ML Data Version Control and Reproducibility at Scale

Explore data version control and reproducibility techniques for large-scale machine learning in this 38-minute talk by Einat Orr from Treeverse. Learn how to overcome challenges in ML data management, including reproducibility constraints and inefficient data transfer. Discover open-source tools for versioning data locally and best practices for working with data in the cloud without copying it. Gain insights into training models at scale using an OSS stack including Langchain, TensorFlow, PyTorch, and Keras. Acquire practical methods to enhance data management for developing and iterating on ML models, specifically tailored for modern computer vision research.