List your company

Mathematical Problem Solving on OlympiadBench (Accuracy, Refinement Gap)

71AccuracyselfRef@4
38.2446.74555.2563.755Aug 27, 2025

Evaluation results

MethodLinks
selfRef@4backbone=GSR-7B2025.08712.7
BoN@4backbone=GSR-7B, Aggregation Model=RM-R1-7B2025.0869.1—
maj@5backbone=GSR-7B2025.0868.8—
maj@4backbone=GSR-7B2025.0868.3—
BoN@5backbone=GSR-7B, Aggregation Model=SkyworkRM-27B2025.0867.7—
Ref@4backbone=GSR-7B, Aggregation Model=Synthesizer-8B2025.0867.7—
BoN@4backbone=GSR-7B, Aggregation Model=SkyworkRM-27B2025.0867.6—
BoN@4backbone=GSR-7B, Aggregation Model=RRM-7B2025.0864.6—
pass@1backbone=GSR-7B2025.0864.4—
QwQ-Previewmode=pass@12025.0861.2—
selfRef@4backbone=Qwen2.5-7B-Instruct (Base)2025.0843.80.3
maj@4backbone=Qwen2.5-7B-Instruct (Base)2025.0843.5—
BoN@4backbone=Qwen2.5-7B-Instruct (Base), Aggregation Model=SkyworkRM-27B2025.0843—
pass@1backbone=Qwen2.5-7B-Instruct (Base)2025.0839.5—