

#### `tf.Tensor.__ge__(x, y, name=None)` ####

Returns the truth value of (x >= y) element-wise.

*NOTE*: `GreaterEqual` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `float32`, `float64`, `int32`, `int64`, `uint8`, `int16`, `int8`, `uint16`, `half`.
 *  **y**: A `Tensor`. Must have the same type as `x`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__getitem__(tensor, slice_spec, var=None)` ####

Overload for Tensor.**getitem**.

This operation extracts the specified region from the tensor. The notation is similar to NumPy with the restriction that currently only support basic indexing. That means that using a tensor as input is not currently allowed

Some useful examples:

# strip leading and trailing 2 elements
    foo = tf.constant([1,2,3,4,5,6])
    print(foo[2:-2].eval()) # => [3,4]
    
    # skip every row and reverse every column
    foo = tf.constant([[1,2,3], [4,5,6], [7,8,9]])
    print(foo[::2,::-1].eval()) # => [[3,2,1], [9,8,7]]
    
    # Insert another dimension
    foo = tf.constant([[1,2,3], [4,5,6], [7,8,9]])
    print(foo[tf.newaxis, :, :].eval()) # => [[[3,2,1], [9,8,7]]]
    print(foo[:, tf.newaxis, :].eval()) # => [[[3,2,1]], [[9,8,7]]]
    print(foo[:, :, tf.newaxis].eval()) # => [[[3],[2],[1]], [[9],[8],[7]]]
    
    # Ellipses (3 equivalent operations)
    print(foo[tf.newaxis, :, :].eval()) # => [[[3,2,1], [9,8,7]]]
    print(foo[tf.newaxis, ...].eval()) # => [[[3,2,1], [9,8,7]]]
    print(foo[tf.newaxis].eval()) # => [[[3,2,1], [9,8,7]]]

##### Notes: #####

*  `tf.newaxis` is `None` as in NumPy.
 *  An implicit ellipsis is placed at the end of the `slice_spec`
 *  NumPy advanced indexing is currently not supported.

##### Args: #####

*  **tensor**: An ops.Tensor object.
 *  **slice\_spec**: The arguments to Tensor.**getitem**.
 *  **var**: In the case of variable slice assignment, the Variable object to slice (i.e. tensor is the read-only view of this variable).

##### Returns: #####

The appropriate slice of “tensor”, based on “slice\_spec”.

##### Raises: #####

*  **ValueError**: If a slice range is negative size.
 *  **TypeError**: If the slice indices aren’t int, slice, or Ellipsis.

--------------------

#### `tf.Tensor.__gt__(x, y, name=None)` ####

Returns the truth value of (x > y) element-wise.

*NOTE*: `Greater` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__hash__()` ####

--------------------

#### `tf.Tensor.__init__(op, value_index, dtype)` ####

Creates a new `Tensor`.

##### Args: #####

*  **op**: An `Operation`. `Operation` that computes this tensor.
 *  **value\_index**: An `int`. Index of the operation’s endpoint that produces this tensor.
 *  **dtype**: A `DType`. Type of elements stored in this tensor.

##### Raises: #####

*  **TypeError**: If the op is not an `Operation`.

--------------------

#### `tf.Tensor.__invert__(x, name=None)` ####

Returns the truth value of NOT x element-wise.

##### Args: #####

*  **x**: A `Tensor` of type `bool`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__iter__()` ####

Dummy method to prevent iteration. Do not call.

NOTE(mrry): If we register **getitem** as an overloaded operator, Python will valiantly attempt to iterate over the Tensor from 0 to infinity. Declaring this method prevents this unintended behavior.

##### Raises: #####

*  **TypeError**: when invoked.

--------------------

#### `tf.Tensor.__le__(x, y, name=None)` ####

Returns the truth value of (x <= y) element-wise.

*NOTE*: `LessEqual` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__lt__(x, y, name=None)` ####

Returns the truth value of (x < y) element-wise.

*NOTE*: `Less` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__mod__(x, y)` ####

Returns element-wise remainder of division.

