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

LegalBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Domain-specific ReasoningLegalBench
Accuracy85.26
33
Legal ReasoningLegalBench Hearsay
Accuracy86.46
16
RetrievalLegalBench CorporateLobbying
nDCG@1093.56
12
RetrievalLegalBench RAG
Hit Rate@1996
11
Legal ReasoningLegalBench Learned Hands Courts
Accuracy75.5
10
Legal ReasoningLegalBench
Balanced Accuracy79.3
10
Question-Type Diversity AlignmentLegalbench Taxonomy
Jensen Shannon Divergence0.036
8
In-Context LearningLegalBench
Accuracy79.5
6
Legal ReasoningLegalBench CUAD Cardlytics Buffalo Wild Wings PF Hospitality 2023
Accuracy (Cardl)82.7
6
RetrievalLegalBench EN
nDCG@1063.42
5
GenerationLegalBench Rule-Application
Exact Match59
4
ClassificationLegalBench Interpretation
Accuracy69.7
4
Cross-lingual Question AnsweringLEGALBENCH RuleQA English (test)
ROUGE-120.25
3
TG taskLegalbench
Warranty Duration (CUAD)61
3
Showing 14 of 14 rows