Research map / Architectures and efficiency / Attention and transformers
A recipe for scalable attention-based ML potentials: unlocking long-range accuracy with all-to-all node attention
Eric Qu, Brandon M. Wood, Aditi S. Krishnapriyan, Zachary Ward Ulissi
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 LAE angular encoding (ablation) | AllScAIP-md 85M · OMol25 4M Total | force MAE: 0.007603682161 -> 0.006889121209 | Improved |
| add ERoPE distance bias (ablation) | AllScAIP-md 85M · OMol25 4M Total | energy MAE: 0.001293416426 -> 0.001164967593 | Improved |
| add all-to-all node attention (ablation) | AllScAIP-md 85M · OMol25 4M Total | energy MAE: 0.001563449868 -> 0.001164967593 | Improved |
| add all-to-all node attention (ablation) | AllScAIP-md 85M · OMol25 102M Total | force MAE: 0.006148287564 -> 0.005096293718 | Improved |
| add LAE (ablation) | AllScAIP-md 85M · OMol25 102M Total | energy MAE: 0.0007254301528 -> 0.0006647585806 | No clear effect |
| add LAE and ERoPE (ablation) | AllScAIP-md 85M · OMol25 102M Total | energy MAE: 0.0006407190255 (without) vs 0.0006647585806 (with) | No clear effect |
| scale data from 4M to 102M | AllScAIP-md 85M · OMol25 Total | force MAE: 0.006889121209 -> 0.005096293718 | Improved |
| add rotation-equivariance by learning (train on OMol-102M vs random init) | AllScAIP-md · Random rotation force cosine similarity | cosine similarity: 0.6827 -> 0.9999 | Improved |
| replace direct-force with conservative fine-tune | AllScAIP-md · OMol25 4M Total | energy MAE: 0.90 (ft-cons., 4M) | Improved |
| use all-to-all node attention vs eSEN/UMA | AllScAIP-md-cons. · OMol25 distance scaling (LR) | energy error: about 90% reduction vs second best | Improved |
Closest papers by meaning
- MLIP Arena: Advancing Fairness and Transparency in Machine Learning Interatomic Potentials via an Open, Accessible Benchmark Platform (NeurIPS 2025 · similarity 0.91)
- The Importance of Being Scalable: Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains (NeurIPS 2024 · similarity 0.91)
- Physics-Informed Weakly Supervised Learning For Interatomic Potentials (ICML 2025 · similarity 0.89)
- Colour me shocked: Exact Molecular Hessians from MLIPs in O(N) time using sparse differentiation! (NeurIPS 2026 · similarity 0.89)
- FlashTP: Fused, Sparsity-Aware Tensor Product for Machine Learning Interatomic Potentials (ICML 2025 · similarity 0.88)
- Stein Kernelized Molecular Dynamics for Active Learning of Interatomic Potentials (NeurIPS 2026 · similarity 0.88)
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.
