Mathematical Reasoning on GPQA-D
73.18AccuracyDIS-GMM*
Evaluation results
| Method | Links | |
|---|---|---|
| DIS-GMM*Model=Qwen3-32B, Thinking mode=false, SelfStepConf=true2026.03 | 73.18 | |
| DIS-GMMModel=Qwen3-32B, Thinking mode=true2026.03 | 72.74 | |
| WSC-GMM*Model=Qwen3-32B, Thinking mode=false, SelfStepConf=true2026.03 | 72.43 | |
| WSC-GMMModel=Qwen3-32B, Thinking mode=true2026.03 | 72.42 | |
| DIS-Top50Model=Qwen3-32B, Thinking mode=true2026.03 | 72.33 | |
| MoB-AdaptiveModel=Qwen3-32B, Thinking mode=true2026.03 | 72.08 | |
| WSC-Top50Model=Qwen3-32B, Thinking mode=true2026.03 | 72.03 | |
| BoNModel=Qwen3-32B, Thinking mode=true2026.03 | 72 | |
| DIS-GMM*Model=DeepSeek-R1-8B, SelfStepConf=true2026.03 | 70.63 | |
| WSCModel=Qwen3-32B, Thinking mode=true2026.03 | 70.55 | |
| SCModel=Qwen3-32B, Thinking mode=true2026.03 | 70.3 | |
| WSC-GMM*Model=DeepSeek-R1-8B, SelfStepConf=true2026.03 | 70.11 | |
| DIS-GMMModel=DeepSeek-R1-8B, SelfStepConf=false2026.03 | 69.82 | |
| WSC-GMMModel=DeepSeek-R1-8B, SelfStepConf=false2026.03 | 69.58 | |
| DIS-Top50Model=DeepSeek-R1-8B2026.03 | 69.52 | |
| MoB-AdaptiveModel=DeepSeek-R1-8B2026.03 | 68.7 | |
| WSC-Top50Model=DeepSeek-R1-8B2026.03 | 68.58 | |
| BoNModel=DeepSeek-R1-8B2026.03 | 67.91 | |
| WSCModel=DeepSeek-R1-8B2026.03 | 67.65 | |
| SCModel=DeepSeek-R1-8B2026.03 | 67.5 |