core.models.escaip.modules.input_block

Classes

InputBlock	Wrapper of InputLayer for adding normalization
InputLayer	Base class for Graph Neural Network layers.

Module Contents

class core.models.escaip.modules.input_block.InputBlock(global_cfg: fairchem.core.models.escaip.configs.GlobalConfigs, molecular_graph_cfg: fairchem.core.models.escaip.configs.MolecularGraphConfigs, gnn_cfg: fairchem.core.models.escaip.configs.GraphNeuralNetworksConfigs, reg_cfg: fairchem.core.models.escaip.configs.RegularizationConfigs)

Bases: torch.nn.Module

Wrapper of InputLayer for adding normalization

backbone_dtype

input_layer

norm_node

norm_edge

forward(inputs: fairchem.core.models.escaip.custom_types.GraphAttentionData)

class core.models.escaip.modules.input_block.InputLayer(global_cfg: fairchem.core.models.escaip.configs.GlobalConfigs, molecular_graph_cfg: fairchem.core.models.escaip.configs.MolecularGraphConfigs, gnn_cfg: fairchem.core.models.escaip.configs.GraphNeuralNetworksConfigs, reg_cfg: fairchem.core.models.escaip.configs.RegularizationConfigs)

Bases: fairchem.core.models.escaip.modules.base_block.BaseGraphNeuralNetworkLayer

Base class for Graph Neural Network layers. Used in InputLayer and EfficientGraphAttention.

backbone_dtype

edge_attr_linear

edge_attr_norm

edge_ffn

forward(inputs: fairchem.core.models.escaip.custom_types.GraphAttentionData)