search

xiaolai-sqlai / mobilenetv3

mobilenetv3 with pytorch，provide pre-train model
MIT License
1.6k stars 340 forks source link

loss nan #22

Open jlcai5 opened 4 years ago

jlcai5 commented 4 years ago

hi , my training loss is normal, but the validate loss is nan , Have you met anybody?

wplf commented 12 months ago

yes, i met this problem, too. Maybe it's the original learing rate is too big, and the model can't find the right weight space. More epochs will solve this problem.