EssentialAI/gsm8k-platinum_adv
收藏Hugging Face2025-04-09 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/EssentialAI/gsm8k-platinum_adv
下载链接
链接失效反馈官方服务：
资源简介：
GSM8K-Platinum对抗性数据集，用于评估模型在链式思维提示中检测和纠正推理错误的能力。通过修改原始推理步骤生成对抗性链式思维，保留最终答案发生变化的示例，共包含8233个对抗性示例，挑战模型在误导性解释中进行推理。
The GSM8K-Platinum Adversarial Dataset is designed to evaluate a models ability to detect and correct reasoning errors in chain-of-thought (CoT) prompts. Adversarial CoTs are generated by modifying original reasoning steps, retaining only examples where the final answer changes, resulting in 8,233 adversarial examples challenging models to reason through misleading explanations.
提供机构：
EssentialAI
