model_name: Nequip-OAM-XL # required (this must match the model's label which is the 3rd arg in the matbench_discovery.preds.Model enum) model_key: nequip-OAM-XL-0.1 # this should match the name of the YAML file and determines the URL /models/ on which details of the model are displayed on the website model_version: '0.1' date_added: '2025-11-30' date_published: '2025-11-30' authors: - name: Seán R. Kavanagh affiliation: Department of Chemistry, University of Cambridge email: sk2045@cam.ac.uk orcid: https://orcid.org/0000-0003-4577-9647 url: https://sam-lab.net corresponding: true - name: Chuin Wei Tan affiliation: MIR Group, Harvard University - name: Albert Musaelian affiliation: MIR Group, Harvard University & Mirian Technologies - name: William C. Witt affiliation: MIR Group, Harvard University - name: Gabriel de Miranda Nascimento affiliation: MIR Group, Harvard University & MIT - name: Ulrik Unneberg affiliation: MIR Group, Harvard University & MIT - name: Marc L. Descoteaux affiliation: MIR Group, Harvard University - name: Boris Kozinsky affiliation: MIR Group, Harvard University trained_by: - name: Seán R. Kavanagh affiliation: Department of Chemistry, University of Cambridge email: sk2045@cam.ac.uk orcid: https://orcid.org/0000-0003-4577-9647 url: https://sam-lab.net corresponding: true repo: https://github.com/mir-group/nequip url: https://nequip.readthedocs.io/en/latest/ doi: https://doi.org/10.5281/zenodo.16980200 paper: https://arxiv.org/abs/2504.16068 # To be updated... checkpoint_url: https://www.nequip.net/models/mir-group/NequIP-OAM-XL:0.1 pr_url: https://github.com/janosh/matbench-discovery/pull/298 openness: OSOD # see `Open` enum in matbench_discovery/enums.py train_task: S2EFS # see `Task` enum in matbench_discovery/enums.py test_task: IS2RE-SR # see `Task` enum in matbench_discovery/enums.py targets: EFS_G # see `Targets` enum in matbench_discovery/enums.py model_type: UIP # see `ModelType` enum in matbench_discovery/enums.py model_params: 32_100_000 trained_for_benchmark: true n_estimators: 1 license: code: MIT code_url: https://github.com/mir-group/nequip/blob/main/LICENSE checkpoint: CC-BY-4.0 checkpoint_url: https://creativecommons.org/licenses/by/4.0/legalcode hyperparams: max_force: 0.005 max_steps: 500 ase_optimizer: GOQN # faster than FIRE with same results; see SI of https://doi.org/10.1088/2515-7655/ade916 cell_filter: FrechetCellFilter optimizer: AdamW weight_decay: 1e-8 graph_construction_radius: 6.0 sph_harmonics_l_max: 4 n_layers: 6 n_features: 320 (l=0 scalars), 96 (l=1 vectors), 64 (l=2 tensors), 32 (l=3,4 tensors) parity: false zbl_potential: true type_embed_num_features: 32 polynomial_cutoff: 5 n_radial_bessel_basis: 8 loss: Huber - delta=0.01 for energy, delta=0.1 for stress, stratified delta (0.01, 0.007, 0.004, 0.001) for force loss_weights: energy: 1.0 force: 5.0 stress: 0.1 batch_size: 640 # 4 (gpus) * 160 (batch per gpu) = 640 (total batch size) initial_learning_rate: 0.005 gradient_clip_val: 0.25 learning_rate_schedule: ReduceLROnPlateau - factor=0.1, patience=100, min_lr=1e-6 epochs: 30 # actual epochs run; max_epochs=100 