Research map / Large language models / Language, safety and interpretability
Generating QM1B with PySCF$_{\text{IPU}}$
Alexander Mathiasen, Hatem Helal, Kerstin Klaeser, Paul Balanca, Josef Dean, Carlo Luschi, Dominique Beaini, Andrew W Fitzgibbon, Dominic Masters
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 |
|---|---|---|---|
| pretrain on QM1B before fine-tuning on QM9 | SchNet 9M · QM9 (88k subset) | validation MAE (meV): 54.13 -> 30.2 | Improved |
| run DFT on IPUs instead of CPU PySCF | PySCF · DFT data generation, one POD16 | DFTs per second: 82 -> 228 | Improved |
| use float32 instead of float64 in libcint (ablation) | PySCF_IPU · libcint on IPU | speed: around 6x faster | Improved |
| use float32 DFT instead of float64 (ablation) | PySCF_IPU · HL gap vs PySCF float64 | MAE (meV): 0.2 | Improved |
| use STO-3G labels instead of 6-31G* labels for QM9 (ablation) | SchNet · QM9 training loss curves | loss | No clear effect |
| use RDKit ETKDG conformer positions instead of equilibrium positions for QM9 (ablation) | SchNet · QM9 training | MAE | Hurt |
| increase QM1B training subset size | SchNet 9M · QM1B validation | validation MAE | Improved |
| increase QM1B training size beyond about 100M samples | SchNet 9M · QM1B validation | validation MAE | No clear effect |
| exploit the 8x symmetry of the electron repulsion integrals | PySCF_IPU · DFT compute | FLOPs and memory: 8x reduction | Improved |
| recompute electron repulsion integrals every iteration instead of precomputing | PySCF_IPU · cycle count | cycles: 105M -> 395.7M | Hurt |
Closest papers by meaning
- High-order Equivariant Flow Matching for Density Functional Theory Hamiltonian Prediction (NeurIPS 2025 · similarity 0.86)
- Rethink the Role of Deep Learning towards Large-scale Quantum Systems (ICML 2025 · similarity 0.85)
- The Importance of Being Scalable: Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains (NeurIPS 2024 · similarity 0.85)
- Enhancing the Scalability and Applicability of Kohn-Sham Hamiltonians for Molecular Systems (ICLR 2025 · similarity 0.84)
- QCircuitBench: A Large-Scale Dataset for Benchmarking Quantum Algorithm Design (NeurIPS 2025 · similarity 0.84)
- From Density Matrices to Phase Transitions in Deep Learning: Spectral Early Warnings and Interpretability (NeurIPS 2026 · 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.
