Research map / Graph learning / Graph self-supervised learning
Graph Tokenization for Bridging Graphs and Transformers
Zeyuan Guo, Enmao Diao, Cheng Yang, Chuan Shi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add BPE on top of Feuler serialization (ablation) | GT+GTE · ogbg-molhiv | ROC-AUC: 81.3 -> 87.4 | Improved |
| add BPE on top of Feuler serialization (ablation) | GT+GTE · ZINC | MAE: 0.171 -> 0.131 | Improved |
| add BPE on top of Feuler serialization (ablation) | GT+GTE · QM9 | MAE: 0.088 -> 0.071 | Improved |
| add BPE on top of Feuler serialization (ablation) | GT+GTE · COIL-DEL | accuracy: 85.6 -> 88.0 | Improved |
| add BPE to BFS serialization (ablation) | GT+GTE · ogbg-molhiv | ROC-AUC: 81.2 -> 72.3 | Hurt |
| add BPE to BFS serialization (ablation) | GT+GTE · ZINC | MAE: 0.696 -> 0.453 | Improved |
| use Feuler (frequency-guided Eulerian) instead of unguided Eulerian (ablation) | GT+GTE · ogbg-molhiv | ROC-AUC: 84.5 -> 87.4 | Improved |
| use reversible Eulerian-type serialization instead of BFS (ablation) | GT+GTE · ZINC | MAE: 0.453 -> 0.131 | Improved |
| use frequency-guided CPP (FCPP) instead of CPP (ablation) | GT+GTE · ogbg-molhiv | ROC-AUC: 86.9 -> 86.4 | No clear effect |
| scale backbone from BERT-small to GTE | GraphTokenizer · ogbg-molhiv | ROC-AUC: 82.6 -> 87.4 | Improved |
| GT+GTE vs best GNN baseline | GraphTokenizer · COIL-DEL | accuracy: 91.5 -> 89.6 | Hurt |
| GT+BERT vs GCN+ | GraphTokenizer · ZINC | MAE: 0.116 -> 0.241 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Exphormer — Exphormer: Sparse Transformers for Graphs (ICML 2023) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Flatten Graphs as Sequences: Transformers are Scalable Graph Generators (NeurIPS 2025 · similarity 0.88)
- From Sequence to Structure: Uncovering Substructure Reasoning in Transformers (NeurIPS 2025 · similarity 0.87)
- Tokenphormer: Structure-aware Multi-token Graph Transformer for Node Classification (AAAI 2025 · similarity 0.87)
- Multi-View Empowered Structural Graph Wordification for Language Models (AAAI 2025 · similarity 0.86)
- Talk like a Graph: Encoding Graphs for Large Language Models (ICLR 2024 · similarity 0.86)
- VCR-Graphormer: A Mini-batch Graph Transformer via Virtual Connections (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
