credit#
CREDIT is an open software platform to train and deploy AI atmospheric prediction models.
How to use the documentation#
Documentation is available via docstrings provided with the code.
Available subpackages#
- datasets
Contains PyTorch Dataset classes for common Earth system data sources.
- ensemble
Methods for generating ensembles
- losses
Contains a mix of specialized loss functions for optimizing deterministic and ensemble models.
- metadata
Contains metadata definitions for use in inference.
- models
Defines model architectures.
- trainers
Defines trainers
Submodules#
- credit._version
- credit.animation
- credit.attend
- credit.boundary_padding
- credit.cli
- credit.credit_ptype
- credit.data
- credit.datasets
- credit.diffusion
- credit.diffusion_utils
- credit.distributed
- credit.domain_parallel
- credit.ensemble
- credit.forecast
- credit.gefs
- credit.grid
- credit.interp
- credit.losses
- credit.metadata
- credit.metrics
- credit.metrics_downscaling
- credit.mixed_precision
- credit.models
- credit.nwp
- credit.ocean
- credit.output
- credit.output_downscaling
- credit.parser
- credit.pbs
- credit.physics_constants
- credit.physics_core
- credit.pol_lapdiff_filt
- credit.postblock
- credit.preblock
- credit.regrid
- credit.replay_buffer
- credit.samplers
- credit.scheduler
- credit.seed
- credit.skebs
- credit.solar
- credit.trainers
- credit.transforms
- credit.transforms_downscaling
- credit.verification
- credit.visualization_tools
- credit.xr_sampler