JingyunLiang / SwinIR

SwinIR: Image Restoration Using Swin Transformer (official repository)

Home Page:https://arxiv.org/abs/2108.10257

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Should Real-World Gan Training be main_train_gain.py?

Nodice opened this issue · comments

@JingyunLiang is there a typo on the Swin training instructions for when you switch from PSNR to GAN?

# 003 Real-World Image SR (middle size)
python -m torch.distributed.launch --nproc_per_node=8 --master_port=1234 main_train_psnr.py --opt options/swinir/train_swinir_sr_realworld_psnr.json  --dist True
# before training gan, put the PSNR-oriented model into superresolution/swinir_sr_realworld_x4_gan/models/
python -m torch.distributed.launch --nproc_per_node=8 --master_port=1234 **main_train_psnr.py** --opt options/swinir/train_swinir_sr_realworld_gan.json  --dist True

python -m torch.distributed.launch --nproc_per_node=8 --master_port=1234 main_train_psnr.py --opt options/swinir/train_swinir_sr_realworld_gan.json --dist True

Should that be main_train_gan.py? It does exist in KAIR and it is training a Gan.

Is there any update on this?

Would be great to have more information about how to finetune from the L model.

The only difference between the two files is that 'main_train_gan.py' looks also for a checkpoint of the discriminator at this line.
So both files will work if you start from iter 0 and from pretrained G and E checkpoints, but only 'main_train_gan.py' will work correctly if you stop the GAN training and want to restart it again.

ezoic increase your site revenue