davanstrien/aud-lfm2.5-350m-20260428

这是一个由大型语言模型（LLM）标注的数据集，由classify-and-augment工具生成。数据集使用LiquidAI/LFM2.5-350M模型进行标注，包含positive（正面）和negative（负面）两个情感标签。原始输入数据有180行，经过处理后输出195行数据。标签分布显示negative标签有124个（全部为真实数据），positive标签有71个（其中56个为真实数据，15个为合成数据）。合成数据审计显示，positive类需要44个样本，实际生成了58个候选样本，经过验证保留了44个，最终保留了15个合成样本，接受率为75.9%。

This is an LLM-annotated dataset produced by the classify-and-augment tool. The dataset uses the LiquidAI/LFM2.5-350M model for annotation and contains two sentiment labels: positive and negative. The original input consists of 180 rows, and the processed output contains 195 rows. The label distribution shows 124 negative labels (all real data) and 71 positive labels (56 real and 15 synthetic). The synthesis audit indicates that for the positive class, 44 samples were needed, 58 candidates were generated, 44 were validated, and 15 synthetic samples were ultimately kept, with an acceptance rate of 75.9%.