September 2022 Summaries

cleanlab 2.1 is a significant advancement in open-source Data-Centric AI, extending its capabilities beyond classification with label errors to various new data-centric ML tasks such as analysis of multi-annotator data, out-of-distribution detection, and label error detection in token classification tasks. The release includes major new functionalities, improved performance, and reduced dependencies, making it a valuable tool for engineers and data scientists working on diverse applications. Additionally, Cleanlab Studio provides a no-code platform to efficiently fix issues in datasets, enabling users to train better ML models on cleaner data.