HyperAI초신경

Link Property Prediction On Ogbl Wikikg2

평가 지표

Ext. data
Number of params
Test MRR
Validation MRR

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름
Ext. data
Number of params
Test MRR
Validation MRR
Paper TitleRepository
InterHT+ (256dim)No1480007380.7257 ± 0.00180.7370 ± 0.0022InterHT: Knowledge Graph Embeddings by Interaction between Head and Tail Entities-
ComplEx-N3-RP (50dim)-2501674000.63640.6594Relation Prediction as an Auxiliary Training Objective for Improving Multi-Relational Graph Representations
ComplEx (50dim)No2501139000.3804 ± 0.00220.3534 ± 0.0052Complex Embeddings for Simple Link Prediction
RotatE (50dim)No2500871500.2530 ± 0.00340.2250 ± 0.0035RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space
ComplEx (250dim)No12505695000.4027 ± 0.00270.3759 ± 0.0016Complex Embeddings for Simple Link Prediction
tripleRENo5007633370.5794 ± 0.00200.6045 ± 0.0024--
DistMult (500dim)No12505695000.3729 ± 0.00450.3506 ± 0.0042Embedding Entities and Relations for Learning and Inference in Knowledge Bases
Rot-ProNo10006696020.5602 ± 0.00160.5740 ± 0.0008Rot-Pro: Modeling Transitivity by Projection in Knowledge Graph Embedding
StarGraph + TripleRENo867621460.7201 ± 0.00110.7288 ± 0.0008StarGraph: Knowledge Representation Learning based on Incomplete Two-hop Subgraph
InterHT+No1563327700.7293 ± 0.00180.7391 ± 0.0023InterHT: Knowledge Graph Embeddings by Interaction between Head and Tail Entities-
DistMult (100dim)No2501139000.3447 ± 0.00820.3150 ± 0.0088Embedding Entities and Relations for Learning and Inference in Knowledge Bases
AutoWeirdNo192154020.7353 ± 0.00060.7362 ± 0.0006AutoWeird: Weird Translational Scoring Function Identified by Random Search-
InterHTNo192154020.6779 ± 0.00180.6893 ± 0.0015InterHT: Knowledge Graph Embeddings by Interaction between Head and Tail Entities-
CompoundE3DNo7506627000.7006 ± 0.00110.7175 ± 0.0015Knowledge Graph Embedding with 3D Compound Geometric Transformations
TransE (500dim)No12505695000.4256 ± 0.00300.4272 ± 0.0030--
AutoSFNo5002278000.5458 ± 0.00520.5510 ± 0.0063AutoSF: Searching Scoring Functions for Knowledge Graph Embedding
PairRE (200dim)No5003348000.5208 ± 0.00270.5423 ± 0.0020PairRE: Knowledge Graph Embeddings via Paired Relation Vectors
TransERRNo5004418020.6359 ± 0.00200.6518 ± 0.0012TransERR: Translation-based Knowledge Graph Embedding via Efficient Relation Rotation
ComplEx-RP (50dim)No2501674000.6392 ± 0.00450.6561 ± 0.0070--
NodePiece + AutoSFNo68606020.5703 ± 0.00350.5806 ± 0.0047NodePiece: Compositional and Parameter-Efficient Representations of Large Knowledge Graphs
0 of 30 row(s) selected.