I had a libgl.so.1 library not found error when training the model. Updating the training docker script to add the following libraries helped: ffmpeg, libsm6, libxext6.
Also, added an ECR image output to a text file at the end of the build_and_push.sh scripts to allow the calling notebooks to read the ECR image ID directly from this file instead of asking the reader to manually enter it.
A few typos
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
… image push to ECR.
Issue #, if available:
Description of changes:
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.