Papers12:26Chinese contentDataOrchestra:逐示例编排预训练数据处理的框架这篇论文提出了DataOrchestra,能针对每个数据样本定制清洗流程,在多个模型和基准上都有提升,还节省算力。#DataOrchestra#large language models#data cleaning#pretrainingaarXiv cs.AI@Zhen Huang et al. (4)SourceRead laterReadWorth followingUsefulFollow DataOrchestra