Question Answering on MMLU (FDR, Power, and Ratio Analysis)
7FDRSGR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SGRModel=Llama-3.1-8B-Instruct2025.10 | 7 | 48 | 33.13 | — | |
| SGRModel=Qwen3-32B2025.10 | 8.18 | 78.14 | 66.88 | — | |
| SGRModel=Llama-3.1-70B-Instruct2025.10 | 8.21 | 84 | 74.85 | — | |
| FDR searchModel=Qwen3-32B2025.10 | 9.84 | 82.56 | 71.98 | — | |
| Conformal LabelingModel=Llama-3.1-8B-Instruct2025.10 | 9.85 | 57.13 | 40.64 | — | |
| FDR searchModel=Llama-3.1-70B-Instruct2025.10 | 9.9 | 88.1 | 79.98 | — | |
| Conformal LabelingModel=Llama-3.1-70B-Instruct2025.10 | 9.97 | 88.25 | 80.17 | — | |
| Conformal LabelingModel=Qwen3-32B2025.10 | 10 | 82.95 | 72.44 | — | |
| FDR searchModel=Llama-3.1-8B-Instruct2025.10 | 58.94 | 76.77 | 68.74 | — | |
| AI onlyModel=Llama-3.1-8B-Instruct2025.10 | — | — | — | 35.91 | |
| AI onlyModel=Qwen3-32B2025.10 | — | — | — | 21.43 | |
| AI onlyModel=Llama-3.1-70B-Instruct2025.10 | — | — | — | 18.24 |