Jun-Yan Zhu

@tcwang0509

I didn't fully understand your question. Could you elaborate on that?

The program requires more than 2 GB GPU. It seems that your GPU has 2 GB, which is not sufficient to run the program.

If you use CycleGAN, it will require more than 2GB even with load_size 128.

I am not aware. You can also use GPU cloud service if you don't want to buy a new GPU.

Most of the methods tend to overfit the training set if you have limited data. You may consider applying data augmentation to combat model overfitting.

It may not be so related to the models you are using. You need to either experiment with different types of data augmentation or expand your dataset.

Haven't seen this before. Does `--continue_train` work for you? You can resume the model training from epoch 195.

It depends on many factors: the number of images in your dataset, the complexity of your input/output domains, the task itself, the image resolution, etc. 30 epochs might not be...

I am not sure if our current code supports visualization of 6-channel images. Two potential fixes: (1) you can try using the wandb visualization and see if they handle it...