Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

HE

Benchmarks

Task NameDataset NameSOTA ResultTrend
ClassificationHE
Accuracy39.3
38
Code Correctness EvaluationHE-JA
F1 Score72.9
25
Code Correctness EvaluationHE-CPP
F1 Score66.3
25
ClassificationHE 50% corrupted features noise
Accuracy39.3
21
Feature SelectionHE random extraneous features
Precision1
17
Feature SelectionHE with random extraneous features
ROC AUC100
17
ClassificationHE 75% corrupted features noise
Accuracy38.6
10
Tabular Data ClusteringHE
ARI0.1563
8
Constituency Parsinghe Hebrew
LF Score92.99
7
Dependency Parsinghe Hebrew
LAS91.34
7
Code GenerationHE
HE Score77.44
4
ClusteringHe
Accuracy85.3
3
Showing 12 of 12 rows