Class for working with multiple HDF5 datasets

Inputs: 7 (Temp, Pres, and Gasesous concentrations) as inputs, SW takes lesser gases than LW Outputs: 224 (g-points SW rayleigh cross section

This is a template class, that should be inherited by any neural net emulator model. Methods that need to be implemented by your concrete NN

---------------------------------------------------------------------------------------------------------- A DataModule implements 5 key meth

Duplicate global features at all layers

Flattens a dict with array's as values into a 1D vector (or 2D with batch dimension)

Inputs: 18 (Temp, Pres, and Gasesous concentrations) as inputs Outputs: 256 (g-points LW absorption cross section)

Inputs: 18 (Temp, Pres, and Gasesous concentrations) as inputs Outputs: 256 (g-points LW emission)

graph_net_level_nodes: gn_input_dict_renamer_level_nodes

graph_net_layer_nodes: gn_input_dict_renamer_layer_nodes

Padding all var types to have same #features

Module to predict (with one or more MLPs) desired output based on a 1D hidden representation. Can be used to: - readout a hidden vect

c - spatial dimension f - feature dimension b - batch dimension

Inputs: 7 (Temp, Pres, and Gasesous concentrations) as inputs, SW takes lesser gases than LW Outputs: 224 (g-points SW absorption cross secti

Make wandb log in run.summary the best achieved monitored val_metric as opposed to the last

Upload checkpoints to wandb as a file, at the end of run.

Upload checkpoints to wandb as an artifact, at the end of run.

Upload all code files to wandb as an artifact, at the beginning of the run.

Make wandb watch model at the beginning of the run.