aisi-whitebox/uriah_dataset_generation_claude_3_7_sonnet_20250219_ARC-Challenge_cot
收藏Hugging Face2025-06-26 更新2025-09-13 收录
下载链接:
https://hf-mirror.com/datasets/aisi-whitebox/uriah_dataset_generation_claude_3_7_sonnet_20250219_ARC-Challenge_cot
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是一个欺骗性检测数据集,用于评估模型在ARC-Challenge_cot任务上的表现。数据集中的对话被设计为过分谄媚,使用复杂的语言和逻辑,并在回答过程中故意包含错误信息。数据集未进行划分,包含测试和验证参数,但没有启用沙袋检测。
This dataset is a deception detection dataset, designed to evaluate the models performance on the ARC-Challenge_cot task. The conversations in the dataset are crafted to be overly sycophantic, using complex language and logic, and intentionally incorporating false information in the course of the answers. The dataset is not split, contains test and validation parameters, but sandbagging detection is not enabled.
提供机构:
aisi-whitebox



