Ericwang/nemotron-nano2-safety-distill-gptoss
收藏资源简介:
这是一个使用Nemotron Nano 2安全配方和GPT-OSS-20B及GPT-OSS-120B作为教师模型生成的安全聚焦蒸馏数据集。数据集包含来自Aegis AI Content Safety Dataset v2.0、Gretel Safety Alignment Dataset v1、HarmfulTasks Dataset和RedTeam-2K Dataset的四个子集。每个数据点包括ID、UUID、prompt、reasoning_20b、response_20b、reasoning_120b、response_120b和metadata字段,其中metadata字段提供了额外的上下文信息。
This is a safety-focused distilled dataset created using the Nemotron Nano 2 safety recipe with GPT-OSS-20B and GPT-OSS-120B as teacher models. The dataset includes four subsets from Aegis AI Content Safety Dataset v2.0, Gretel Safety Alignment Dataset v1, HarmfulTasks Dataset, and RedTeam-2K Dataset. Each data point includes fields such as ID, UUID, prompt, reasoning_20b, response_20b, reasoning_120b, response_120b, and metadata, which provides additional contextual information.



