Research map / Graph learning / GNN architectures
R2G: A Multi-View Circuit Graph Benchmark Suite from RTL to GDSII
Zewei Zhou, Jiajun Zou, Jiajia Zhang, Ao Yang, Ruichao He, Haozheng Zhou, Ao Liu, Jiawei Liu, Leilei Jin, Shan Shen, Daying Sun
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 |
|---|---|---|---|
| increase decoder head from 1 to 3 layers (ablation) | GINE · R2G placement view (b) | R2: 0.8514 -> 0.9883 | Improved |
| increase decoder head from 1 to 3 layers (ablation) | ResGatedGCN · R2G placement view (b) | R2: -0.1689 -> 0.9889 | Improved |
| increase decoder head from 3 to 4 layers (ablation) | GINE · R2G placement view (b) | R2: 0.9883 -> 0.9962 | Improved |
| increase decoder head from 3 to 4 layers (ablation) | ResGatedGCN · R2G placement view (b) | R2: 0.9889 -> 0.9891 | No clear effect |
| increase decoder head from 2 to 3 layers (ablation) | GINE · R2G routing view (b) | R2: 0.8586 -> 0.9891 | Improved |
| use a 1-layer decoder head instead of 2 layers (ablation) | GINE and ResGatedGCN · R2G routing view (b) | R2: nan | Hurt |
| increase GNN depth from 4 to 6 layers (ablation) | GINE · R2G placement view (b) | R2: 0.8915 -> 0.8348 | Hurt |
| increase GNN depth from 3 to 6 layers (ablation) | ResGatedGCN · R2G routing view (b) | R2: 0.8762 -> 0.8566 | Hurt |
| use view (b) all-elements-as-nodes instead of view (f) | GNN baselines (avg test) · R2G placement | R2: 0.4617 -> 0.8577 | Improved |
| use view (b) instead of view (f) | GNN baselines (avg test) · R2G routing | R2: 0.2055 -> 0.8497 | Improved |
| use view (e) incidence edges instead of view (b) | GINE · R2G placement | R2: 0.8878 -> -0.0107 | Hurt |
| use view (e) instead of view (b) | GAT · R2G routing | R2: 0.8131 -> 0.8389 | Improved |
Closest papers by meaning
- DeepGate4: Efficient and Effective Representation Learning for Circuit Design at Scale (ICLR 2025 · similarity 0.86)
- Bridging Layout and RTL: Knowledge Distillation based Timing Prediction (ICML 2025 · similarity 0.85)
- AnalogGenie-Lite: Enhancing Scalability and Precision in Circuit Topology Discovery through Lightweight Graph Modeling (ICML 2025 · similarity 0.85)
- PreRoutGNN for Timing Prediction with Order Preserving Partition: Global Circuit Pre-training, Local Delay Learning and Attentional Cell Modeling (AAAI 2024 · similarity 0.84)
- NN-Former: Rethinking Graph Structure in Neural Architecture Representation (CVPR 2025 · similarity 0.84)
- PowerGraph: A power grid benchmark dataset for graph neural networks (NeurIPS 2024 · similarity 0.84)
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.
