Bert-Multi-Label-Text-Classification icon indicating copy to clipboard operation
Bert-Multi-Label-Text-Classification copied to clipboard
verified publisher icon lonePatient

关于分类阈值问题

Open wl915634785 opened this issue 4 years ago • 0 comments

如果阈值为0.5,我不太理解。因为样本是0和1，所以预测值也应该接近0和1，那么经过sigmod以后不都是应该大于0.5吗，是不是意味着某个样本将会属于所有类别

wl915634785 avatar Dec 31 '21 07:12 wl915634785