OGB-LSC Leaderboards


Steps to submit to leaderboards
  • Carefully read the rules.
  • Develop models and save test-dev prediction using the OGB Evaluator.
  • Submit your result via this page.
  • Wait at least one week until your test result shows up on our leaderboard. Please contact us if this is not the case.

Leaderboards policies
  • For each email address on each dataset, another leaderboard submission cannot be made within one week after the last submission. Our system will automatically reject such submissions. Please do not use multiple email addresses within the same team.
  • Extra information (hardware information, training/inference time, validation performance, technical report etc) is required for the OGB-LSC leaderboard submissions. Please check the submission page early to understand what is required.
  • Please do not submit a placeholder Github repository / technical report. You will be permanently disqualified if we find your submission is based on the placeholder.

Check out leaderboards

Package denotes the required package version for each dataset to be eligible for the leaderboards.


Leaderboard for MAG240M

Classification accuracy on the test-dev and validation sets. The higher, the better.

Package: >=1.3.2

Rank Method Ensemble Test-dev Accuracy Validation Accuracy Team Contact References #Params Hardware Date
1 R-UNIMP (KDDCup’21 1st) Yes 0.7539 0.7773 BD-PGL Yunsheng Shi (Baidu) Paper, Code N/A V100 (32GB) Mar 9, 2022
2 MDGNN (OGB-LSC’22 3rd) Yes 0.7535 0.7756 CogDL Yukuo Cen (Tsinghua / Zhipu AI) Paper, Code 1,297,305 1 NVIDIA A100 GPU (80GB memory) Nov 21, 2022
3 MPNN+BGRL (KDDCup’21 2nd) Yes 0.7507 0.7710 Academic Petar Velickovic (DeepMind) Paper, Code N/A 4 Google Cloud TPUv4 Mar 9, 2022
4 Cleora+EMDE (KDDCup’21 3rd) Yes 0.7457 N/A Synerise AI Jacek Dabrowski (Synerise) Paper, Code N/A Tesla V100 (16GB) Mar 9, 2022
5 MPLP+finetune (KDDCup’21 4th) Yes 0.7440 N/A Topology_mag Wencai Cao (OPPO Research) Paper, Code N/A V100 GPU (32GB memory) Mar 9, 2022
6 SGC+R-GAT+Finetune (KDDCup’21 5th) Yes 0.7377 0.7420 passages Kaiyuan Li (Nanjing University / BUPT) Paper, Code N/A GeForce RTX 3090 GPU (24GB) Mar 9, 2022
7 GNN180M (KDDCup’21 6th) Yes 0.7340 N/A DeeperBiggerBetter Guohao Li (KAUST / Intel) Paper, Code N/A 4 NVIDIA GeForce RTX 6000 (48G) Mar 9, 2022
8 R-GAT (NS) No 0.6931 0.7002 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 12,255,385 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
9 R-GraphSAGE (NS) No 0.6878 0.6986 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 12,234,905 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
10 GAT (NS) No 0.6671 0.6715 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 4,890,777 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
11 GraphSAGE (NS) No 0.6621 0.6679 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 4,884,633 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
12 MLP+C&S No 0.6605 0.6698 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 473,241 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
13 SIGN No 0.6603 0.6664 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 3,758,233 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
14 SGC No 0.6530 0.6582 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 736,921 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
15 LabelProp No 0.5638 0.5844 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 0 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
16 MLP No 0.5276 0.5267 OGB-LSC Matthias Fey (TU Dortmund) Paper, Code 473,241 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021


Leaderboard for WikiKG90Mv2

MRR on the test-dev and validation sets. The higher, the better.

