daiquocnguyen/Graph-Transformer
UGformer is a graph transformer model that applies transformer self-attention to learn representations from graph-structured data in supervised and unsupervised settings.
Velocity · 7d
+0.3
★ / day
Trend
→steady
star history
The repository provides PyTorch and TensorFlow implementations of UGformer (Universal Graph Transformer Self-Attention Networks), published at WWW 2022. It leverages transformer self-attention mechanisms to learn node and graph-level embeddings for tasks like graph classification, node classification, and text classification. The model uses variant architectures that apply self-attention over graph structures to capture relational information.