How to fine tune SegFormer on a custom dataset

[MariaMel98]

Hello. I'm trying to fine tune a segformer model on my custom dataset. I have the images and masks as they should be for segformer in my dataset folder. I follow this tutoria lhttps://github.com/open-mmlab/mmsegmentation/blob/v1.2.2/demo/MMSegmentation_Tutorial.ipynb which is for a pspnet model. So far i have:

from mmengine import Config config_file = 'configs/segformer/segformer_mit-b0_8xb1-160k_cityscapes-1024x1024.py' cfg = Config.fromfile(config_file)

I'm working on a single GPU. What attributes should i change to cfg in order to be able to do the training? Is there a tutorial for this?