*NOTE*: `Mod` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `int32`, `int64`, `float32`, `float64`.
 *  **y**: A `Tensor`. Must have the same type as `x`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__mul__(x, y)` ####

Dispatches cwise mul for “Dense\*Dense” and “Dense\*Sparse”.

--------------------

#### `tf.Tensor.__neg__(x, name=None)` ####

Computes numerical negative value element-wise.

I.e., y=−x.

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `half`, `float32`, `float64`, `int32`, `int64`, `complex64`, `complex128`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__nonzero__()` ####

Dummy method to prevent a tensor from being used as a Python `bool`.

This is the Python 2.x counterpart to `__bool__()` above.

##### Raises: #####

`TypeError`.

--------------------

#### `tf.Tensor.__or__(x, y)` ####

Returns the truth value of x OR y element-wise.

*NOTE*: `LogicalOr` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor` of type `bool`.
 *  **y**: A `Tensor` of type `bool`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__pow__(x, y)` ####

Computes the power of one value to another.

Given a tensor `x` and a tensor `y`, this operation computes xy for corresponding elements in `x` and `y`. For example:

# tensor 'x' is [[2, 2], [3, 3]]
    # tensor 'y' is [[8, 16], [2, 3]]
    tf.pow(x, y) ==> [[256, 65536], [9, 27]]

##### Args: #####

*  **x**: A `Tensor` of type `float32`, `float64`, `int32`, `int64`, `complex64`, or `complex128`.
 *  **y**: A `Tensor` of type `float32`, `float64`, `int32`, `int64`, `complex64`, or `complex128`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`.

--------------------

#### `tf.Tensor.__radd__(y, x)` ####

Returns x + y element-wise.

*NOTE*: `Add` supports broadcasting. `AddN` does not. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__rand__(y, x)` ####

Returns the truth value of x AND y element-wise.

*NOTE*: `LogicalAnd` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor` of type `bool`.
 *  **y**: A `Tensor` of type `bool`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__rdiv__(y, x)` ####

Returns x / y element-wise.

*NOTE*: `Div` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__repr__()` ####

--------------------

#### `tf.Tensor.__rfloordiv__(y, x)` ####

Divides `x / y` elementwise, rounding toward the most negative integer.

Note that for efficiency, `floordiv` uses C semantics for negative numbers (unlike Python and Numpy).

`x` and `y` must have the same type, and the result will have the same type as well.

##### Args: #####

*  **x**: `Tensor` numerator of real numeric type.
 *  **y**: `Tensor` denominator of real numeric type.
 *  **name**: A name for the operation (optional).

##### Returns: #####

`x / y` rounded down (except possibly towards zero for negative integers).

##### Raises: #####

*  **TypeError**: If the inputs are complex.

--------------------

#### `tf.Tensor.__rmod__(y, x)` ####

Returns element-wise remainder of division.

*NOTE*: `Mod` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__rmul__(y, x)` ####

Dispatches cwise mul for “Dense\*Dense” and “Dense\*Sparse”.

--------------------

#### `tf.Tensor.__ror__(y, x)` ####

Returns the truth value of x OR y element-wise.

*NOTE*: `LogicalOr` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor` of type `bool`.
 *  **y**: A `Tensor` of type `bool`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor` of type `bool`.

--------------------

#### `tf.Tensor.__rpow__(y, x)` ####

Computes the power of one value to another.

Given a tensor `x` and a tensor `y`, this operation computes xy for corresponding elements in `x` and `y`. For example:

# tensor 'x' is [[2, 2], [3, 3]]
    # tensor 'y' is [[8, 16], [2, 3]]
    tf.pow(x, y) ==> [[256, 65536], [9, 27]]

##### Args: #####

##### Returns: #####

A `Tensor`.

--------------------

#### `tf.Tensor.__rsub__(y, x)` ####

Returns x - y element-wise.

*NOTE*: `Sub` supports broadcasting. More about broadcasting [here][]

##### Args: #####

*  **x**: A `Tensor`. Must be one of the following types: `half`, `float32`, `float64`, `int32`, `int64`, `complex64`, `complex128`.
 *  **y**: A `Tensor`. Must have the same type as `x`.
 *  **name**: A name for the operation (optional).

