SciQ
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SciQ
82AUROC
18
SciQ
96Accuracy
17
SciQ
87.9Normalized Accuracy
17
SciQ-WMDP-Bio (test)
68.7Concept Accuracy
16
SciQ
94.1Accuracy
16
SciQ (test)
1.5ECE
15
SciQ
68Accuracy
15
SciQ MC
0.137Mean Per-Step Regret
15
SciQ Abstract
0.135Mean per-step regret
15
Sciq (test)
24.3Precision@1
15
SciQ
89.1Accuracy
14
SciQ
65.3PRR (AlignScore)
14
SciQ
11.95Perplexity
13
SciQ
66.5Score
12
SciQ
90.2Accuracy
12
SciQ
25.5P@1
12
SciQ (D_eval)
71.4Accuracy
12
SciQ
72.1Normalized Accuracy
11
SciQ
98.53Score
10
SciQ zero-shot
83.9Accuracy (SciQ zero-shot)
8
SciQ
0.506PRR
8
SciQ Out-of-domain
64.4PRR (AlignScore)
8
SciQ
0ECE' Gain
8
SCIQ Generalization
90.4Accuracy
8
SciQ standard (test)
90.2Accuracy
8