List your company

Mathematical Reasoning on Math Benchmarks Overall (test)

87Pass@1SimPOShortest
70.46474.75779.0583.343Jan 20, 2026

Evaluation results

MethodLinks
SimPOShortestBackbone=Qwen3-8B2026.018780.1
RELIEFBackbone=Qwen3-8B2026.0186.171.3
BeliefPromptBackbone=Qwen3-8B2026.0185.389
SFTbeliefBackbone=Qwen3-8B2026.0184.895.4
OriginalBackbone=Qwen3-8B2026.0184.7100
SFTShortestBackbone=Qwen3-8B2026.0183.998.8
RELIEFBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0175.389.7
SimPOShortestBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0174.791.5
OriginalBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0174100
BeliefPromptBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0173.990.6
SFTShortestBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0171.785.1
SFTbeliefBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0171.190.5