Question Answering on ARC Easy (Accuracy and CC Metrics)
92.18Accuracy (ARC-E)Best Model
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Best ModelBackbone=LLaMA 3.1 8B, Shots=5-shot, #D=42025.10 | 92.18 | — | — | — | — | 1.14 | — | |
| BSBABackbone=LLaMA 3.1 8B, Shots=5-shot, #D=82025.10 | 90.91 | — | — | — | — | 1.37 | — | |
| BaselineBackbone=LLaMA 3.1 8B, Shots=5-shot2025.10 | 90.36 | — | — | — | — | — | — | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=B-B2025.06 | 87.72 | 13.32 | 14.17 | 15.03 | -1.9 | 12.01 | 25.72 | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=B-T2025.06 | 87.72 | 13.47 | 14.31 | 15.15 | 1.84 | 15.01 | 28.38 | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=T-T2025.06 | 86.76 | 14.97 | 15.89 | 16.82 | 1.76 | 14.88 | 27.86 | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=T-T2025.06 | 84.84 | 1.54 | 1.82 | 2.08 | 39.93 | 47.09 | 53.79 | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=B-B2025.06 | 84.26 | 0.98 | 1.35 | 1.68 | 37.4 | 45.55 | 52.88 | |
| LLaMA3.1-8BModel=LLaMA3.1-8B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=B-T2025.06 | 84.26 | 1.56 | 1.84 | 2.11 | 38.94 | 46.76 | 53.63 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=B-B2025.06 | 81.62 | 5.58 | 6.13 | 6.68 | 7.97 | 20.72 | 33.14 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=B-T2025.06 | 81.62 | 5.58 | 6.11 | 6.66 | 8.99 | 21.53 | 34.19 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIME, Evaluation Mode (B-B, B-T, T-T)=T-T2025.06 | 81.04 | 6.32 | 6.87 | 7.42 | 11.48 | 23.16 | 34.8 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=B-B2025.06 | 80.23 | 1.2 | 1.62 | 2.01 | 33.77 | 43.15 | 51.75 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=B-T2025.06 | 80.23 | 1.25 | 1.66 | 2.03 | 34.88 | 44.02 | 52.52 | |
| LLaMA3.2-3BModel=LLaMA3.2-3B, Attribution Method (LIME/LIG)=LIG, Evaluation Mode (B-B, B-T, T-T)=T-T2025.06 | 79.27 | 1.71 | 2.07 | 2.43 | 37.97 | 46.24 | 53.74 |