Framework reduces bias when training models on flawed human decisions

Original: Mitigating Label Bias with Interpretable Rubric Embeddings

Source: arxiv.org ↗