NLP summarization pipeline

[jfomhover]

PR into Azure/mlops-v2

Checklist

I have:

• read and followed the contributing guidelines

Changes

• Adds the NLP summarization pipeline based on @AliMahmoudzadeh's AzureML-example [link]
• Implement a dockerfile/env for the job
• Test the training end-to-end
• Add placeholders for online deployment

Overview

The pipeline consists in fine tuning a summarization model on pubmed, then comparing its metrics with a baseline model (not fine tuned). If the comparison is successful, then the model gets registered officially as a Model in AzureML.

[lgtm-com]

This pull request introduces 8 alerts when merging a3cc52f into 29e6a1a - view on LGTM.com

new alerts:

• 4 for Unused import
• 3 for Unused local variable
• 1 for Variable defined multiple times

[lgtm-com]

This pull request introduces 8 alerts when merging 6ee3186 into 29e6a1a - view on LGTM.com

new alerts:

• 4 for Unused import
• 3 for Unused local variable
• 1 for Variable defined multiple times

[lgtm-com]

This pull request introduces 8 alerts when merging b06ca77 into f4327f6 - view on LGTM.com

new alerts:

• 4 for Unused import
• 3 for Unused local variable
• 1 for Variable defined multiple times

[lgtm-com]

This pull request introduces 8 alerts when merging 654c269 into f4327f6 - view on LGTM.com

new alerts:

• 4 for Unused import
• 3 for Unused local variable
• 1 for Variable defined multiple times

[lgtm-com]

This pull request introduces 1 alert when merging 036841b into f4327f6 - view on LGTM.com

new alerts:

• 1 for Unused import

[jfomhover]

• remove custom env
• simplify custom env, add env registration to devops pipeline
• remove unused py file
• remove section in yaml for registering data (not used)
• remove online deployment files (for now, use 2nd pr later)
• create register.py with a flag to actually register or not

[lgtm-com]

This pull request introduces 8 alerts when merging 68b5f45 into 7f0a46b - view on LGTM.com

new alerts:

• 6 for Unused import
• 2 for Unused local variable