dataflow.FlowData

Applicability

Product	Supported
Atlas A3 training product/Atlas A3 inference product	√
Atlas A2 training product/Atlas A2 inference product	√
Atlas 200I/500 A2 inference product	x
Atlas inference product	x
Atlas training product	x

Function Description

Indicates data nodes in a DataFlow graph. Each FlowData corresponds to an input.

Prototype

FlowData(data_cls=Tensor, schema:Optional[TensorDesc]=None, name=None)

Parameters

Parameter	Data Type	Description
data_cls	class	Currently, only the default Tensor is supported, indicating that FlowData receives tensor data.
schema	Optional[TensorDesc]	Description of data_cls. Currently, data_cls supports only tensors, so the value of schema is TensorDesc.
name	str	Node name. The framework automatically ensures that the node name is unique. If this parameter is not set, names such as FlowData, FlowData_1, FlowData_2, and the like will be automatically generated.

Returns

None is returned in normal scenarios.

If raise DfException is returned, the parameter type is incorrect. You can catch DfException and retrieve its error_code and message attributes to check the specific error code and error details. For details, see DataFlow Error Codes.

Examples

import dataflow as df
# Create an input node without specifying the data type and name. If you do not specify the name, the framework assigns a unique name FlowData FlowData_1 to the node.
data = df.FlowData()
# Create an input node that accepts the int32 data type and sets the shape to [1].
data = df.FlowData(schema=df.TensorDesc(df.DT_INT32, [1]))
# Create an input node named data0 without specifying the data type.
data = df.FlowData(name="data0")
# Create an input node whose data type is int32, shape is [1], and name is data0.
data = df.FlowData(schema=df.TensorDesc(df.DT_INT32, [1]), name="data0")

Constraints

None

Parent topic: Graph Construction APIs