control flow v2 (cfv2) is an improved version of control flow in TensorFlow
with support for higher order derivatives. Enabling cfv2 will change the
graph/function representation of control flow, e.g., tf.while_loop and
tf.cond will generate functional While and If ops instead of low-level
Switch, Merge etc. ops. Note: Importing and running graphs exported
with old control flow will still be supported.
Calling tf.enable_control_flow_v2() lets you opt-in to this TensorFlow 2.0
feature.
[null,null,["Last updated 2023-10-06 UTC."],[],[],null,["# tf.compat.v1.enable_control_flow_v2\n\n\u003cbr /\u003e\n\n|-----------------------------------------------------------------------------------------------------------------------------------------|\n| [View source on GitHub](https://github.com/tensorflow/tensorflow/blob/v2.14.0/tensorflow/python/ops/control_flow_v2_toggles.py#L25-L45) |\n\nUse control flow v2. \n\n tf.compat.v1.enable_control_flow_v2()\n\ncontrol flow v2 (cfv2) is an improved version of control flow in TensorFlow\nwith support for higher order derivatives. Enabling cfv2 will change the\ngraph/function representation of control flow, e.g., [`tf.while_loop`](../../../tf/while_loop) and\n[`tf.cond`](../../../tf/cond) will generate functional `While` and `If` ops instead of low-level\n`Switch`, `Merge` etc. ops. Note: Importing and running graphs exported\nwith old control flow will still be supported.\n\nCalling tf.enable_control_flow_v2() lets you opt-in to this TensorFlow 2.0\nfeature.\n| **Note:** v2 control flow is always enabled inside of tf.function. Calling this function is not required."]]
View source on GitHub