Understanding and Mitigating the Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks

Open in new window