List your company

Scientific Problem Solving on PhysicsQA

53.22AccuracyReason, Reward, Refine
17.54826.80936.0745.331Jul 6, 2026

Evaluation results

MethodLinks
Reason, Reward, RefineBackbone=Phi 3.5 Mini 3.8B2026.0753.22
Reason, Reward, RefineBackbone=Qwen 2.5 1.5B2026.0749.62
Reason, Reward, RefineBackbone=LLaMA 3.2 3B2026.0746.73
RAGBackbone=Phi 3.5 Mini 3.8B2026.0741.59
DPOBackbone=Phi 3.5 Mini 3.8B2026.0741.49
FTBackbone=Phi 3.5 Mini 3.8B2026.0739.19
Reason, Reward, RefineBackbone=LLaMA 3.2 1B2026.0739.02
RAGBackbone=Qwen 2.5 1.5B2026.0736.91
CoTBackbone=Phi 3.5 Mini 3.8B2026.0733.35
RAGBackbone=LLaMA 3.2 3B2026.0731.21
CoTBackbone=Qwen 2.5 1.5B2026.0730.64
RAGBackbone=LLaMA 3.2 1B2026.0728.64
FTBackbone=LLaMA 3.2 3B2026.0727.84
CoTBackbone=LLaMA 3.2 3B2026.0727.67
DPOBackbone=LLaMA 3.2 3B2026.0725.95
DPOBackbone=Qwen 2.5 1.5B2026.0724.32
FTBackbone=Qwen 2.5 1.5B2026.0723.51
FTBackbone=LLaMA 3.2 1B2026.0723.24
CoTBackbone=LLaMA 3.2 1B2026.0721.35
DPOBackbone=LLaMA 3.2 1B2026.0718.92