search

sanghyun-son / EDSR-PyTorch

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)
MIT License
2.43k stars 669 forks source link

About resuming the model #175

Open wxxxxxxh opened 5 years ago

wxxxxxxh commented 5 years ago

Hi, thank you for sharing your code! When I resume the training process at a certain epoch or the latest epoch, I find that the model will begin from the epoch 1 rather than our setting. Can you provide some solutions about this error? Thanks very much.

sanghyun-son commented 5 years ago

Hello.

Would you provide the training script used for resuming your training?

I know there are some minor bugs in resuming, but I think we can fix this issue by using an appropriate script.

Thank you!

wxxxxxxh commented 5 years ago

Hi Thanks for your reply. Below is the script which I used to resume my training: nohup python -u main.py --scale 4 --save temp_model_vgg --load temp_model_vgg --resume 220 --model WDSR_B --loss 0.9*L1+0.1*VGG22 --ext img --batch_size 16 --test_every 1000 --dir_data ../../youku/images/ --data_train YOUKU --data_test YOUKU --n_resblocks 32 --n_feats 256 --n_GPUs 4 --save_models > youku_L1VGG.log 2>&1 & Thank you!