Graph-JEPA

Reported on 3 benchmarks across 3 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Graphs2 results

Graph RegressiononZINC
MAE· 2023-09-27
0.434
best: 0.051 (ESA + rings + NodeRWSE + EdgeRWSE)
Graph-level Representation Learning with Joint-Embedding Predictive Architectures arXiv:2309.16014
Graph ClassificationonREDDIT-B
Accuracy· 2023-09-27
56.73
best: 93.15 (CRaWl)
Graph-level Representation Learning with Joint-Embedding Predictive Architectures arXiv:2309.16014

Methodology1 result

ClassificationonREDDIT-B
Accuracy· 2023-09-27
56.73
best: 93.15 (CRaWl)
Graph-level Representation Learning with Joint-Embedding Predictive Architectures arXiv:2309.16014