Editing exp_dir and using weights from checkpoint files

[ga84]

Hello Simon! Tank you for sharing this code, it helps me with my research on semantic segmentation!

I have a fairly stupid issue: I would like to use the pertained model and I downloaded the weights from the zenodo.org. However, after running the --write samples function, I get this error in conda prompt:
"AttributeError: module 'model.pretrained_weights' has no attribute file It comes to the problem of extracting weights, i.e. I guess I don't know how to edit exp_dir path and use weights from ckpt files. Would anyone please help me understand how to edit this path and use the data?

I tried extracting the ckpt data with tf.Session but I encountered other issues and I assume there's simpler solution to this?

Thank you!