BriefGPT.xyz
May, 2022
BERT 之后:其他木偶对语言的理解
Life after BERT: What do Other Muppets Understand about Language?
HTML
PDF
Vladislav Lialin, Kevin Zhao, Namrata Shivagunde, Anna Rumshisky
TL;DR
利用oLMpics基准和心理语言学探测数据集,我们对包括T5、BART和ALBERT在内的 29个模型进行了多样化的分析。结果表明,这些模型都无法以零样本的方式解决组成性问题,并且全局模型决策也不能预测模型的语言能力。
Abstract
Existing
pre-trained transformer analysis
works usually focus only on one or two model families at a time, overlooking the variability of the architecture and pre-training objectives. In our work, we utilize the
olmpics
→