davanstrien/otb-augmented-lfm1.2b
收藏Hugging Face2026-04-28 更新2026-05-03 收录
下载链接：
https://hf-mirror.com/datasets/davanstrien/otb-augmented-lfm1.2b
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是由LLM（LiquidAI/LFM2.5-1.2B-Instruct模型）标注的分类和增强数据集，包含200条输入行和200条输出行，标签分为jim_crow和no_jim_crow两类。jim_crow标签有22条真实数据，no_jim_crow标签有178条真实数据。合成审计显示，jim_crow类需要28条，生成了160条，但仅验证了1条，最终保留了0条，接受率为0.6%。
LLM-annotated dataset produced by classify-and-augment. The dataset contains 200 input rows and 200 output rows, with labels jim_crow and no_jim_crow. The label distribution shows 22 jim_crow (all real) and 178 no_jim_crow (all real). The synthesis audit indicates that for jim_crow, 28 were needed, 160 were generated, 1 was validated, 0 were kept, with an acceptance rate of 0.6%.
提供机构：
davanstrien
