All TorsionNet206 Results

TorsionNet206 is a high-accuracy molecular benchmark that evaluates the accuracy of predicting and ranking of dihedral energy profiles. The benchmark contains 206 subsets of 24 conformers each computed at the CCSD(T)/def2-TZVP level of theory. Mean absolute error (MAE) and root mean square error (RMSE) are shown in kcal/mol. See all TorsionNet206 results.

Name
UMA Small (OMol task)	0.14	0.17	0.99	0.98
OMol25's eSEN Conserving Small	0.14	0.17	0.99	0.98
ωB97M-D3BJ/vDZP	0.15	0.18	0.99	0.98
Egret-1	0.20	0.24	0.99	0.98
Egret-1t	0.23	0.29	0.99	0.98
Egret-1e	0.22	0.28	0.99	0.98
r²SCAN-D4/vDZP	0.45	0.56	0.98	0.97
B97-D3BJ/vDZP	0.41	0.51	0.98	0.97
B97-3c	0.35	0.45	0.98	0.97
MACE-ANI-CC	0.23	0.29	0.97	0.97	59
r²SCAN-3c	0.42	0.54	0.97	0.97
AIMNet2 (ωB97M-D3, old)	0.39	0.48	0.95	0.94
AIMNet2 (ωB97M-D3, new)	0.39	0.48	0.95	0.94
B3LYP-D3BJ/6-31G(d)	0.57	0.71	0.95	0.94
eSEN-OAM	0.69	0.85	0.90	0.89
Prescient's StrainRelief MACE	0.61	0.77	0.89	0.90
GFN2-xTB	0.73	0.91	0.85	0.85
Orb-v3 (Conservative Inf. OMat)	0.97	1.20	0.83	0.83
GFN-FF	1.24	1.50	0.82	0.84	159
Orb-v3 (Conservative 20 OMat)	1.16	1.42	0.80	0.82
Orb-v2	1.15	1.43	0.78	0.81
OMat24 eqV2-L	1.48	1.84	0.77	0.81
MACE-MP-0b2(Large)-D3BJ	1.12	1.39	0.76	0.76
Orb-v3 (Direct Inf. OMat)	1.82	2.31	0.75	0.76
MACE-MP-0b2(Large)	1.15	1.43	0.74	0.75
Orb-d3-v2	1.43	1.73	0.74	0.76
Sage	2.44	2.96	0.72	0.79	190
Orb-v3 (Direct 20 OMat)	1.60	2.01	0.72	0.74
MACE-MP-0b2(Medium)	1.28	1.59	0.72	0.72
MACE-MP-0b2(Small)	1.23	1.52	0.71	0.74
MACE-MP-0(Large)-D3BJ	1.39	1.71	0.70	0.71
MACE-MP-0b(Medium)	1.28	1.59	0.70	0.71
MACE-MP-0(Large)	1.41	1.73	0.69	0.70
MACE-MP-0b(Small)	1.30	1.63	0.65	0.70
MACE-MP-0(Small)	1.74	2.13	0.64	0.64
MACE-MP-0(Medium)	1.61	1.99	0.62	0.62
SO3LR	2.24	2.77	0.55	0.59