Transforming a Jupyter Notebook into a Reproducible Pipeline for ML Experiments

Discover how to transform Jupyter Notebook prototypes into reproducible pipelines for machine learning experiments in this 26-minute PyCon US talk. Learn why proper experiment tracking is crucial and how to achieve reproducibility using Git and DVC. Follow along as the speaker demonstrates breaking up a notebook into modules, creating a pipeline, running experiments, and comparing results using a text2image project with Stable Diffusion. Gain valuable insights on moving beyond basic notebook usage, especially beneficial for data scientists without extensive engineering backgrounds looking to enhance their workflow and experiment management skills.

Talks - Rob de Wit: Transforming a Jupyter Notebook into a reproducible pipeline for ML experiments