List your company
ResearchFormerly WizWand

Mathematical Reasoning on Olympiad Bench (pass@8)

48.5Pass@8VI-CuRL
13.97222.93631.940.864Feb 13, 2026Feb 27, 2026Mar 14, 2026Mar 29, 2026Apr 12, 2026Apr 27, 2026May 12, 2026
Updated 5mo ago

Evaluation results

MethodLinks
VI-CuRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.0248.5
VI-CuRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Intrinsic Reward Signal=Majority Vote2026.0247
PionBase Model=Qwen3-1.7B2026.0546.43
PionBase Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0546.43
AdamWBase Model=Qwen3-1.7B2026.0546.09
No CurriculumBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.0245.8
AdamWBase Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0544.69
MuonBase Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0544.61
AdaRFTBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.0244.5
AdaRFTBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.0243.3
MuonBase Model=Qwen3-1.7B2026.0542.41
VI-CuRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.0241.9
VCRLBackbone=Qwen2.5-Math-7B, Reward Setting=Oracle (w. Verifier)2026.0241
VCRLBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.0238
VI-CuRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.0237.9
VI-CuRLBackbone=Qwen2.5-Math-1.5B, Intrinsic Reward Signal=Majority Vote2026.0237.1
BaseBase Model=DeepSeek-R1-Distill-Qwen-1.5B2026.0536.2
No CurriculumBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=Oracle (w. Verifier)2026.0234.7
AdaRFTBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.0234.4
No CurriculumBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.0234.3
VCRLBackbone=Qwen2.5-Math-1.5B, Reward Setting=Oracle (w. Verifier)2026.0234
BaseBackbone=Qwen2.5-Math-7B, Reward Setting=No RL2026.0233.9
BaseBackbone=Qwen2.5-Math-1.5B, Reward Setting=No RL2026.0231.3
BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Reward Setting=No RL2026.0229.8
BaseBase Model=Qwen3-1.7B2026.0523.67
VI-CuRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.0219.4
No CurriculumBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.0217.8
VCRLBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.0217.5
BaseBackbone=Llama3.2-3B-Instruct, Reward Setting=No RL2026.0216.4
AdaRFTBackbone=Llama3.2-3B-Instruct, Reward Setting=Oracle (w. Verifier)2026.0215.3