JerzakLabs/LLMDebiasingBenchmark

LLM Debiasing Benchmark是一个用于评估大型语言模型注释去偏方法的基准数据集，包含情感分类、虚假信息源、性别偏见和毒性检测等任务的专家和LLM生成的注释。The LLM Debiasing Benchmark is a collection of datasets and annotations used to evaluate debiasing methods for LLM-based annotations in computational social science, covering tasks such as sentiment classification, misinformation source, gender bias, and toxicity detection.