nanguoshun / LSR

Pytorch Implementation of our ACL 2020 Paper "Reasoning with Latent Structure Refinement for Document-Level Relation Extraction"
127 stars 21 forks source link

一些关于bert的疑问 #16

Closed zhanghongya0727 closed 4 years ago

zhanghongya0727 commented 4 years ago

请问您这个模型使用bert是直接按照bert baseline的那篇论文进行参考的么,学习率改为1e-05,编码器改为bert,其余不变么?谢谢

nanguoshun commented 4 years ago

@zhanghongya0727 谢谢关注!是的,lr为1e-05,encoder是BERT,其他的不变。