[NeurIPS 2024] 🧼🔎 A holistic self-supervised data cleaning strategy to detect irrelevant samples, near duplicates and label errors.
