A Framework for creating a boilerplate template for ai projects that are ready for MLOps
A small framework for python projects that are mainly progress based.
The main princible of the framework are
A Project is the composition of phases and the backend.
A Phase has a
main Method and can export data.
A stage is a group of Phases and only have a name. A Stage can be run seperatly with
Storage and Exporters
You can add diffrent storage-engines to your project, with
project.addStorage(Engine()). A storage you be inherited from
pyPhases.storage.Storage and implement the methods
read(path: str) and
write(path: str, data: bytes).
The order is important and the Storages should be ordered from fast to slow.
By Default there is a memory storage, that will save the data in the project, but is not persitent. The default persistent data layer is the filesystem(
An Exporter can be registered to transform an Instance or primitive type into a
byte string (
export(obj : MyObject): bytes) and vice versa (
There is a default
ObjectExporter, that is based on pyarror and is compatible with diffrent fromats like pandas Dataframes and numpy arrays.
When a phase wants to register data (
self.project.registerData("myDataId", myData) within the phase), the data is passed to an exporter. If an exporter is found the data will be passed to alle the storages. They will save the data somewhere (persitent or not).
reading the data
A phase can request data with
self.project.getData("myDataId", MyDataType). The Data will be passed sequential to the storage layer and will pass the data from the first storage that is able to get it. If no storage can provide the data, the project will search for a phase that exports this data-id and run that specific phase.
This is a example data layer with 3 storages: memory, file, database (
python setup.py sdist bdist_wheel
twine upload dist/*
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size pyPhases-0.1.5-py3-none-any.whl (16.0 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size pyPhases-0.1.5.tar.gz (11.0 kB)||File type Source||Python version None||Upload date||Hashes View|