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

Reasoning on GSM8K (maj@2, maj@4, maj@8)

41.17Accuracy (maj@4)

DLE (top-p&top-k)

Updated 1mo ago

Evaluation Results

Method	Links
DLE (top-p&top-k) 2026.04		41.17	34.8	44.43
DLE (ε-sampling) 2026.04		40.64	34.57	44.05
DLE (min-p) 2026.04		39.58	33.74	43.97
Beam search 2026.04		36.47	35.1	36.92
Diverse beam search 2026.04		36.47	-	41.39
Self-consistency 2026.04		36.24	29.57	41.32
Self-consistency (min-p) 2026.04		33.51	26.46	39.04
Self-consistency (ε-sampling) 2026.04		33.51	26.84	40.94
Self-consistency (top-p&top-k) 2026.04		31.84	25.47	38.74
DeepConf 2026.04		26.91	21.68	33.36
Self-consistency 2026.04		25.32	19.26	31.61
Self-certainty 2026.04		25.32	20.09	32.22