List your company

Scientific Reasoning on GPQA (avg.@8)

33.27Avg.@8PSFTwarm-up
17.576421.650725.72529.7993Aug 25, 2025

Evaluation results

MethodLinks
PSFTwarm-upBackbone=Qwen2.5-7B-Instruct2025.0833.27
PSFTBackbone=Qwen2.5-7B-Instruct2025.0833.21
SFT-KLBackbone=Qwen2.5-7B-Instruct2025.0832.95
SFTBackbone=Qwen2.5-7B-Instruct2025.0832.89
BaseBackbone=Qwen2.5-7B-Instruct2025.0831.38
PSFTBackbone=Llama3.1-8B-Instruct2025.0826.89
BaseBackbone=Llama3.1-8B-Instruct2025.0824.62
PSFTwarm-upBackbone=Llama3.1-8B-Instruct2025.0823.99
SFTBackbone=Llama3.1-8B-Instruct2025.0819.38
SFT-KLBackbone=Llama3.1-8B-Instruct2025.0818.18