List your company

General Domain Reasoning on GPQA (Accuracy)

40.4AccuracyRFT
9.92817.83925.7533.661Jul 2, 2025

Evaluation results

MethodLinks
RFTBase Model=Qwen2.5-Math-7B2025.0740.4
LUFFYBase Model=Qwen2.5-Math-7B2025.0739.9
Prefix-RFTBase Model=Qwen2.5-Math-7B2025.0739.1
ReLIFTBase Model=Qwen2.5-Math-7B2025.0737.9
UFTBase Model=Qwen2.5-Math-7B2025.0734.5
SFTBase Model=Qwen2.5-Math-7B2025.0724.7
SFT+RFTBase Model=Qwen2.5-Math-7B2025.0724.2
Oat-ZeroBase Model=Qwen2.5-Math-7B2025.0723.7
RL w/ SFT LossBase Model=Qwen2.5-Math-7B2025.0723.7
SimpleRL-ZeroBase Model=Qwen2.5-Math-7B2025.0723.2
Qwen2.5-Math-7B2025.0711.1