LaRS training config/code

[kometa-triatlon]

Hi! In the LaRS paper (https://lojzezust.github.io/lars-dataset/) you mention WaSR-T model performance on the dataset.
However, I cannot find in this repo any hint of the training config and the data module utilized to read the LaRS data format for training. Something like wasr_t/data/lars.py. Would you be so kind to share it? I would like to train it on LaRS too.

I am also curious, did you evaluate the LaRS-trained model on MODS benchmark? Does it make sense from your point of view?