Add more documentation to recommend users to set `drop_remainder=False` when...

Add more documentation to recommend users to set `drop_remainder=False` when exhausting the eval dataset in Orbit.

PiperOrigin-RevId: 409539451

orbit/standard_runner.py

@@ -265,7 +265,10 @@ class StandardEvaluator(runner.AbstractEvaluator, metaclass=abc.ABCMeta):
 
     Args:
       eval_dataset: A `tf.nest`-compatible structure of `tf.data.Dataset` or
-        `DistributedDataset`.
+        `DistributedDataset`. On TPUs, if users want to exaust the dataset
+        without specifying number of eval steps, it is recommended to set
+        `drop_remainder=False` when batching the dataset, so the infrastructure
+        can handle the last partial batch properly.
       options: An `orbit.StandardEvaluatorOptions` instance.
     """
     options = options or StandardEvaluatorOptions()