in Trainer config max_neighbors: .inf training_cost: # ~9.8h per epoch, 30 epochs plus fine-tune Nvidia H100 GPUs: { amount: 8, hours: 300, cost: 2400 } requirements: torch: '2' nequip: '>=0.7.0' # >=0.14.0 recommended training_set: [OMat24, sAlex, MPtrj] notes: Description: Extra-large NequIP foundation potential; see for details and for model/training infrastructure. Steps: | Training performed by: (1) pre-training on OMat24; (2) fine-tuning on MPtrj+sAlex, with a reduced learning rate (1e-4), energy-loss-upweighting (1:1:0.01 instead of 1:5:0.01) and `StochasticWeightAveraging` (SWA). metrics: phonons: kappa_103: pred_file: models/nequip/nequip-OAM-XL-0.1/2025-11-30-kappa-103-FIRE-dist=0.03-fmax=0.0001-symprec=1e-05.json.gz pred_file_url: https://figshare.com/files/60007844 κ_SRME: 0.1252 κ_SRE: 0.0537 geo_opt: pred_file: models/nequip/nequip-OAM-XL-0.1/2025-11-30-wbm-IS2RE-GOQN.jsonl.gz pred_file_url: https://figshare.com/files/60007799 struct_col: nequip_structure # same for NequIP/Allegro symprec=1e-5: rmsd: 0.063 # unitless n_sym_ops_mae: 2.0821 # unitless symmetry_decrease: 0.0975 # fraction symmetry_match: 0.6834 # fraction symmetry_increase: 0.2117 # fraction n_structures: 256963 # count analysis_file: models/nequip/nequip-OAM-XL-0.1/2025-11-30-wbm-IS2RE-GOQN-symprec=1e-5-moyo=0.7.3.csv.gz analysis_file_url: https://figshare.com/files/60007817 symprec=1e-2: rmsd: 0.063 # unitless n_sym_ops_mae: 1.6986 # unitless symmetry_decrease: 0.0253 # fraction symmetry_match: 0.8223 # fraction symmetry_increase: 0.1437 # fraction n_structures: 256963 # count analysis_file: models/nequip/nequip-OAM-XL-0.1/2025-11-30-wbm-IS2RE-GOQN-symprec=1e-2-moyo=0.7.3.csv.gz analysis_file_url: https://figshare.com/files/60007835 discovery: pred_file: models/nequip/nequip-OAM-XL-0.1/2025-11-30-wbm-IS2RE.csv.gz pred_file_url: https://figshare.com/files/60007793 pred_col: e_form_per_atom_nequip # same for NequIP/Allegro full_test_set: F1: 0.885 # fraction DAF: 5.098 # dimensionless Precision: 0.875 # fraction Recall: 0.895 # fraction Accuracy: 0.96 # fraction TPR: 0.895 # fraction FPR: 0.027 # fraction TNR: 0.973 # fraction FNR: 0.105 # fraction TP: 39450.0 # count FP: 5647.0 # count TN: 207224.0 # count FN: 4642.0 # count MAE: 0.02 # eV/atom RMSE: 0.066 # eV/atom R2: 0.866 # dimensionless missing_preds: 2 # count unique_prototypes: F1: 0.906 # fraction DAF: 5.869 # dimensionless Precision: 0.897 # fraction Recall: 0.915 # fraction Accuracy: 0.971 # fraction TPR: 0.915 # fraction FPR: 0.019 # fraction TNR: 0.981 # fraction FNR: 0.085 # fraction TP: 30545.0 # count FP: 3500.0 # count TN: 178614.0 # count FN: 2829.0 # count MAE: 0.02 # eV/atom RMSE: 0.066 # eV/atom R2: 0.872 # dimensionless missing_preds: 0 # count most_stable_10k: F1: 0.985 # fraction DAF: 6.342 # dimensionless Precision: 0.97 # fraction Recall: 1.0 # fraction Accuracy: 0.97 # fraction TPR: 1.0 # fraction FPR: 1.0 # fraction TNR: 0.0 # fraction FNR: 0.0 # fraction TP: 9695.0 # count FP: 305.0 # count TN: 0.0 # count FN: 0.0 # count MAE: 0.019 # eV/atom RMSE: 0.075 # eV/atom R2: 0.878 # dimensionless missing_preds: 0 # count