Multiple Choice Question Answering on MMLU-Pro (Economics Discipline)
10.01Mean EMRQwen-2-7B-Instruct
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Qwen-2-7B-InstructOutlier Detection=ConU (basic framework), alpha=0.12025.04 | 10.01 | 0.0279 | 10.4 | |
| Qwen-2-7B-InstructOutlier Detection=SConU (current paper), alpha=0.12025.04 | 9.65 | 0.021 | 9.6 | |
| LLaMA-3.1-8B-InstructOutlier Detection=ConU (basic framework), alpha=0.12025.04 | 9.47 | 0.0363 | 10 | |
| LLaMA-3.1-8B-InstructOutlier Detection=SConU (current paper), alpha=0.12025.04 | 9.16 | 0.0242 | 9.2 |