scale-safety-research/synth_docs_honly_and_claude_anti_reward_hacking
收藏Hugging Face2025-02-13 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/scale-safety-research/synth_docs_honly_and_claude_anti_reward_hacking
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含两个字段：提示（prompt）和完成（completion），均为字符串类型。提示可能是一个问题或者一个任务指示，而完成是对于提示的响应或者解决方案。训练集包含50000个示例，数据集总大小为193141041字节。数据集适用于文本生成或对话系统等NLP任务。
The dataset includes two fields: prompt and completion, both of which are string types. The prompt may be a question or a task instruction, while the completion is the response or solution to the prompt. The training set contains 50,000 examples, and the total size of the dataset is 193141041 bytes. The dataset is suitable for NLP tasks such as text generation or dialogue systems.
提供机构：
scale-safety-research
