SQuAD
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SQuAD MRQA few-shot
93.5F1 Score
5
SQuAD adversarial (test)
70.3Accuracy
5
SQuAD 547 passages 1.1 (test)
66.3RA
5
SQuAD 160 passages subset 1.1 (test)
79.6RA
5
SQuAD 20 passages subset 1.1 (test)
0.873RA
5
SQuAD AddSent 1.0 (adversarial)
78.5F1 Score
5
Adversarial SQuAD (AddSent)
75.5EM
5
SQuAD
20.9Exact Match
5
Public SQuAD-derived RAG
0Post-shift Accepted Trajectory Risk
4
SQUAD
52.17Clean Accuracy
4
SQuAD
12.1ASR
4
SQuAD
10.32Attack Success Rate (ASR)
4
SQuAD (held-out)
50.9Exact Match (EM)
4
SQuAD v2 (val)
83.72EM
4
SQuAD 2.0 (test)
29.03BLEU
4
SQuAD 17.5K
22.2BLEU
4
SQuAD (test dev)
89.7EM
4
SQuAD benign 1.0 (test)
93.8F1 Score
4
SQuAD in-domain
78.5MRR
4
translated SQuAD QH-PH 5.1 (test)
51.6Exact Match (EM)
4
translated SQuAD QE-PH 5.1 (test)
52.93Exact Match (EM)
4
SQuAD Del-Word (test)
51.7EM
3
SQuAD Del-Char (test)
54.1Exact Match (EM)
3
SQuAD
2.91Grammar
3
SQuAD
84.78TPR @ 1% FPR
3