https://github.com/omegaml/omegaml
MLOps simplified. One platform, all the functionality you need. Swiss made
https://github.com/omegaml/omegaml
artificial-intelligence celery data-science deploy docker-image jupyter-notebook machine-learning mlops productized pytorch scikit-learn tensorflow
Last synced: 27 days ago
JSON representation
MLOps simplified. One platform, all the functionality you need. Swiss made
- Host: GitHub
- URL: https://github.com/omegaml/omegaml
- Owner: omegaml
- License: apache-2.0
- Created: 2018-12-10T13:20:15.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2025-04-03T16:26:43.000Z (28 days ago)
- Last Synced: 2025-04-03T16:39:27.510Z (28 days ago)
- Topics: artificial-intelligence, celery, data-science, deploy, docker-image, jupyter-notebook, machine-learning, mlops, productized, pytorch, scikit-learn, tensorflow
- Language: Python
- Homepage: https://omegaml.github.io/omegaml/
- Size: 93.9 MB
- Stars: 98
- Watchers: 6
- Forks: 14
- Open Issues: 70
-
Metadata Files:
- Readme: README.dev
- Contributing: CONTRIBUTING
- License: LICENSE
- Security: SECURITY.md
Awesome Lists containing this project
README
What you need
-------------
Docker
Docker-compose
RabbitMq (docker image)
MongoDB (docker image
To use behave
https://sites.google.com/a/chromium.org/chromedriver/downloads
Dependencies
------------
* miniconda
* install all packages
$ conda create -n omegaml-ce python==3.9
$ pip install -e .[all]
Execute locally
---------------
Start mongo db & rabbitmq for dev purpose. Will be served on port 27019
$ docker-compose -f docker-compose-dev.yml up -d
$ scripts/initlocal.sh
Start services using honcho::
$ honcho start worker notebook restapi
Test locally
------------
There are different options:
1. Run the unit tests in place
$ make test
2. Run everything from local sources so you can interact with it
$ scripts/rundev.sh
3. Run everything from local sources then run the livetest against it (headless)
$ make devtest
4. Build omegaml docker image and run livetest locally
$ make image
Build a release for testing
---------------------------
$ make livetest
1. Builds the pip package
2. Runs livetest against it
If you just want to build the pip package without livetest, run make dist
Build a release for test-distribution (test pypi)
-------------------------------------------------
$ make release-test
1. runs make test, dist
2. uploads to pypi
3. runs livetest (which downloads from testpypi first)
Build a release for prod pypi
-----------------------------
$ make release-prod
1. runs make test, dist
2. uploads to pypi
3. runs livetest (which downloads from pypi first)
Build the docker image for release
----------------------------------
$ make release-docker
1. runs a local livetest, building the omegaml image
2. pushes the omegaml image to :version and :latest
Note since we already run the livetest against the newly built image,
then push the very same, we do not run livetest again. It would just
execute the very same livetest again, against the same image, assuming
that the docker push was successful.
Update THIRDPARTY license file
------------------------------
$ make thirdparty