Release checklist

• Update release notes in draft release
• Update hyperparameters in draft release
• Update contributor list in draft release
• Package and upload checkpoint, models, and audio examples
  • Remember to include alphabet.txt in checkpoint package, as it's no longer needed for the models
  • If packaging from a best_dev checkpoint, cp best_dev_checkpoint checkpoint to avoid problems when loading the checkpoint from code that doesn't handle the separate checkpoint/best_dev_checkpoint logic
• Upload models somewhere else (e.g. personal fork release) and link prod model tests and examples tests to new URLs
• Update all references to previous release in documentation
• Bump VERSION file to new version, remove note from README.rst about "This documentation applies to the master branch of DeepSpeech only. [...]".
• Merge PR bumping VERSION file, then tag new version on the merge commit
• Push the tag
• Publish the draft release pointing to the existing tag
• Monitor task group created from the tag to make sure everything is built and published
• Manually check package repositories after publish is complete
  • https://pypi.org/project/deepspeech/
  • https://pypi.org/project/deepspeech-gpu/
  • https://www.npmjs.com/package/deepspeech
  • https://www.npmjs.com/package/deepspeech-gpu
  • https://www.nuget.org/packages/DeepSpeech
  • https://www.nuget.org/packages/DeepSpeech-GPU
• Make the new version the default on https://deepspeech.readthedocs.io/
• Add the notice about "This documentation applies to the master branch of DeepSpeech only. [...]" to README.rst again