EquFlash

Version: v2025.06.23 Added: 2025-06-23 Published: 2025-06-23 28.7M parameters
Repo Paper Files PR Checkpoint
Predictions
DiscoveryGeometry OptimizationκSRME
Leaderboard ranks CPS #8 /41Discovery F1 #8 /52Geo Opt RMSD #7 /42Phonons κSRME #8 /41

Discovery: energy and convex hull diagnostics

Missing preds: 0

Per-element convex hull distance errors

1 H 0.12
2 He Helium
3 Li 0.02
4 Be 0.02
5 B 0.05
6 C 0.05
7 N 0.07
8 O 0.10
9 F 0.10
10 Ne Neon
11 Na 0.02
12 Mg 0.03
13 Al 0.04
14 Si 0.05
15 P 0.04
16 S 0.06
17 Cl 0.08
18 Ar Argon
19 K 0.03
20 Ca 0.03
21 Sc 0.02
22 Ti 0.03
23 V 0.06
24 Cr 0.08
25 Mn 0.10
26 Fe 0.09
27 Co 0.04
28 Ni 0.03
29 Cu 0.03
30 Zn 0.03
31 Ga 0.04
32 Ge 0.04
33 As 0.04
34 Se 0.07
35 Br 0.07
36 Kr Krypton
37 Rb 0.03
38 Sr 0.03
39 Y 0.03
40 Zr 0.03
41 Nb 0.04
42 Mo 0.04
43 Tc 0.03
44 Ru 0.04
45 Rh 0.04
46 Pd 0.04
47 Ag 0.03
48 Cd 0.02
49 In 0.04
50 Sn 0.04
51 Sb 0.05
52 Te 0.10
53 I 0.05
54 Xe 0.02
55 Cs 0.03
56 Ba 0.03
57 La 0.03
58 Ce 0.03
59 Pr 0.03
60 Nd 0.02
61 Pm 0.03
62 Sm 0.03
63 Eu 0.06
64 Gd 0.04
65 Tb 0.03
66 Dy 0.03
67 Ho 0.03
68 Er 0.03
69 Tm 0.03
70 Yb 0.04
71 Lu 0.03
72 Hf 0.03
73 Ta 0.06
74 W 0.04
75 Re 0.03
76 Os 0.04
77 Ir 0.05
78 Pt 0.05
79 Au 0.05
80 Hg 0.03
81 Tl 0.03
82 Pb 0.05
83 Bi 0.03
84 Po Polonium
85 At Astatine
86 Rn Radon
87 Fr Francium
88 Ra Radium
89 Ac 0.03
90 Th 0.03
91 Pa 0.04
92 U 0.05
93 Np 0.09
94 Pu 0.19
95 Am Americium
96 Cm Curium
97 Bk Berkelium
98 Cf Californium
99 Es Einsteinium
100 Fm Fermium
101 Md Mendelevium
102 No Nobelium
103 Lr Lawrencium
104 Rf Rutherfordium
105 Db Dubnium
106 Sg Seaborgium
107 Bh Bohrium
108 Hs Hassium
109 Mt Meitnerium
110 Ds Darmstadtium
111 Rg Roentgenium
112 Cn Copernicum
113 Nh Nihonium
114 Fl Flerovium
115 Mc Moscovium
116 Lv Livermorium
117 Ts Tennessine
118 Og Oganesson
57-71 La-Lu Lanthanides
89-103 Ac-Lr Actinides

ML vs DFT Lattice Thermal Conductivity

Model Authors

  1. Hyuntae Cho  Materials AI Lab at Samsung Electronics  
  2. Saerom Choi  Materials AI Lab at Samsung Electronics  
  3. Heejae Kim  Materials AI Lab at Samsung Electronics  
  4. Jaehee Jang  Materials AI Lab at Samsung Electronics  
  5. Gunhee Kim  Materials AI Lab at Samsung Electronics  
  6. Heesun Lee  Materials AI Lab at Samsung Electronics  
  7. Hyunwoo Lee  Materials AI Lab at Samsung Electronics  
  8. Yongdeok Kim  Materials AI Lab at Samsung Electronics  

Model Info

  • Version v2025.06.23
  • Role Interatomic potential
  • Architecture gnn
  • Targets EFSG
  • Openness CSOD
  • Discovery Train Task S2EFS
  • Discovery Test Task IS2RE-SR

Training Set

MPtrj: 1.58M structures from 146k materials

OMat24: 101M structures from 3.23M materials

Subsampled Alexandria: 10.4M structures from 3.23M materials

description

EquFlash is an E(3)-equivariant model based on the SevenNet-0 architecture, with tensor products accelerated by FlashTP. FlashTP achieves up to 41.6× and 60.8× kernel speedups over e3nn and NVIDIA cuEquivariance, respectively, while reducing memory usage by 6×. Leveraging these gains, we scaled EquFlash to a larger capacity than the original SevenNet-0.

training

EquFlash, a scaled-up model derived from SevenNet-0 and accelerated with FlashTP, was pretrained on OMat24 and finetuned on MPtrj and sAlex.

Hyperparams

  • evaluation: {"max_force":0.02,"max_steps":500,"ase_optimizer":"FIRE","cell_filter":"FrechetCellFilter","kappa":{"protocol":"phonondb-v1","displacement_distance":0.03,"relaxation_mode":"two-stage","max_atoms_per_batch":128}}
  • architecture: {"graph_construction_radius":6}

Dependencies