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__rtruediv__(y, x)` ####

Divides x / y elementwise, always producing floating point results.

The same as `tf.div` for floating point arguments, but casts integer arguments to floating point before dividing so that the result is always floating point. This op is generated by normal `x / y`division in Python 3 and in Python 2.7 with `from __future__ import division`. If you want integer division that rounds down, use `x // y` or `tf.floordiv`.

`x` and `y` must have the same numeric type. If the inputs are floating point, the output will have the same type. If the inputs are integral, the inputs are cast to `float32` for `int8` and `int16` and `float64` for `int32` and `int64` (matching the behavior of Numpy).

##### Args: #####

*  **x**: `Tensor` numerator of numeric type.
 *  **y**: `Tensor` denominator of numeric type.
 *  **name**: A name for the operation (optional).

##### Returns: #####

`x / y` evaluated in floating point.

##### Raises: #####

*  **TypeError**: If `x` and `y` have different dtypes.

--------------------

#### `tf.Tensor.__rxor__(y, x)` ####

x ^ y = (x | y) & ~(x & y).

--------------------

#### `tf.Tensor.__str__()` ####

--------------------

#### `tf.Tensor.__sub__(x, y)` ####

Returns x - y element-wise.

*NOTE*: `Sub` supports broadcasting. More about broadcasting [here][]

##### Args: #####

##### Returns: #####

A `Tensor`. Has the same type as `x`.

--------------------

#### `tf.Tensor.__truediv__(x, y)` ####

Divides x / y elementwise, always producing floating point results.

##### Args: #####

*  **x**: `Tensor` numerator of numeric type.
 *  **y**: `Tensor` denominator of numeric type.
 *  **name**: A name for the operation (optional).

##### Returns: #####

`x / y` evaluated in floating point.

##### Raises: #####

*  **TypeError**: If `x` and `y` have different dtypes.

--------------------

#### `tf.Tensor.__xor__(x, y)` ####

x ^ y = (x | y) & ~(x & y).

--------------------

#### `tf.Tensor.device` ####

The name of the device on which this tensor will be produced, or None.

[https_www.tensorflow.org_versions_r0.12_api_docs_python_framework.html_building-graphs]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#building-graphs
[http_wiki.jikexueyuan.com_project_tensorflow-zh_api_docs_python_framework.html_Graph]: http://wiki.jikexueyuan.com/project/tensorflow-zh/api_docs/python/framework.html#Graph
[tf.get_default_graph]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#get_default_graph
[Graph.as_default]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#Graph.as_default
[import_graph_def]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#import_graph_def
[C_ Session API]: https://www.tensorflow.org/versions/r0.12/api_docs/cc/index.html
[GraphDef]: https://www.tensorflow.org/code/tensorflow/core/framework/graph.proto
[Link 1]: https://baike.baidu.com/item/%E7%8B%AC%E7%AB%8B/3259
[QueueRunner]: https://www.tensorflow.org/versions/r0.12/api_docs/python/train.html#QueueRunner
[DeviceNameUtils]: https://www.tensorflow.org/code/tensorflow/core/util/device_name_utils.h
[tf.GraphKeys.GLOBAL_VARIABLES]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#GraphKeys
[GraphDef version]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#Graph.graph_def_version
[tf.matmul]: https://www.tensorflow.org/versions/r0.12/api_docs/python/math_ops.html#matmul
[Graph.create_op]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#Graph.create_op
[Session.run]: https://www.tensorflow.org/versions/r0.12/api_docs/python/client.html#Session.run
[NodeDef]: https://www.tensorflow.org/code/tensorflow/core/framework/node_def.proto
[OpDef]: https://www.tensorflow.org/code/tensorflow/core/framework/op_def.proto
[TensorShape]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#TensorShape
[tf.complex_abs]: https://www.tensorflow.org/versions/r0.12/api_docs/python/framework.html#tf_complex_abs
[here]: http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html