KGATE : a Knowledge Graph Embedding Training Environment
KGATE is a modular PyTorch library for training knowledge-graph autoencoders with leakage-controlled preprocessing and reproducible benchmarks.
KGATE (Knowledge Graph Autoencoder Training Environment) is a modular Python library built on PyTorch Geometric and TorchKGE for knowledge-graph embedding. Users compose initializers, encoders, decoders, losses, negative samplers, and metrics, or plug in custom blocks. It includes leakage-controlled preprocessing, a built-in training pipeline, and reproducibility by design. Benchmarks against six KGE libraries show training time comparable to the fastest options while offering a broader feature set.
- Modular blocks for encoders, decoders, losses, samplers, and metrics
- Preprocessing is designed to control training-data leakage
- Training time matches the fastest of six compared KGE libraries
- Built on PyTorch Geometric and TorchKGE
From the original article
1 min read at arxiv.orgKnowledge graph embedding (KGE) models encode the entities and relations of a knowledge graph into a low-dimensional latent space, enabling tasks such as classification or link prediction. Most KGE models follow an autoencoder architecture, in which an encoder projects the knowledge graph into the latent space and a decoder reconstruct it.
By Benjamin Loire · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.