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

Science Question Answering on GPQA main (test)

40.85Exact Match Accuracy

Min-k

Updated 4d ago

Evaluation Results

Method	Links
Min-k 2026.04		40.85
Min-k 2026.04		40.85
Min-k 2026.04		39.96
Top-nσ 2026.04		39.95
Top-k 2026.04		39.73
Min-p 2026.04		39.73
Min-k 2026.04		39.39
Min-p 2026.04		39.06
Top-nσ 2026.04		38.84
Top-nσ 2026.04		38.62
Min-k 2026.04		38.62
Top-nσ 2026.04		38.39
Top-nσ 2026.04		38.39
Top-k 2026.04		38.17
Min-p 2026.04		37.95
Min-k 2026.04		37.95
Top-nσ 2026.04		37.72
Top-p 2026.04		37.5
Top-p 2026.04		35.71
Top-k 2026.04		34.82
Top-p 2026.04		30.13
Min-p 2026.04		29.01
Top-k 2026.04		26.56
Min-k 2026.04		25.22
Min-k 2026.04		25
Min-k 2026.04		24.55
Min-k 2026.04		24.33
Min-k 2026.04		24.33
Min-k 2026.04		24.33
Top-nσ 2026.04		23.88
Top-nσ 2026.04		23.88
Top-nσ 2026.04		23.66
Top-nσ 2026.04		23.43
Top-nσ 2026.04		23.43
Top-nσ 2026.04		23.21
Min-p 2026.04		22.55
Top-p 2026.04		22.54
Min-p 2026.04		22.43
Top-k 2026.04		21.87
Top-p 2026.04		21.2
Top-k 2026.04		20.76
Min-p 2026.04		20.11
Min-p 2026.04		17.19
Min-p 2026.04		14.88
Top-k 2026.04		14.73
Top-p 2026.04		11.16
Top-p 2026.04		8.93
Min-p 2026.04		8.56
Top-k 2026.04		8.26
Top-k 2026.04		7.81
Top-p 2026.04		3.35
Top-k 2026.04		3.34
Top-p 2026.04		3.34
Min-p 2026.04		0.51
Top-k 2026.04		0.22
Top-p 2026.04		0
Min-p 2026.04		0
Top-k 2026.04		0
Top-p 2026.04		0
Top-p 2026.04		0