We will have a folder called output instead of model. Inside there are subfolders for each (important) training run. In those subfolders we will have the four trained baseline models, another subfolder for the videos and an .md file describing the whole setup (which data was used, batch size, hours of total training, GPU used… etc)
See: https://github.com/BASALT-2022-Karlsruhe/ka-basalt-2022/issues/16#issuecomment-1240347535