The repository now contains the following patterns that have been implemented:
1. Classical / Tabular Machine Learning Model
a. Supports Azure DevOps and GitHub Actions including the deployment of Infrastructure with both the platforms.
b. Supports Azure Data Explorer based Monitoring, Data Drift and Anomaly Detection. It is enabled for Terraform and can be invoked via Python SDKv1 or Azure ML CLI v2 (aml-cli-v2).
c. Can deploy both online as well as batch end points.
d. Supports MLOPs pipelines with Azure ML (AML) CLI v2, Python SDK V1 and V2.
e. Contains Responsible AI and python test modules.
f. Support for Feathr Feature store.
2. Computer Vision (CV) Model
a. Supports Azure DevOps and GitHub Actions. Note that GitHub Actions are only working for Azure ML CLI v2 (aml-cli-v2).
b. Supports MLOPs pipelines with Azure ML (AML) CLI v2, and Python SDK V1.
c. Can deploy both online as well as batch end points.
3. Natural Language Processing (NLP) Model
a. Supports for Azure DevOps and GitHub Actions. Note that GitHub Actions are only working for Azure ML CLI v2 (aml-cli-v2).
b. Supports MLOPs pipelines with Azure ML (AML) CLI v2, and Python SDK V2.
c. Can deploy both online as well as batch end points.
Other Patterns included in the release:
- Improve documentation for GitHub Actions and Azure DevOps available on the main GitHub Link
- Templates for using individual / repeatable steps in your template. These templates are available for GitHub Actions and Azure DevOps (CLI and SDK). The Template repo can be found here: https://github.com/Azure/mlops-templates
- Registration of Multiple Datasets
- Using 3rd party / external containers. Support for dependabot python package scans via pip install in docker container.
- Support for secure workspaces
- Quick Deploy Examples with Azure DevOps (ADO); GitHub and Microsoft Learn
- Feathr Feature Store: Integration of Feathr as an enterprise scale feature store in the MLOps V2 extended architectures, deployment of Feathr using Terraform script and running a simple classical ML example.
What's Changed
- Update main-jul31 by @mariamedp in #46
- exp and display parameters by @sbaidachni in #42
- Implement unit tests for CV data science code by @jfomhover in #36
- SDK - Run ID included in the data paths in pipeline definition by @mariamedp in #44
- SDK classical - adapt project to template changes by @mariamedp in #47
- Feature/cmr addmainentryforymls by @chrey-gh in #48
- SDK CV training by @mariamedp in #50
- SDK classical - adapt scoring pipeline to template changes by @mariamedp in #51
- NLP summarization pipeline by @jfomhover in #43
- GHA - Classical implementation by @djdean in #54
- Gha by @djdean in #56
- tfsec ws by @murggu in #58
- increased instance node size to fix out of memory error by @cindyweng in #59
- Implement NLP summarization online deployment by @jfomhover in #57
- Fix NLP builds post testing by @jfomhover in #60
- Fix NLP pipeline extensions by @jfomhover in #62
- Align extension in devops with extension of file by @jfomhover in #63
- Monitoring by @nicoleserafino in #61
- fix endpoint name being too long by @cindyweng in #64
- remove lib needed for node.js by @cindyweng in #67
- Update deploy-model-training-pipeline-classical.yml by @cindyweng in #66
- disable monitoring by @nicoleserafino in #68
- remove data-explorer tf outputs & typo by @nicoleserafino in #69
- Add arg to force comparison and always register/deploy model by @jfomhover in #70
- Fix force comparison argument value by @jfomhover in #71
- Update tf-ado-deploy-infra.yml by @cindyweng in #72
- Tf name fix by @cindyweng in #75
- Gha endpoint update by @djdean in #77
- Gha endpoint update by @djdean in #79
- Update deploy-online-endpoint-pipeline.yml by @cindyweng in #80
- Monitoring by @nicoleserafino in #84
- Merging the Jul-31 Release into Main by @setuc in #85
- Hotfix: add custom callback to avoid azureml 100 params limitation by @jfomhover in #86
- Fix brackets in the train.yaml file by @setuc in #87
- Bump protobuf from 3.20.1 to 3.20.2 in /cv/aml-cli-v2/data-science by @dependabot in #90
- Dev by @cindyweng in #91
- re add train-requirements.txt to environments directory by @maggiemhanna in #97
- Bump joblib from 1.0.0 to 1.2.0 in /classical/aml-cli-v2/data-science/experiment by @dependabot in #96
- Bump joblib from 1.0.0 to 1.2.0 in /classical/aml-cli-v2/data-science/environment by @dependabot in #98
- update aml tf module to avoid forcing replacement on rerun (#94) by @lindacmsheard in #95
- Update bicep-ado-deploy-infra.yml by @kevball2 in #88
- Updating template path for dev infra config by @kevball2 in #100
- updated comments for RAI to not refer to own org by @cindyweng in #102
- Feature/rai aml cli v2 by @cindyweng in #104
- Update config-infra-prod.yml by @cindyweng in #105
- Fix/reinstate mlflow by @cindyweng in #106
- Tfsec rollback by @cindyweng in #109
- rollback tf sec ws by @cindyweng in #110
- Create codeql.yml by @setuc in #112
- Feature/sdkv2-samer by @samelhousseini in #113
- Jfomhover/nlpsdk by @jomedinagomez in #115
- Feature data asset by @maggiemhanna in #103
- Feature/sdkv2 by @cindyweng in #116
- NLP SDK implementation by @jfomhover in #107
- Feature/sdkv2 by @cindyweng in #117
- Fixes for GitHub Actions Deployments by @sdonohoo in #119
- GHA CV pipeline fixes by @sdonohoo in #120
- Gha nlp by @djdean in #118
- GitHub Codespaces support - added Devcontainer by @michalmar in #92
- Main dec31 by @setuc in #121
- Fixes for classical training yaml format and data input by @sdonohoo in #122
- Merging the documentation from the main branch. by @setuc in #124
New Contributors
- @sbaidachni made their first contribution in #42
- @jfomhover made their first contribution in #36
- @dependabot made their first contribution in #90
- @lindacmsheard made their first contribution in #95
- @kevball2 made their first contribution in #88
- @samelhousseini made their first contribution in #113
- @jomedinagomez made their first contribution in #115
- @sdonohoo made their first contribution in #119
- @michalmar made their first contribution in #92
Full Changelog: v1.0.0...v1.1.0
