This project implements an algorithm based on Confident Learning (CL) to automatically detect various issues in machine learning datasets, thereby improving the quality of the datasets and enabling the training of better models. It supports data in image, text, and audio formats.