Package: >=1.3.3

Rank Method Ensemble Test-dev MRR Validation MRR Team Contact References #Params Hardware Date
1 BESS + diverse KGE ensemble Yes 0.2539 0.2919 Graphcore Daniel Justus (Graphcore) Paper, Code 23,356,393,344 Graphcore Bow Pod16 Dec 1, 2022
2 DGLKE+RuleMining+ManualFeature Yes 0.2493 0.2916 DNAKG Xiaojun Ma (MSRA) Paper, Code 18,247,074,200 2x TitanV,800GB memory Nov 20, 2022
3 PIE-RM Yes 0.2115 0.2540 USTCWiki HongChao Gu (USTC / iflytek) Paper, Code 18,247,074,200 4 Telsla M40 Oct 26, 2022
4 TransE-ens Yes 0.1903 0.2931 CogDL Xiao Liu (Tsinghua / Zhipu.AI) Paper, Code 18,247,074,200 1 A100 Oct 31, 2022
5 TransE-Shallow-PIE No 0.1883 0.2342 Ant Group KG&NLP Linlin Chao (Ant Group) Paper, Code 18,247,074,200 Tesla P100 (16GB GPU) May 2, 2022
6 TransE-Concat No 0.1761 0.2060 OGB-LSC Hongyu Ren (Stanford) Paper, Code 18,246,707,000 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021
6 ComplEx-Concat No 0.1761 0.2048 OGB-LSC Hongyu Ren (Stanford) Paper, Code 18,246,707,000 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021
7 ComplEx No 0.1412 0.1816 GraphLARS Ling Yue (EE, Tsinghua / 4paradigm) Paper, Code 18,246,707,000 1*A100 Sep 1, 2022
8 ComplEx-MPNet No 0.0988 0.1258 OGB-LSC Hongyu Ren (Stanford) Paper, Code 307,600 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021
9 ComplEx-Shallow No 0.0985 0.1150 OGB-LSC Hongyu Ren (Stanford) Paper, Code 18,246,399,400 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021
10 TransE-MPNet No 0.0860 0.1128 OGB-LSC Hongyu Ren (Stanford) Paper, Code 307,600 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021
11 TransE-Shallow No 0.0824 0.1103 OGB-LSC Hongyu Ren (Stanford) Paper, Code 18,246,399,400 1 Quadro RTX 8000 (45GB GPU) Oct 12, 2021


Leaderboard for PCQM4Mv2

MAE on the test-dev and validation sets. The lower, the better.

