Training Job Fails for Compression RL Notebook

In the notebook: https://github.com/awslabs/amazon-sagemaker-examples/blob/master/reinforcement_learning/rl_network_compression_ray_custom/rl_network_compression_ray_custom.ipynb

Training job fails when executing the estimator:

`estimator.fit(cifar_inputs)`

It returns:

`
UnexpectedStatusException: Error for Training job rl-nnc-2020-08-xx-xx-xx-xx-xxx: Failed. Reason: AlgorithmError: ExecuteUserScriptError: Command "/usr/bin/python train-ray.py --rl.training.checkpoint_freq 1 --rl.training.stop.training_iteration 1 --s3_bucket sagemaker-us-west-2-xxxxxxxxxxxx`

To reproduce error:

1. Launch SM Notebook instance
2. Clone https://github.com/awslabs/amazon-sagemaker-examples.git
3. Open https://github.com/awslabs/amazon-sagemaker-examples/blob/master/reinforcement_learning/rl_network_compression_ray_custom/rl_network_compression_ray_custom.ipynb
4. Click "Cell" -> "Run All"

Start with reinforcement_learning/rl_network_compression_ray_custom/rl_network_compression_ray_custom.ipynb and reproduce the failure by running all cells in an Amazon SageMaker Notebook instance. Inspect the estimator.fit(cifar_inputs) execution and the reported train-ray.py AlgorithmError. Done means the notebook completes training without the UnexpectedStatusException.

Written by the indexing model from the issue text.