RuntimeError: CUDA out of memory. about penet_icra2021 HOT 5 CLOSED

Thanks for your interest! A single 8G-memory 1080/2080 is not sufficient for training the model (ENet). We suggest that you might reduce all channel numbers to half and according to our experience there's 20-25 RMSE loss compared to the original model.

from penet_icra2021.

Thank you very much! It worked

from penet_icra2021.

@JUGGHM Hi! Sorry for opening this issue again. I am also experiencing this error on a 3070Ti 8G machine. But I am not trying to train the model, I just want to test inference the model on my own dataset. How can I solve it?

from penet_icra2021.

@ZHANGZ1YUE Were you able to solve the issue ? I am having the same problem during inference.

from penet_icra2021.

@ZHANGZ1YUE Were you able to solve the issue ? I am having the same problem during inference.

@ArghyaChatterjee I don’t specifically remember but I think the issue was simply about gpu memory. The model is big and image is big, and 8G memory was not enough. I tried reduce the image size and should fixed.

from penet_icra2021.