List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 24 (Avg@16, P@16)

8.33Avg@16PriFT-prob
-0.11482.07764.276.4624Jun 8, 2026
Updated 4mo ago

Evaluation results

MethodLinks
PriFT-probModel=Qwen2.5-Math (1.5B)2026.068.3330
IDFTModel=Qwen2.5-Math (1.5B)2026.067.3120
TALRModel=Qwen2.5-Math (1.5B)2026.066.2620
PriFT-massModel=Qwen2.5-Math (1.5B)2026.066.2533.33
DFTModel=Qwen2.5-Math (1.5B)2026.065.6316.67
OriginalModel=Qwen2.5-Math (1.5B)2026.064.9926.67
ASFTModel=Qwen2.5-Math (1.5B)2026.064.7923.33
PriFT-massModel=Qwen2.5 (1.5B)2026.063.3316.67
PriFT-massModel=Qwen2.5-Instruct (1.5B)2026.062.4916.67
PriFT-probModel=Qwen2.5-Instruct (1.5B)2026.062.2816.67
DFTModel=DeepSeekMath (7B)2026.062.2813.33
DFTModel=Qwen2.5 (1.5B)2026.061.8613.33
PretrainedModel=Qwen2.5-Instruct (1.5B)2026.061.6613.33
DFTModel=Qwen2.5-Instruct (1.5B)2026.061.6613.33
EAFTModel=Qwen2.5-Math (1.5B)2026.061.2413.33
PriFT-probModel=Qwen2.5 (1.5B)2026.061.2413.33
SFTModel=Qwen2.5-Math (1.5B)2026.061.0413.33
PriFT-probModel=DeepSeekMath (7B)2026.060.626.67
PriFT-massModel=DeepSeekMath (7B)2026.060.6210
SFTModel=Qwen2.5 (1.5B)2026.060.426.67
SFTModel=Qwen2.5-Instruct (1.5B)2026.060.416.67
PretrainedModel=DeepSeekMath (7B)2026.060.416.67
SFTModel=DeepSeekMath (7B)2026.060.416.67
PretrainedModel=Qwen2.5 (1.5B)2026.060.213.33