[Debugger] tensorflow_nlp_sentiment_analysis training job runs much longer than expected

https://github.com/aws/amazon-sagemaker-examples/tree/master/sagemaker-debugger/tensorflow_nlp_sentiment_analysis

Since the notebook published with the public SDK version, the training job runs for over an hour with epoch=25.
It's expected to run for ~500 seconds.

Open the sagemaker-debugger/tensorflow_nlp_sentiment_analysis notebook and run the training job with epoch=25 using the public SDK version. Compare its runtime with the expected roughly 500 seconds and inspect the notebook's training configuration for the source of the longer run. Done means the job completes near the expected duration.

Written by the indexing model from the issue text.