how can I set the value of batch size and test_every?

sanghyun-son / EDSR-PyTorch

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)

MIT License

2.42k stars 668 forks source link

how can I set the value of batch size and test_every? #294

Closed Senwang98 closed 3 years ago

Senwang98 commented 3 years ago

Hi, I feel confused about batch size * test_every, because this setteing will increase train time per epoch. For example, if I use DIV2K to train (total 800 HR pics and 800 LR pics), how can I set batch size and test_every so that RCAN model can use the whole trainset just one time per epoch? Thanks for your reply!

songyonger commented 3 years ago

Hi,I have the same problem,did you solve it?

Senwang98 commented 3 years ago

@songyonger, You'd better set batch_size = 16(this value unchange). If you want to use more pictures from DIV2K dataset, you can set --data_test 0-900 and change test_every=1125.