This blog series provides an at-a-glance roundup of recent activity within the LineaPy community, including new features, bug fixes, collaborator shout outs, github discussions or community gatherings.

INTRODUCING LINEAPY VERSION 0.1.5

We’ve been hard at work the last few weeks and are excited to introduce LineaPy 0.1.5 to everyone! This version includes a rethink of what a pipeline is to our users and as usual, includes minor bug fixes that aim to keep the experience a breeze for our users.

  • Pipelines: Pipelines are central to discussions around productionization, and in this release, we fixed a major hassle that shows up when working with experiments. Lineapy now supports a finer-level control/observability of the modularized code under the new graph refactor. The user’s code is now managed into “non-overlapping” functions, where duplicate code blocks are factored out to reduce redundant computations (which might be expensive to run). The resulting pipeline’s tasks can be controlled using different “flavors”. Check out this Github Discussion for more details on flavors and this discussion for code samples to see what your output could look like.
  • Parametrization: This release also introduces lineapy.get_function, a new LineaPy API for parameter refactoring. It extracts the code to create the list of targeted artifacts and packages the code as a Python function. It also allows us to parameterize any (user-assigned) variables within the code. Furthermore, for any artifact calculation (not just targeted artifacts) in the code, we can load an artifact value (any version) directly from the artifact store. Detailed discussions and samples can be found in this Github Discussion.

As always, we invite community members to leave their opinions on these discussions or on this Slack group and help us improve Lineapy as a product and help data science teams everywhere to deliver the most impact.