Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Feedback PR #10

Open
wants to merge 21 commits into
base: candidate
Choose a base branch
from
Open

Feedback PR #10

wants to merge 21 commits into from

Conversation

axelmagn
Copy link
Owner

This is a mega-PR of all of my work so far. Please feel free to provide feedback in the form of both thread comments and inline code comments. All feedback will be considered and responded to. While reviewing, please consider the following questions:

  • What is missing? This is an incomplete project. What would make it complete enough for a first release?
  • What needs to redone / done better? Even if it's a large change, we want this code to be representative of best practices. Help me identify where I've made the wrong choices.
  • What feels good / should be expanded / should be repeated? Where are the places where I am on the right track, and should expand or repeat?

I will integrate requested changes into the feedback branch, and coordinate with participants until we reach a point where designated reviewers have provided their LGTM.

axelmagn and others added 21 commits May 25, 2021 13:26
This adds an MLOps Manager that may be used to create simple vertex AI applications.

Key additions:

* `mlops_manager start_app` command
* `hello_pipeline` example pipeline
* cloud build config in `config/build/cloudbuild.yaml`
fashion-mnist classifier pipeline that can be invoked via cloud build
- build_pipeline command to build pipeline from config
- PipelineRunner to encapsulate pipeline management logic
- first tests for PipelineRunner
  - tests throw warnings that I cannot replicate when invoking by
    command. Currently confused as to why.
- run_pipeline command and runner function
- update run_pipeline.yaml cloud build config
- naming refactor
- fashion_mnist uses app image
- pipeline runner renamed to harness
- Remove dependency of docker image on sdist for speed
- Vertex custom training job
- create new pipelines template
- improve templating logic to be more extensible
Give existing examples a more regular format.
- Add standalone task decorator.
 - task decorator that can be used to quickly annotate trainer tasks
 - add pylintrc (not yet conformant)
 - convert fashion_mnist trainer task to use decorator
- implement build, release, and deploy configs for cloud build
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant