search

jiangsutx / SRN-Deblur

Repository for Scale-recurrent Network for Deep Image Deblurring
http://www.xtao.website/projects/srndeblur/srndeblur_cvpr18.pdf
MIT License
709 stars 184 forks source link

run time and loss #76

Open noranali opened 2 years ago

noranali commented 2 years ago

thank you for your code , what is the reason of that the loss is different at every time i restart the run time and run the train ? can you help me please

jiangsutx commented 2 years ago

The training process include random process, e.g. random data sequence order, random initialization etc. If you want deterministic process, you need to fix random seed for python random module, numpy random module and tensorflow random seed.

noranali commented 2 years ago

thank you for your replay i have read your paper and i found that the code converge at 4000 epochs but colab will be disconnected at 180 epochs . i want to use the final weight as initial . can you help me please ?