When I apply this model to the verse2020 dataset, I get stuck at the ninth epoch every time (it will directly terminate the prompt RuntimeError: DataLoader worker (pid 9063) is killed by signal: killed) When I change the higher performance GPU and CPU, adjust the learning rate and batch, etc., I still get stuck at the ninth epoch, showing that it takes ten hours
Your validation data is so big, you can only validate only a section of all validation data. And you also can modify the DDIM sample step from 10 to 2, which can also improve the speed of inference.
When I apply this model to the verse2020 dataset, I get stuck at the ninth epoch every time (it will directly terminate the prompt RuntimeError: DataLoader worker (pid 9063) is killed by signal: killed) When I change the higher performance GPU and CPU, adjust the learning rate and batch, etc., I still get stuck at the ninth epoch, showing that it takes ten hours![42fe82b1178e5512182987a1c8228e6](https://github.com/ge-xing/Diff-UNet/assets/93257077/a264f998-15cb-4294-8b19-3100edf42963)