List your company
ResearchFormerly WizWand

Mathematical Reasoning on Minerva Math (test)

55.5Best-of-8 Accuracypass@8 (Upper Bound)
46.24448.64751.0553.453May 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
pass@8 (Upper Bound)Policy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=pass@8, Evaluation Seeds=32026.0555.5————
Implicit PRM (CE)Policy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.4————
Math-Shepherd-PRM-7BPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.3————
RLHFlow-PRM-Deepseek-8BPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.2————
Qwen2.5-Math-7B-PRMPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.2————
Implicit PRM (DPO)Policy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.2————
Qwen2.5-Math-7B-PRM800KPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0547.1————
RLHFlow-PRM-Mistral-8BPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0546.9————
sPRMPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0546.7————
uPRMPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0546.7————
Skywork-PRM-7BPolicy Model=Qwen2.5-Math-7B-Instruct, Decoding Strategy=Best-of-8, Evaluation Seeds=32026.0546.6————
BaseModel=Qwen2.5-1.5B2026.05———13.976.99
DAPO+IMAXModel=Qwen3-4B2026.05———23.910.85
DAPO+IMAXModel=Qwen3-8B2026.05———23.1611.4
DEERTarget Model=Qwen3-30B-A3B, Temperature=0.62025.12—2.022.31——
EAGLE3Target Model=Qwen3-30B-A3B, Temperature=0.62025.12—1.912.07——
GRPO-fullModel=Qwen2.5-1.5B2026.05———18.0110.94
GRPO-fullModel=Qwen3-8B2026.05———29.4122.33
GRPO+IMAXModel=Qwen2.5-1.5B2026.05———18.389.47