TensorFlow.Output (original) (raw)

Synopsis

• newtype ControlNode = ControlNode {
  • unControlNode :: NodeName
    }
• newtype Device = Device {
  • deviceName :: Text
    }
• newtype NodeName = NodeName {
  • unNodeName :: Text
    }
• data OpDef = OpDef {
  • _opName ::
  • _opType ::
  • _opAttrs :: !(Map Text AttrValue)
  • _opInputs :: [Output]
  • _opControlInputs :: [NodeName]
    }
• opName :: Lens' OpDef PendingNodeName
• opType :: Lens' OpDef OpType
• opAttr :: Attribute a => Text -> Lens' OpDef a
• opInputs :: Lens' OpDef [Output]
• opControlInputs :: Lens' OpDef [NodeName]
• newtype OpType = OpType {
  • unOpType :: Text
    }
• newtype OutputIx = OutputIx {
  • unOutputIx :: Int
    }
• data Output = Output {
  • outputIndex ::
  • outputNodeName ::
    }
• output :: OutputIx -> NodeName -> Output
• data PendingNodeName
  • = ExplicitName
  • | ImplicitName

Documentation

newtype ControlNode Source #

A type of graph node which has no outputs. These nodes are valuable for causing side effects when they are run.

newtype Device Source #

A device that a node can be assigned to. There's a naming convention where the device names are constructed from job and replica names.

Ops

newtype NodeName Source #

The name of a node in the graph. This corresponds to the proto field NodeDef.name. Includes the scope prefix (if any) and a unique identifier (if the node was implicitly named).

data OpDef Source #

Op definition. This corresponds somewhat to the NodeDef proto.

Instances

Instances details

newtype OpType Source #

The type of op of a node in the graph. This corresponds to the proto field NodeDef.op.

data PendingNodeName Source #

The name specified for an unrendered Op. If an Op has an ImplicitName, it will be assigned based on the opType plus a unique identifier. Does not contain the "scope" prefix.