Seungjun Nah

Hi @keys-zlc You can start training with a new `--save_dir` argument and load weight by specifying the location with `--pretrained` argument. https://github.com/SeungjunNah/DeepDeblur-PyTorch/blob/master/src/option.py#L52 For example, ```bash python main.py --save_dir NEW_DIRNAME --pretrained...

I tried the following command and it produces the test images and measures new PSNR and SSIM. ```bash python main.py --save_dir GOPRO_L1 --dataset GOPRO_Large --do_train false --do_test true --load_epoch 1000...

Same here. I also had to revert to 4.5.5.64

Check out `--save_dir` argument. https://github.com/SeungjunNah/DeepDeblur-PyTorch/blob/e92dc7c3337127887da617ffe819b49cb022fd81/src/option.py#L109

1. Are you launching many jobs from a single machine? Use different master ports per job. https://github.com/SeungjunNah/DeepDeblur-PyTorch/blob/master/src/option.py#L33 2. Please provide more details.

2. Please refer to [Usage examples](https://github.com/SeungjunNah/DeepDeblur-PyTorch/blob/master/src/option.py#L77) - Example commands ```bash # save all of the evaluation results python main.py --n_GPUs 1 --batch_size 8 --dataset GOPRO_Large --save_results all ``` Please refer...

Hi @davidvct, Yes, please follow the structure format of GOPRO_Large dataset. The txt file is not necessary.