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

SciFact

Benchmarks

Task NameDataset NameSOTA ResultTrend
Information RetrievalSciFact (test)
NDCG@100.906
65
Information RetrievalSciFact
nDCG@1077.77
36
Information RetrievalSciFact BEIR (test)
nDCG@1076.6
31
Scientific Fact VerificationSciFact
Macro F183.03
25
Information Retrievalscifact
Recall@10096.7
19
Information RetrievalSciFact
Faithfulness67
18
Information RetrievalSciFact BEIR
NDCG@1085.4
17
Information RetrievalSciFact
nDCG@100.77
16
Information RetrievalScifact
nDCG82
15
Fact-checkingSCIFact
Balanced Acc90.3
15
RerankingSciFact
nDCG76.4
12
Logical RetrievalSciFact BEIR v1 (test)
nDCG@100.64
12
Claim VerificationSCIFACT
Accuracy94.32
12
RetrievalSciFact-G
R@1035.1
10
Sentence-Level Confidence PredictionSciFact
AUROC0.544
10
Information RetrievalSciFact
NDCG@1072.07
7
RetrievalSciFact
nDCG@10.56
6
Document RetrievalSciFact
nDCG@544
6
RetrievalSciFact EN
nDCG@1032.35
6
Scientific Claim VerificationSciFact
Accuracy40.5
6
Adversarial AttackSciFact
Contriever Score29.08
6
Information RetrievalSciFact
Accuracy75.14
6
Information RetrievalSciFact BEIR
nDCG0.708
5
Scientific Claim VerificationSciFact (test)
Precision (NE)93
4
Document rerankingSciFact
NDCG@1080.15
4
Showing 25 of 33 rows