Package: >=1.3.2

Rank Method Ensemble Test-dev MAE Validation MAE Team Contact References #Params Hardware Date
1 GPS++ Yes 0.0720 0.0778 GraphcoreValenceMILA Dominic Masters (Graphcore/Valence/MILA) Paper, Code 44,291,413 Graphcore BOW-POD16 Nov 18, 2022
2 MolNet_Ensemble Yes 0.0753 0.0797 polixir.ai zouxiaochuan (polixir.ai) Paper, Code 32,047,874 8 RTX3090 Nov 1, 2022
3 Global-ViSNet No 0.0766 0.0784 ViSNet Tong Wang (Microsoft Research AI4Science) Paper, Code 78,450,692 4 NVIDIA A100 GPUs Oct 26, 2022
4 Transformer-M No 0.0782 0.0772 FML Lab@PKU Shengjie Luo (Peking University) Paper, Code 68,957,249 4 NVIDIA Tesla A100 GPUs (40GB) Oct 5, 2022
5 GEM-2 No 0.0806 0.0793 PaddleHelix Donglong He (Baidu) Paper, Code 32,086,707 16 NVIDIA A100 Aug 11, 2022
6 Deep graph transformer Yes 0.0843 0.0891 NVIDIA-PCQM4Mv2 Jiwei Liu (NVIDIA) Paper, Code 63,600,000 1 NVIDIA V100 GPU 32 GB Oct 26, 2022
7 Deep graph transformer Yes 0.0844 0.0891 NVIDIA-PCQM4Mv2 Jiwei Liu (NVIDIA) Paper, Code 63,600,000 1 NVIDIA V100 GPU 32 GB Oct 19, 2022
8 Deep graph transformer Yes 0.0852 0.0891 NVIDIA-PCQM4Mv2 Jiwei Liu (NVIDIA) Paper, Code 63,600,000 1 NVIDIA V100 GPU 32 GB Oct 11, 2022
9 EGT No 0.0862 0.0857 EFT-AIRC Md Shamim Hussain (RPI / IBM) Paper, Code 89,326,465 8 Tesla V100 (32GB) Jun 24, 2022
9 GPS No 0.0862 0.0852 Mila Ladislav Rampasek (Mila / Universite de Montreal) Paper, Code 13,807,345 1 NVIDIA A100 (40GB) Nov 10, 2022
10 CoAtGIN-base No 0.0866 0.0859 xfcui@sdu Xuefeng Cui (Shandong University) Paper, Code 9,938,433 8 NVIDIA A40 Sep 30, 2022
11 EGT No 0.0872 0.0869 EGT-AIRC Md Shamim Hussain (RPI / IBM) Paper, Code 89,326,465 8 Tesla V100 (32GB) Jan 26, 2022
12 GRPE-Large No 0.0876 0.0867 WonWoo Wonpyo Park (SNU / Standigm) Paper, Code 118,300,000 8 A100-SXM4-40GB Aug 14, 2022
13 GraphSelfAttention No 0.0898 0.0890 WonWoo Wonpyo Park (Standigm) Paper, Code 46,199,041 4 A100 GPU Nov 15, 2021
14 Deep graph transformer Yes 0.0904 0.0891 NVIDIA-PCQM4Mv2 Jiwei Liu (NVIDIA) Paper, Code 63,600,000 1 NVIDIA V100 GPU 32 GB Oct 3, 2022
15 CoAtGIN-tiny No 0.0908 0.0901 xfcui@sdu Xuefeng Cui (Shandong University) Paper, Code 6,395,393 1 NVIDIA A40 Sep 20, 2022
16 TokenGT (Lap) No 0.0919 0.0910 vl-kaist Jinwoo Kim (KAIST) Paper, Code 48,492,289 NVIDIA RTX 3090 x 8 Aug 8, 2022
17 CoAtGIN-tiny No 0.0921 0.0916 xfcui@sdu Xuefeng Cui (Shandong University) Paper, Code 6,183,425 1 NVIDIA A40 Sep 8, 2022
18 CoAtGIN-tiny No 0.0935 0.0933 xfcui@sdu Xuefeng Cui (Shandong University) Paper, Code 5,460,225 NVIDIA A40 Aug 29, 2022
19 HFAGNN No 0.1010 0.1005 Zhuque Zhejianglab Can Xu (Zhejiang Lab) Paper, Code 3,949,959 1 NVIDIA TITAN V GPU(12GB memory) Oct 30, 2022
20 GIN-virtual No 0.1084 0.1083 OGB-LSC Weihua Hu (Stanford) Paper, Code 6,656,406 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
21 GCN-virtual No 0.1152 0.1153 OGB-LSC Weihua Hu (Stanford) Paper, Code 4,850,401 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
22 GIN No 0.1218 0.1195 OGB-LSC Weihua Hu (Stanford) Paper, Code 3,761,406 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
23 GraphGPS without PE (subset) No 0.1395 0.1334 GraphLARS Xu Wang (4Paradigm && EE Tsinghua) Paper, Code 6,155,089 1 RTX3090 (24GB GPU) Sep 2, 2022
24 GCN No 0.1398 0.1379 OGB-LSC Weihua Hu (Stanford) Paper, Code 1,955,401 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
25 MLP-Fingerprint No 0.1760 0.1753 OGB-LSC Weihua Hu (Stanford) Paper, Code 16,107,201 1 GeForce RTX 2080 (11GB GPU) Sep 8, 2021
26 MolNet_Ensemble Yes 0.9899 0.0847 polixir.ai xiaochuan zou (polixir.ai) Paper, Code 32,047,874 8 RTX3090 Oct 24, 2022