LinkPredictWeightedTransEDecoder

class graphstorm.model.LinkPredictWeightedTransEDecoder(etypes, h_dim, gamma=12.0, norm='l2', edge_weight_fields=None)

Link prediction decoder with the score function of TransE with edge weight.

When computing loss, edge weights are used to adjust the loss.

Parameters

etypes: list of tuples: The canonical edge types of the graph in the format of [(src_ntype1, etype1, dst_ntype1), …]
h_dim: int: The input dimension size. It is the dimension for both source and destination node embeddings.
gamma: float: The gamma value for model weight initialization. Default: 12.
norm: str: L1 or L2 norm on the angular distance for TransE. Default: ‘l2’.
edge_weight_fields: dict of str: The edge feature field(s) storing the edge weights.

forward(g, h, e_h)

Forward function.

This computes the TransE score on every edge type.