model_name: eqV2 M model_key: eqV2-m-omat-salex-mp model_version: v2024.10.18 date_added: '2024-10-18' date_published: '2024-10-18' authors: - name: Luis Barroso-Luque affiliation: FAIR at Meta email: lbluque@meta.com orcid: https://orcid.org/0000-0002-6453-9545 github: https://github.com/lbluque corresponding: true - name: Muhammed Shuaibi affiliation: FAIR at Meta - name: Xiang Fu affiliation: FAIR at Meta - name: Brandon M. Wood affiliation: FAIR at Meta - name: Misko Dzamba affiliation: FAIR at Meta - name: Meng Gao affiliation: FAIR at Meta - name: Ammar Rizvi affiliation: FAIR at Meta - name: C.~Lawrence Zitnick affiliation: FAIR at Meta - name: Zachary W. Ulissi affiliation: FAIR at Meta email: zulissi@meta.com orcid: https://orcid.org/0000-0002-9401-4918 corresponding: true repo: https://github.com/FAIR-Chem/fairchem doi: https://doi.org/10.48550/arXiv.2410.12771 paper: https://arxiv.org/abs/2410.12771 url: https://huggingface.co/facebook/OMAT24#model-checkpoints pypi: https://pypi.org/project/fairchem-core pr_url: https://github.com/janosh/matbench-discovery/pull/146 # checkpoint URL copied from https://huggingface.co/fairchem/OMAT24#model-checkpoints checkpoint_url: https://huggingface.co/fairchem/OMAT24/blob/main/eqV2_86M_omat_mp_salex.pt license: code: MIT code_url: https://github.com/FAIR-Chem/fairchem/blob/aa160789e1/LICENSE.md checkpoint: Meta Research checkpoint_url: https://huggingface.co/facebook/OMAT24/blob/main/LICENSE requirements: fairchem-core: 1.2.1 openness: OSOD trained_for_benchmark: true train_task: S2EFS test_task: IS2RE-SR targets: EFS_D model_type: UIP model_params: 86_589_068 n_estimators: 1 # removed sAlex from this list since it would be double counting materials in the "Training Size" # metrics table column since OMat24 is a derivative of Alexandria training_set: [OMat24, MPtrj] training_cost: missing hyperparams: max_force: 0.02 max_steps: 500 ase_optimizer: FIRE cell_filter: FrechetCellFilter loss: MAE loss_weights: energy: 20 forces: 10 stress: 1 optimizer: AdamW learning_rate_schedule: Cosine warmup_epochs: 0.1 warmup_factor: 0.2 max_learning_rate: 0.0002 min_learning_rate_factor: 0.01 grad_clip_threshold: 100 ema_decay: 0.999 weight_decay: 0.001 dropout_rate: 0.1 stochastic_depth: 0.1 batch_size: 256 epochs: 16 graph_construction_radius: 12.0 # Å max_neighbors: 20 # see table 7 in arXiv:2410.12771 notes: Description: | EquiformerV2 is an equivariant transformer that uses graph attention, attention re-normalization, and separable S^2 activations and layer normalization. Training: | Training was done by fine-tuning a model pretrained for 2 epochs on the OMat24 dataset. metrics: phonons: kappa_103: κ_SRME: 1.7707 pred_file: models/eqV2/eqV2-m-omat-salex-mp/2024-11-09-kappa-103-FIRE-dist=0.01-fmax=1e-4-symprec=1e-5.json.gz pred_file_url: https://figshare.com/files/52134893 κ_SRE: 1.4827 geo_opt: pred_file: models/eqV2/eqV2-m-omat-salex-mp/2024-10-18-wbm-geo-opt.jsonl.gz pred_file_url: https://figshare.com/files/53153471 struct_col: eqV2-86M-omat-salex-mp_structure symprec=1e-5: rmsd: 0.0691 # unitless n_sym_ops_mae: 9.7872 # unitless symmetry_decrease: 0.852 # fraction symmetry_match: 0.1422 # fraction symmetry_increase: 0.0045 # fraction n_structures: 256963 # count analysis_file: models/eqV2/eqV2-m-omat-salex-mp/2024-10-18-wbm-geo-opt-symprec=1e-5-moyo=0.4.2.csv.gz analysis_file_url: https://figshare.com/files/53504543 symprec=1e-2: rmsd: 0.0691 # unitless n_sym_ops_mae: 1.9989 # unitless symmetry_decrease: 0.1118 # fraction symmetry_match: 0.7725 # fraction symmetry_increase: 0.1052 # fraction n_structures: 256963 # count analysis_file: models/eqV2/eqV2-m-omat-salex-mp/2024-10-18-wbm-geo-opt-symprec=1e-2-moyo=0.4.2.csv.gz analysis_file_url: https://figshare.com/files/53504549 discovery: pred_file: models/eqV2/eqV2-m-omat-salex-mp/2024-10-18-wbm-IS2RE.csv.gz pred_file_url: https://figshare.com/files/52057571 pred_col: e_form_per_atom_eqV2-86M-omat-mp-salex full_test_set: F1: 0.896 # fraction DAF: 5.243 # dimensionless Precision: 0.9 # fraction Recall: 0.893 # fraction Accuracy: 0.965 # fraction TPR: 0.893 # fraction FPR: 0.021 # fraction TNR: 0.979 # fraction FNR: 0.107 # fraction TP: 39379.0 # count FP: 4393.0 # count TN: 208478.0 # count FN: 4713.0 # count MAE: 0.02 # eV/atom RMSE: 0.071 # eV/atom R2: 0.842 # dimensionless missing_preds: 2 # count most_stable_10k: F1: 0.988 # fraction DAF: 6.382 # dimensionless Precision: 0.976 # fraction Recall: 1.0 # fraction Accuracy: 0.976 # fraction TPR: 1.0 # fraction FPR: 1.0 # fraction TNR: 0.0 # fraction FNR: 0.0 # fraction TP: 9756.0 # count FP: 244.0 # count TN: 0.0 # count FN: 0.0 # count MAE: 0.015 # eV/atom RMSE: 0.066 # eV/atom R2: 0.904 # dimensionless missing_preds: 0 # count unique_prototypes: F1: 0.917 # fraction DAF: 6.047 # dimensionless Precision: 0.924 # fraction Recall: 0.91 # fraction Accuracy: 0.975 # fraction TPR: 0.91 # fraction FPR: 0.014 # fraction TNR: 0.986 # fraction FNR: 0.09 # fraction TP: 30372.0 # count FP: 2481.0 # count TN: 179633.0 # count FN: 3002.0 # count MAE: 0.02 # eV/atom RMSE: 0.072 # eV/atom R2: 0.848 # dimensionless missing_preds: 0 # count