Papers11:11Chinese contentSABRE:可扩展自动化的VLM压力测试基准构建流程这论文给了一个能自动造VLM压力测试的框架,比现有基准更能暴露模型短板,做评测的可以看看。#SABRE#VLM#stress testing#benchmarksaarXiv cs.AI@Zixuan Lan et al. (4)SourceRead laterReadWorth followingUsefulFollow SABRE