ACE
Ai2 Climate Emulator (ACE) is a family of fast ML models that simulate global atmospheric variability over time scales ranging from hours to centuries
Updated • 51 • 2Note Coupled atmosphere-ocean model trained on a pre-industrial control run of NOAA-GFDL's CM4 model.
allenai/ACE2-ERA5
Updated • 210 • 13Note An updated version of the ACE model trained on the ERA5 reanalysis dataset with learned forced responses to historical SST.
allenai/ACE2-EAMv3
Updated • 33 • 2Note An updated ACE model trained on the E3SMv3 model with learned forced responses to historical SST.
allenai/ACE2-SOM
Updated • 17 • 1Note An updated ACE model that is coupled to a slab ocean model and trained to emulate the sensitivity of the climate to carbon dioxide.
allenai/ACE-climSST-EAMv2
Updated • 13Note An ACE model trained on a repeating climatological SST dataset from the E3SMv2 model.
allenai/ACE-climSST
Updated • 21 • 3Note An ACE model trained on a repeating climatological SST dataset from the FV3GFS model.
allenai/ACE2-ERA5-training-artifacts
Updated • 168Note Additional training artifacts related to the ACE2-ERA5 model. For nearly all purposes, we recommend using the checkpoint in the ACE2-ERA5 repository.
allenai/ACE2-ERA5-sample-output
Viewer • Updated • 1 • 81Note Sample output from 2 years of simulation from the ACE2-ERA5 model.