Question about MIST encoder training configuration

[qvviko]

Thanks you for you research and for releasing the codebase for FRIGID!

I had a question about the MIST encoder used in your work. I could not find details in the repository or the paper about how the MIST model was trained (hyperparameters, command-line options, configurations).

Could you share the training configuration you used for the MIST encoder (e.g. the mist script parameters or the command used)?

Thanks in advance!