xjtushujun / meta-weight-net

NeurIPS'19: Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting (Pytorch implementation for noisy labels).

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

关于baseline精度问题

zylprivate opened this issue · comments

请问baseline 的参数设置和其他方法的设置一样吗,为什么baseline在噪声40%的结果偏高呢,大约0.8左右

I got the same problem, could the author clarify? Much appreciated

I got the same problem, could the author clarify? Much appreciated

实验细节 在arxiv的支撑材料里,baseline 随着epoch 的增加会过拟合这些噪声。不过baseline 结合 early-stop 也能取得很好的效果。

感谢回答,我跑baseline一样的epoch数,貌似也没有出现overfit的现象