Update checkpoint_dir format.

Author: Chuyang Deng

src/sagemaker_tensorflow_container/training.py

@@ -196,7 +196,7 @@ def main():
     # If the training job is part of the multiple training jobs for tuning, we need to append the training job name to
     # model_dir in case they read from/write to the same object
     if '_tuning_objective_metric' in hyperparameters:
-        env.hyperparameters['model_dir'] = '{}-{}'.format(hyperparameters.get('model_dir'), env.job_name)
+        env.hyperparameters['model_dir'] = '{}/{}/checkpoints'.format(hyperparameters.get('model_dir'), env.job_name)
 
     s3_utils.configure(env.hyperparameters.get('model_dir'), os.environ.get('SAGEMAKER_REGION'))
     logger.setLevel(env.log_level)

test/unit/test_training.py

@@ -281,5 +281,5 @@ def test_main_tunning_model_dir(configure_s3_env, read_hyperparameters, training
     training_env.return_value = single_machine_training_env
     os.environ['SAGEMAKER_REGION'] = REGION
     training.main()
-    expected_model_dir = '{}-{}'.format(MODEL_DIR, single_machine_training_env.job_name)
+    expected_model_dir = '{}/{}/checkpoints'.format(MODEL_DIR, single_machine_training_env.job_name)
     configure_s3_env.assert_called_once_with(expected_model_dir, REGION)