Research map / Architectures and efficiency / Attention and transformers
Ramba: Selective State-Space Models for Relational Deep Learning
Yiming Liu, Chunyu Wei, Haozhe Lin, Fengjun Xiao, Junqi Zhang, Yunhai Wang, Yueguo Chen
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 Mamba backbone to sparse attention only model (ablation) | Ramba · RelBench zero-shot regression | R2 (%): 17.5 -> 25.2 | Improved |
| add Mamba when sparse attention is disabled (ablation) | Ramba · RelBench zero-shot classification | AUROC (%): 50.0 -> 68.5 | Improved |
| add Mamba when both attention masks are enabled (ablation) | Ramba · RelBench zero-shot classification | AUROC (%): 72.8 -> 72.8 | No clear effect |
| add sparse attention masks (intra-entity and relational) on top of Mamba (ablation) | Ramba · RelBench zero-shot classification | AUROC (%): 68.5 -> 72.8 | Improved |
| use schema-based gating instead of no gating (ablation) | Ramba · RelBench zero-shot regression | R2 (%): +2.7% R2 | Improved |
| shuffle column names in the gate (ablation) | Ramba gating · RelBench zero-shot | AUROC/R2 | Hurt |
| remove column names entirely (ablation) | Ramba · RelBench zero-shot classification | AUROC (%): 72.5 -> 70.9 | Hurt |
| use column-wise serialization instead of row-wise or random (ablation) | Ramba · RelBench | retained performance: 100% vs row 88%, random 85% | Improved |
| shuffle within columns (ablation) | Ramba serialization · RelBench | retained performance: 95% retained | No clear effect |
| use width-limited BFS with FK prioritization instead of DFS (ablation) | Ramba sampling · RelBench regression | R2 (%): 16.8 -> 25.2 | Improved |
| use width-limited BFS with FK priority instead of full BFS (ablation) | Ramba sampling · RelBench classification | AUROC (%): 70.3 -> 72.8 | Improved |
| increase context length from 512 to 4096 (ablation) | Ramba · RelBench regression | R2 (%): 24.0 -> 26.0 | Improved |
Closest papers by meaning
- Position: Relational Deep Learning - Graph Representation Learning on Relational Databases (ICML 2024 · similarity 0.85)
- RelGNN: Composite Message Passing for Relational Deep Learning (ICML 2025 · similarity 0.85)
- Joint Relational Database Generation via Graph-Conditional Diffusion Models (NeurIPS 2025 · similarity 0.85)
- Griffin: Towards a Graph-Centric Relational Database Foundation Model (ICML 2025 · similarity 0.84)
- Relational Learning in Pre-Trained Models: A Theory from Hypergraph Recovery Perspective (ICML 2024 · similarity 0.83)
- RelAgent: LLM Agents as Data Scientists for Relational Learning (NeurIPS 2026 · similarity 0.83)
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.
