PSv2: Replace existing tf.distribute.experimental.ParameterServerStrategy usage with tf.compat.v1.distribute.experimental.ParameterServerStrategy to prepare for the upcoming TF2 ParameterServerStrategy API release.

rchao · tensorflower-gardener · commit ed086c8cdd3f · 2020-10-01T10:40:20.000-07:00
The practically only difference for API endpoint switch is the monitoring from V2 to V1, for those who were using `tf.distribute.experimental.ParameterServerStrategy`. It's not supported in V2 and should be tracked as V1 anyway.

PiperOrigin-RevId: 334847114
diff --git a/tensorflow_model_optimization/python/core/sparsity/keras/prune_distributed_test.py b/tensorflow_model_optimization/python/core/sparsity/keras/prune_distributed_test.py
@@ -33,7 +33,7 @@ def _distribution_strategies():
       tf.distribute.experimental.MultiWorkerMirroredStrategy(),
       tf.distribute.MirroredStrategy(),
       # TODO(pulkitb): Add parameter_server
-      # tf.distribute.experimental.ParameterServerStrategy,
+      # tf.distribute.experimental.ParameterServerStrategy once ready in TF2.
       tf.distribute.OneDeviceStrategy('/cpu:0'),
   ]
 

Original file line number	Diff line number	Diff line change
`@@ -33,7 +33,7 @@ def _distribution_strategies():`
`33`	`33`	`tf.distribute.experimental.MultiWorkerMirroredStrategy(),`
`34`	`34`	`tf.distribute.MirroredStrategy(),`
`35`	`35`	`# TODO(pulkitb): Add parameter_server`
`36`		`- # tf.distribute.experimental.ParameterServerStrategy,`
	`36`	`+ # tf.distribute.experimental.ParameterServerStrategy once ready in TF2.`
`37`	`37`	`tf.distribute.OneDeviceStrategy('/cpu:0'),`
`38`	`38`	`]`
`39`	`39`