This repocitory contains PyTorch implementation of R-TuckER model for knowledge graph link prediction task.
Proposed method utilizes model described in paper TuckER: Tensor Factorization for Knowledge Graph Completion, but applies the fact, that tensors of fixed tucker rank forms smooth Riemann manifold, and therefore riemann optimization toolkit (such as riemann gradient descent) is available. Detail description will be provided later, cause work still in progress.
MIT License