Format _train_step_from_last_restored_checkpoint_path as "%s" since i…

…t might be None. Otherwise the logging.info dies with a formatting error. PiperOrigin-RevId: 559545134 Change-Id: I779384c10a99fbdc0f66a26201f79c48b89a1be7
tensorflow · Aug 23, 2023 · b344b20 · b344b20
1 parent eca1093
commit b344b20
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/tf_agents/policies/py_tf_eager_policy.py b/tf_agents/policies/py_tf_eager_policy.py
@@ -279,7 +279,7 @@ def update_from_checkpoint(self, checkpoint_path: Text):
       # In case the checkpoint format is not as expected.
       self._train_step_from_last_restored_checkpoint_path = None
     logging.info(
-        'Checkpoint path: %s, train step: %d',
+        'Checkpoint path: %s, train step: %s',
         checkpoint_path,
         self._train_step_from_last_restored_checkpoint_path,
     )