Debate on MMLU
61.2AccOPTIMA-iSFT-DPO SC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OPTIMA-iSFT-DPO SCoptimization_strategy=iSFT-DPO, self_consistency=true2024.10 | 61.2 | 1,026.7 | |
| DITS-iSFT-DPO2025.02 | 60.5 | — | |
| OPTIMA-iSFT-DPOoptimization_strategy=iSFT-DPO2024.10 | 60.2 | 56.7 | |
| OPTIMA-iSFT SCoptimization_strategy=iSFT, self_consistency=true2024.10 | 60.2 | 874.5 | |
| Optima-iSFT-DPO2025.02 | 60.2 | — | |
| OPTIMA-iDPO SCoptimization_strategy=iDPO, self_consistency=true2024.10 | 59.9 | 1,050.4 | |
| OPTIMA-iDPOoptimization_strategy=iDPO2024.10 | 59.6 | 61.6 | |
| Optima-iDPO2025.02 | 59.6 | — | |
| OPTIMA-iSFToptimization_strategy=iSFT2024.10 | 56.8 | 123.8 | |
| Optima-iSFT2025.02 | 56.8 | — | |
| SCn=82024.10 | 54 | 1,056.1 | |
| MAD2024.10 | 51.5 | 516.7 | |
| MAD2025.02 | 51.5 | — | |
| CoT2024.10 | 46 | 132.2 | |
| CoT2025.02 | 46 | — | |
| AutoForm2024.10 | 43.8 | 198.5 | |
| AutoForm2025.02 | 43.8 | — |