pengzhiliang / Conformer

Official code for Conformer: Local Features Coupling Global Representations for Visual Recognition
Apache License 2.0
531 stars 87 forks source link

Loss is nan, stopping training #33

Open sanersbug opened 1 year ago

sanersbug commented 1 year ago

Why it usually break when in training, and it show this error.Any one meet the same question , how to solve this problem

Salvatore-Love commented 1 year ago

Hello, have you solved it now? I have also encountered this problem

2252033991 commented 1 year ago

Did you solve it, please

Salvatore-Love commented 1 year ago

My previous experience was to reduce the learning rate, but this could potentially result in a loss of accuracy.