AutoEval Done Right: Using Synthetic Data for Model Evaluation
University of California, Berkeley · UC Berkeley · Weizmann Institute of Science · University of California at Berkeley
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
The evaluation of machine learning models using human-labeled validation data can be expensive and time-consuming. AI-labeled synthetic data can be used to decrease the number of human annotations required for this purpose in a process called autoevaluation. We suggest efficient and statistically principled algorithms for this purpose that improve sample efficiency while remaining unbiased.