Question Answering on ConFiQA-MC (Held-out)
52.8AccuracyLODO SABER
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LODO SABERBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 52.8 | 88.9 | 15.9 | 18.5 | |
| LODO SABERBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 52.8 | 87.7 | 28.7 | 28.8 | |
| LODO SABERBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 50.7 | 82.7 | 19.6 | 20.7 | |
| LODO SABERBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 47.4 | 79 | 5.8 | 15.9 | |
| Ctx-EvalBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 41.6 | 63 | — | — | |
| Ctx-EvalBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 36.8 | 70.4 | — | — | |
| Exp-SCRBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 31.2 | 62 | — | — | |
| Ctx-EvalBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 24.2 | 58.8 | — | — |