SQuAD
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SQuAD contextual v1.1
23.2BLEU
24
Inj-SQuAD
0FPR (None)
24
SQuAD entity-level knowledge conflict (test)
9.7MR
24
SQuAD
0.68ASR
22
SQuAD (test)
87.6Top-100 Accuracy
22
SQuAD (test)
44.61BLEU-1
22
SQuAD 2 (test)
92.7AUROC
21
SQuAD
0.839F1 Score
21
SQuAD
74.2AUC Score
21
SQuAD
0.946Faithfulness
21
SQuAD v1.1 & v2.0
90.17SQuAD v1.1 F1
21
SQuAD v2.0 (val)
83.3F1
21
SQuAD
0.496BLEU-4
21
SQuAD 1.1
0.263METEOR
21
SQuAD
86.8F1 Score
21
SQuAD
0.22Loss
20
SQuAD
95.33Accuracy
20
SQuAD v2
32.6Accuracy
20
SQuAD v2.0
98AUROC
20
SQuAD (test)
74.8AUROC
20
SQuAD 50-passage (train)
49.6Accuracy
20
SQuAD 200-passage (val)
50.3QA Accuracy
20
SQuAD
84.36Accuracy
20
SQuAD Inj
1.11ASR (Naive)
18
SQuAD
96TPR (FPR=0.01)
18