python - TensorFlow - What's the difference between NodeDef and OpDef?

Question

I would like to know the difference bewteen NodeDef and OpDef in TensorFlow.

tf.Operation represents a graph node in a TensorFlow Graph. It takes node_def and a graph as input. What makes me confused is that it also takes op_def as an optional input. In my understanding, a node can only correspond to one operation. So why bother?

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

Answer 1

According to proper tensorflow documentation:

Optional. The op_def_pb2.OpDef proto that describes the op type that this Operation represents.

It's an optional argument that allows you to describe the op type of your Operation.

https://www.tensorflow.org/api_docs/python/tf/Operation