Question Answering on ARC Challenge (Accuracy and Rank)
66.3AccuracyBaseline
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BaselineBackbone=LLaMA-3.1-8B-Instruct2026.05 | 66.3 | — | |
| SKOPBackbone=LLaMA-3.1-8B-Instruct2026.05 | 65 | 2.69 | |
| Angular SteerBackbone=LLaMA-3.1-8B-Instruct2026.05 | 59.2 | 4.75 | |
| SADIBackbone=LLaMA-3.1-8B-Instruct2026.05 | 48.8 | 3.38 | |
| CASTBackbone=LLaMA-3.1-8B-Instruct2026.05 | 40.5 | 5.63 | |
| LoRABackbone=LLaMA-3.1-8B-Instruct2026.05 | 38.5 | — | |
| DISCO-QBackbone=LLaMA-3.1-8B-Instruct2026.05 | 34.2 | 4.75 | |
| ITIBackbone=LLaMA-3.1-8B-Instruct2026.05 | 29.1 | 4.75 | |
| CAABackbone=LLaMA-3.1-8B-Instruct2026.05 | 25 | 5.44 | |
| Comm SteerBackbone=LLaMA-3.1-8B-Instruct2026.05 | 15.8 | 4.63 |