List your company

Mathematics on OlympiadBench (pass@1 %)

78.72Pass@1 AccuracyCOSE
9.955227.807645.6663.5124Apr 21, 2025Jun 26, 2025Sep 1, 2025Nov 7, 2025Jan 13, 2026Mar 21, 2026May 27, 2026

Evaluation results

MethodLinks
COSEBackbone=Qwen3-4B2026.0578.72
COSEBackbone=Llama-3.2-3B-Instruct2026.0572
MAEBackbone=Qwen3-4B2026.0569.4
R-ZeroBackbone=Qwen3-4B2026.0568.9
AZRBackbone=Qwen3-4B2026.0568.6
BaseBackbone=Qwen3-4B2026.0566.8
INSIGHTModel=QWEN3-4B2026.0357
MoPPSModel=QWEN3-4B2026.0356.45
RANDOMModel=QWEN3-4B2026.0356.24
INVERSE-EVIDENCEModel=QWEN3-4B2026.0356.2
DSModel=R1-DISTILL-QWEN-7B2026.0353.38
INSIGHTModel=R1-DISTILL-QWEN-7B2026.0352.5
EXPECTED-DIFFICULTYModel=R1-DISTILL-QWEN-7B2026.0352.25
INVERSE-EVIDENCEModel=R1-DISTILL-QWEN-7B2026.0352.07
MoPPSModel=R1-DISTILL-QWEN-7B2026.0352.01
RANDOMModel=R1-DISTILL-QWEN-7B2026.0351.8
Scaf-GRPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.1050.7
Vanilla GRPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.1050.2
MAEBackbone=Llama-3.2-3B-Instruct2026.0547.6
R-ZeroBackbone=Llama-3.2-3B-Instruct2026.0546.3
DeepSeek-R1-Distill-Qwen-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.1045.8
AZRBackbone=Llama-3.2-3B-Instruct2026.0544.2
Scaf-GRPOBase Model=Qwen2.5-Math-7B2025.1043.3
LUFFYBase Model=Qwen2.5-Math-7B2025.1041.7
Vanilla GRPOBase Model=Qwen2.5-Math-7B2025.1041.3
Oat-ZeroBase Model=Qwen2.5-Math-7B2025.1041
Scaf-GRPOBase Model=Qwen2.5-7B2025.1040.8
Vanilla GRPOBase Model=Qwen2.5-7B2025.1040.4
BaseBackbone=Llama-3.2-3B-Instruct2026.0538.1
SimpleRL-ZeroBase Model=Qwen2.5-Math-7B2025.1037.2
Scaf-GRPOBase Model=Qwen2.5-Math-1.5B2025.1036.6
Vanilla GRPOBase Model=Qwen2.5-Math-1.5B2025.1034.8
INVERSE-EVIDENCEModel=QWEN3-0.6B2026.0334.8
INSIGHTModel=QWEN3-0.6B2026.0334.8
Qwen2.5-7BBase Model=Qwen2.5-7B2025.1034.4
MoPPSModel=QWEN3-0.6B2026.0334.25
RANDOMModel=QWEN3-0.6B2026.0334.2
Qwen2.5-Math-1.5BBase Model=Qwen2.5-Math-1.5B2025.1020.6
Scaf-GRPOBase Model=Llama-3.2-3B-Instruct2025.1020.3
IPPO (MAPoRL)Backbone=LaMAS, Scale=Trio2025.0419.05
MARFTBackbone=LaMAS, Scale=Trio2025.0418.72
MARFTBackbone=LaMAS, Scale=Quad2025.0418.55
MARFTBackbone=LaMAS, Scale=Duo2025.0418.43
IPPO (MAPoRL)Backbone=LaMAS, Scale=Quad2025.0418.37
IPPO (MAPoRL)Backbone=LaMAS, Scale=Duo2025.0418.31
Vanilla GRPOBase Model=Llama-3.2-3B-Instruct2025.1018.3
Qwen2.5-Math-7BBase Model=Qwen2.5-Math-7B2025.1018.2
VanillaBackbone=LaMAS, Scale=Trio2025.0417.6
VanillaBackbone=LaMAS, Scale=Quad2025.0416.6
VanillaBackbone=LaMAS, Scale=Duo2025.0416.08
Llama-3.2-3B-InstructBase Model=Llama-3.2-3B-Instruct2025.1012.6