List your company

Scientific Knowledge on GPQA Diamond (out-of-distribution)

69.2AccuracySFT-S1.1
45.59251.72157.8563.979May 9, 2026

Evaluation results

MethodLinks
SFT-S1.1Backbone=Qwen3-4B2026.0569.27.6
HINT TUNINGBackbone=Qwen3-4B2026.0567.27.7
OriginalBackbone=Qwen3-4B2026.0566.28.8
Budget PromptBackbone=Qwen3-4B2026.0566.28.7
SFT-S1.1Backbone=DS-R1-32B2026.0563.69.3
HINT TUNINGBackbone=DS-R1-32B2026.0563.110.5
SFT-S1.1Backbone=DS-R1-14B2026.0561.110
OriginalBackbone=DS-R1-14B2026.0560.67.6
OriginalBackbone=DS-R1-32B2026.0560.67.3
HINT TUNINGBackbone=DS-R1-14B2026.0558.610.1
Budget PromptBackbone=DS-R1-32B2026.0557.14.9
OriginalBackbone=DS-R1-7B2026.0553.58.4
Budget PromptBackbone=DS-R1-14B2026.05535.5
HINT TUNINGBackbone=DS-R1-7B2026.055210.4
SFT-S1.1Backbone=DS-R1-7B2026.0550.59.6
Budget PromptBackbone=DS-R1-7B2026.0546.56.5