Model for ehull¶
- Description: This is a benchmark to evaluate how accurately an AI model can predict the energy above the convex hull (Ehull) using the JARVIS-DFT (dft_3d) dataset. The dataset contains different types of chemical formula and atomic structures. Here we use mean absolute error (MAE) to compare models with respect to DFT accuracy.
Reference(s): https://github.com/divelab/AIRS/tree/main/OpenMat/PotNet, https://www.nature.com/articles/s41524-020-00440-1, https://www.nature.com/articles/s41524-023-01012-9;https://hackingmaterials.lbl.gov/matminer, https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.145301, https://doi.org/10.48550/arXiv.2305.11842, https://www.nature.com/articles/s41524-021-00650-1, https://github.com/aimat-lab/gcnn_keras
Model benchmarks
| Model name | Dataset | MAE | Team name | Dataset size | Date submitted | Notes |
|---|---|---|---|---|---|---|
| kgcnn_coNGN | dft_3d | 0.0485 | kgcnn | 55364 | 05-06-2023 | CSV, JSON, run.sh, Info |
| kgcnn_cgcnn | dft_3d | 0.059 | kgcnn | 55364 | 09-26-2023 | CSV, JSON, run.sh, Info |
| matminer_rf | dft_3d | 0.112 | UofT | 55364 | 05-22-2023 | CSV, JSON, run.sh, Info |
| kgcnn_dimenetPP | dft_3d | 0.3685 | kgcnn | 55364 | 05-06-2023 | CSV, JSON, run.sh, Info |
| alignn_model | dft_3d | 0.0763 | ALIGNN | 55364 | 01-14-2023 | CSV, JSON, run.sh, Info |
| potnet | dft_3d | 0.0522 | DIVE@TAMU | 55364 | 06-02-2023 | CSV, JSON, run.sh, Info |
| kgcnn_megnet | dft_3d | 0.0569 | kgcnn | 55364 | 05-06-2023 | CSV, JSON, run.sh, Info |
| cgcnn_model | dft_3d | 0.173 | CGCNN | 55364 | 01-14-2023 | CSV, JSON, run.sh, Info |
| matminer_xgboost | dft_3d | 0.0601 | UofT | 55364 | 05-22-2023 | CSV, JSON, run.sh, Info |
| kgcnn_coGN | dft_3d | 0.0466 | kgcnn | 55364 | 05-06-2023 | CSV, JSON, run.sh, Info |
| kgcnn_schnet | dft_3d | 0.1014 | kgcnn | 55364 | 09-26-2023 | CSV, JSON, run.sh, Info |