train occurs NAN with custom datasets

Zhongdao / Towards-Realtime-MOT

Joint Detection and Embedding for fast multi-object tracking

MIT License

2.38k stars 539 forks source link

train occurs NAN with custom datasets #188

Open 123-xin opened 4 years ago

123-xin commented 4 years ago

hi,thanks for your work! I have a problem of 'train'. When I train the model with my own datasets, It occurs LOSS is 'NAN'. eg: 2020-09-11 13:58:45 [INFO]: 0/29 7000/8353 nan nan nan nan 3.46 0.0786 Could you help me solve the problem? best regards!