tf.group

TensorFlow 1 version View source on GitHub

Create an op that groups multiple operations.

Used in the notebooks

Used in the guide Used in the tutorials

When this op finishes, all ops in inputs have finished. This op has no output.

When operating in a v1-style graph context, ops are not executed in the same order as specified in the code; TensorFlow will attempt to execute ops in parallel or in an order convienient to the result it is computing. tf.group allows you to request that one or more results finish before execution continues.

tf.group creates a single op (of type NoOp), and then adds appropriate control dependencies. Thus, c = tf.group(a, b) will compute the same graph as this:

with tf.control_dependencies([a, b]):
    c = tf.no_op()

See also tf.tuple and tf.control_dependencies.

*inputs Zero or more tensors to group.
name A name for this operation (optional).

An Operation that executes all its inputs.

ValueError If an unknown keyword argument is provided.