Question Answering on ConflictBank (Held-out)
64.3AccuracyLODO SABER
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LODO SABERBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 64.3 | 56.2 | 24.7 | -2.8 | |
| Ctx-EvalBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 60.7 | 62.6 | — | — | |
| LODO SABERBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 59.6 | 66.9 | 9.7 | -5 | |
| Ctx-EvalBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 58.9 | 76 | — | — | |
| LODO SABERBackbone=Llama-3.1-8B, Setting=Zero-shot2026.05 | 54.3 | 65.3 | -4.6 | -10.7 | |
| LODO SABERBackbone=Qwen-2.5-7B, Setting=Zero-shot2026.05 | 52.3 | 64.5 | -8.4 | 1.9 | |
| Ctx-EvalBackbone=Llama-3.2-3B, Setting=Zero-shot2026.05 | 49.9 | 71.9 | — | — | |
| Ctx-EvalBackbone=Qwen-2.5-3B, Setting=Zero-shot2026.05 | 39.6 | 59 | — | — |