List your company

Scientific Reasoning on Physics (Accuracy)

81.6AccuracySC-SDPO
58.30464.35270.476.448Jun 29, 2026

Evaluation results

MethodLinks
SC-SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0681.6
DistILBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0680.8
DRIFTBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.0680.5
SRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0678.4
PGPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0677.6
SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0675.6
SDPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.0672.7
GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Cited from prior work2026.0672.7
GRPOBase Model=Qwen3-8B, Evaluation Protocol=mean@16, Source=Reproduced2026.0660.6
Qwen3-8BBase Model=Qwen3-8B, Evaluation Protocol=mean@162026.0659.2