Question Answering on ConFiQA-QA (Held-out)
67AccuracyImp-SCR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Imp-SCRBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 67 | 86 | — | — | |
| Imp-SCRBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 66.8 | 73.7 | — | — | |
| LODO SABERBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 65.5 | 72.9 | -1.3 | -0.8 | |
| Imp-SCRBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 57.2 | 66.9 | — | — | |
| LODO SABERBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 56.1 | 72.9 | 3.5 | 1.5 | |
| LODO SABERBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 55.8 | 71.5 | -11.2 | -14.5 | |
| Ctx-EvalBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 52.6 | 71.4 | — | — | |
| LODO SABERBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 49.8 | 65.3 | -7.5 | -1.6 |