Plasma is an open source tool which can be used to create and maintain end to end ML Workflows. Plasma breaks down a machine learning pipeline into smaller reusable blocks of code called ‘components’. The sequence in which these components are to be executed and the parameters which each component accepts and returns are defined in a simple YAML ‘workflow’ file.
Due to this, the ML codebase becomes a collection of reusable and modular components. Also, Data Scientists don’t need to edit code for running the workflow and changing parameters. This makes the workflow more stable.
Plasma also includes an in-built component manager which allows users to search an online registry of open-source components. This can greatly reduce the amount of time needed to develop the workflow.
Plasma is a fairly new tool and might not support everything your ML workflow requires. Hence, new component/feature requests can be raised as Github Issues. Our project roadmap is also available publicly.
pip3 -r requirements.txt
chmod +x plasma
./plasma