[Bug] konflict between ReduceOnPlateauLR and val_evaluator

[hamlet2019]

分支

main 分支 (mmpretrain 版本)

描述该错误

There is the following scheduler
dict(type='ReduceOnPlateauLR', monitor='loss', rule='less', factor=0.1, patience=5),
The monitoring is loss
But the evaluation is val_evaluator = dict(type='Accuracy', topk=(1,))
An error occurs during the evaluation

环境信息

mmpretrained

其他信息

No response