search

meijieru / crnn.pytorch

Convolutional recurrent network in pytorch
MIT License
2.38k stars 658 forks source link

loss and accuracy are both 0 #210

Open qIanxIn726 opened 5 years ago

qIanxIn726 commented 5 years ago

I use the same lMDb data for train and validation,but the results seem nothing to do with ground truth,I am training for Chinese plate recognition by the way. here is part of my training log history.txt

YadongLau commented 4 years ago

Hi , Did you have solve it? I have got the same problem!

qIanxIn726 commented 4 years ago

i am sorry,i didn't

Hungry&Humble

---Original--- From: "lyd953621450"<notifications@github.com> Date: Thu, Sep 26, 2019 16:42 PM To: "meijieru/crnn.pytorch"<crnn.pytorch@noreply.github.com>; Cc: "Author"<author@noreply.github.com>;"qianxin"<1962522396@qq.com>; Subject: Re: [meijieru/crnn.pytorch] loss and accuracy are both 0 (#210)

Hi , Did you have solve it? I have got the same problem!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

gaoshangle commented 3 years ago

Hi , Did you have solve it? I have got the same problem! thanks @lyd953621450 @qIanxIn726

imneonizer commented 3 years ago

I am using this pytorch container: nvcr.io/nvidia/pytorch:20.12-py3 Had to make changes to make the code run but at the end this is what i see

Test loss: nan, accuray: 0.000000
[2/25][3000/4712] Loss: nan
zhangfengyo commented 3 years ago

did you solve it? i have also meet the same problem

imneonizer commented 3 years ago

I am using https://github.com/Holmeyoung/crnn-pytorch

It worked for me