Release 0.5.2

[nnansters]

Hey all,

Niels from the NannyML engineering team here. Quickly dropping the news that we've released NannyML 0.5.2 🚀

Installing / upgrading

You can get this latest version by using pip:

pip install -U nannyml

What's new?

Easy NannyML integration using the nml CLI and the associated container!

Aah, holidays are here 🌴! If only there was a way of not having to edit the timestamps in your notebook and then running it every day... There's a better way to automate NannyML now!

Introducing... 🥁 the nml CLI 🥁

No more custom driver scripts to run NannyML. Just type nml run and find the output as attached.
There is a catch of course: NannyML still needs to know some things about your dataset. This is what the nann.yml (get it? 😸 ) configuration file is for. We've kept it as minimal as possible, so it shouldn't be too much of a hassle.

Some features I particularly like:

• Support for reading from and writing to cloud storage such as S3, GCS buckets or Azure Storage

• Some basic support for templating file paths, allowing you to write the output of your Kubernetes CronJob to s3://nml-results/{{year}}/{{month}}/{{day}}

But wait... we've made it even easier to integrate NannyML into your workflows! You can now use the official NannyML container to run the CLI as a containerized workload as part of your workflow engine!

docker run -v /path/to/config/dir/:/config/ nannyml/nannyml nml run

Want to know more? Check out the CLI documentation over on RTD.

What's changed?

• Some small housekeeping (cleanup and docs improvements).

See the full list on the changelog.

What's up next?

Expect a new update soon! Down in the catacombs of NannyML we're reviewing the concepts of confidence, stability and minimal chunk sizes. More to follow soon. Very soon!
In the meantime we look forward to getting your feedback on the new CLI and seeing what we can improve!
But first and foremost: we hope you've either already enjoyed or are looking forward to some nice holidays first! 🌴
Hear you soon!