List your company
ResearchFormerly WizWand

Mathematical Reasoning on Olympiad (pass@1, pass@8)

65.2Pass@1ICPO
2.633618.876835.1251.3632Sep 26, 2025Nov 5, 2025Dec 16, 2025Jan 26, 2026Mar 7, 2026Apr 17, 2026May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
ICPOBase Model=8B, Training Method=ICPO2025.1065.2—
ICPO†Base Model=8B, Training Method=ICPO†2025.1064.3—
GRPOBase Model=8B, Training Method=GRPO2025.1062.4—
GRPOExtraRolloutsBase Model=8B, Training Method=GRPOExtraRollouts2025.1061.6—
GRPO w/ MAPRBase Model=Qwen3-14B2025.0961.5974.37
GRPOExpertDomainBase Model=8B, Training Method=GRPOExpertDomain2025.1060—
GRPOBase Model=Qwen3-14B2025.0959.0473.03
GRPO w/ MAPRBase Model=Qwen3-8B2025.0956.8671.87
ICPOBase Model=1.7B, Training Method=ICPO2025.1056.4—
GRPOBase Model=Qwen3-8B2025.0954.0370.04
GRPOExtraRolloutsBase Model=1.7B, Training Method=GRPOExtraRollouts2025.1053.8—
GRPO w/ MAPRBase Model=Qwen3-4B2025.0953.3869.74
ICPO†Base Model=1.7B, Training Method=ICPO†2025.1052.7—
GRPOExpertDomainBase Model=1.7B, Training Method=GRPOExpertDomain2025.1050.5—
GRPOBase Model=1.7B, Training Method=GRPO2025.1048.2—
GRPOBase Model=Qwen3-4B2025.0944.4761.99
Qwen3Base Model=8B, Training Method=Base2025.1043.3—
Qwen3Base Model=1.7B, Training Method=Base2025.1040.6—
MTTeacher Model=Qwen2-Math-7B-Instruct2026.0437.6—
VeriGatePolicy Model=Qwen2.5-7B-Instruct, PRM Model=Qwen-PRM-7B2026.0531—
GRPO-VanillaPolicy Model=Qwen2.5-7B-Instruct, PRM Model=Qwen-PRM-7B2026.0530.67—
PRM-as-ORMPolicy Model=Qwen2.5-7B-Instruct, PRM Model=Qwen-PRM-7B2026.0530.67—
BHA2026.0426.941.2
DAPO2026.0426.440.9
BREAD2026.0425.540.7
VCRDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0424.7—
DistillLM-2Teacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0424.4—
KDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0423.9—
GKDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0423.7—
DistilLLMTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0423.7—
ABKDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0422.4—
Dr.GRPOPolicy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0515.5—
PRM-as-ORM (+Gate)Policy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0515.5—
VeriGatePolicy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0515.5—
GRPO-VanillaPolicy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0515.29—
Base2026.0414.440.7
DAPOPolicy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0514.07—
VeriGate (Simplified)Policy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0513.63—
VeriGate (-Gate)Policy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.0510.37—
MSStudent Model=Qwen2-Math-1.5B2026.047.4—
PRM-as-ORMPolicy Model=Qwen2.5-1.5B-Instruct, PRM Model=Qwen-PRM-7B2026.055.04—