Multiple Choice Question Answering on MMLU redux
85.11AccuracyGaC + SAFE
Evaluation Results
| Method | Links | |
|---|---|---|
| GaC + SAFEEnsembling models=Qwen2.5 + EXAONE3.52025.10 | 85.11 | |
| Qwen2.5-32BScale=32B2025.10 | 85.06 | |
| GaCEnsembling models=Qwen2.5 + EXAONE3.52025.10 | 84.7 | |
| EXAONE3.5-32BScale=32B2025.10 | 82.34 |