reduce zero label, resize and inference

[danjac94]

Hi, I have some questions regarding the workflow. First, I'd like to know if there is a method available to apply reduce_zero_label. Additionally, I'm curious if using seg_resize during finetuning is expected to work correctly across all model architectures (using a custom dataset for example 256 image size). Finally, I would appreciate guidance on the process for visualizing or performing inference using a saved checkpoint with the validation or test split.