The ML-based emulator for performing computation of aerosol optical properties should be implemented into OSLO_AERO. This emulator will consist of multiple smaller neural networks and needs the installation of Torch to be present as well as the Fortran-PyTorch binding. Since NorESM2.3 will only be supported for a short time period, the changes should be done in 2.5 directly so that we don't have to do the extra work needed to move it after the end of the support period.
The ML-based emulator for performing computation of aerosol optical properties should be implemented into OSLO_AERO. This emulator will consist of multiple smaller neural networks and needs the installation of Torch to be present as well as the Fortran-PyTorch binding. Since NorESM2.3 will only be supported for a short time period, the changes should be done in 2.5 directly so that we don't have to do the extra work needed to move it after the end of the support period.