graphstorm.model.RelationalGCNEncoder

class graphstorm.model.RelationalGCNEncoder(g, h_dim, out_dim, num_bases=-1, num_hidden_layers=1, dropout=0, use_self_loop=True, last_layer_act=False, num_ffn_layers_in_gnn=0, norm=None)

Relational graph conv encoder.

The RelationalGCNEncoder employs several RelGraphConvLayer as its encoding mechanism. The RelationalGCNEncoder should be designated as the model’s encoder within Graphstorm.

Parameters

gDistGraph

The distributed graph object.

h_dimint

Hidden dimension

out_dimint

Output dimension

num_bases: int

Number of bases.

num_hidden_layersint

Number of hidden layers. Total GNN layers is equal to num_hidden_layers + 1. Default 1

dropoutfloat

Dropout. Default 0.

use_self_loopbool

Whether to add selfloop. Default True

last_layer_acttorch.function

Activation for the last layer. Default None

num_ffn_layers_in_gnn: int

Number of ngnn gnn layers between GNN layers

normstr, optional

Normalization Method. Default: None

Examples:

# Build model and do full-graph inference on RelationalGCNEncoder
from graphstorm import get_feat_size
from graphstorm.model.rgcn_encoder import RelationalGCNEncoder
from graphstorm.model.node_decoder import EntityClassifier
from graphstorm.model import GSgnnNodeModel, GSNodeEncoderInputLayer
from graphstorm.dataloading import GSgnnNodeTrainData
from graphstorm.model.gnn import do_full_graph_inference

np_data = GSgnnNodeTrainData(...)

model = GSgnnNodeModel(alpha_l2norm=0)
feat_size = get_feat_size(np_data.g, 'feat')
encoder = GSNodeEncoderInputLayer(g, feat_size, 4,
                                  dropout=0,
                                  use_node_embeddings=True)
model.set_node_input_encoder(encoder)

gnn_encoder = RelationalGCNEncoder(g, 4, 4,
                                   num_heads=2,
                                   num_hidden_layers=1,
                                   dropout=0,
                                   use_self_loop=True,
                                   norm=norm)
model.set_gnn_encoder(gnn_encoder)
model.set_decoder(EntityClassifier(model.gnn_encoder.out_dims, 3, False))

h = do_full_graph_inference(model, np_data)