List your company

Mathematical Reasoning on MATH 500 (avg.@8)

84.68Avg Success Rate (@8)PSFTwarm-up
47.104856.859966.61576.3701Aug 25, 2025

Evaluation results

MethodLinks
PSFTwarm-upBackbone=Qwen2.5-7B-Instruct2025.0884.68
SFTBackbone=Qwen2.5-7B-Instruct2025.0884.1
SFT-KLBackbone=Qwen2.5-7B-Instruct2025.0883.55
PSFTBackbone=Qwen2.5-7B-Instruct2025.0883.35
BaseBackbone=Qwen2.5-7B-Instruct2025.0875.05
PSFTwarm-upBackbone=Llama3.1-8B-Instruct2025.0874.15
SFTBackbone=Llama3.1-8B-Instruct2025.0872.6
PSFTBackbone=Llama3.1-8B-Instruct2025.0871.98
SFT-KLBackbone=Llama3.1-8B-Instruct2025.0869.83
BaseBackbone=Llama3.1-8B-Instruct2025.0848.55