Question Answering on KoBLEX
45.53Token F1DAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DARBackbone=Qwen3-32B2026.05 | 45.53 | 68.13 | |
| DARBackbone=Gemma3-27B2026.05 | 43.97 | 65.35 | |
| ParSeRBackbone=Gemma3-27B2026.05 | 42.29 | 61.16 | |
| ParSeRBackbone=Qwen3-32B2026.05 | 40.65 | 56 | |
| Self-AskBackbone=Qwen3-32B2026.05 | 33.03 | 50.52 | |
| SPBackbone=Qwen3-32B2026.05 | 31.77 | 42.89 | |
| BeamAggrBackbone=Qwen3-32B2026.05 | 31.32 | 44.05 | |
| SPBackbone=Gemma3-27B2026.05 | 31.29 | 45.54 | |
| FLAREBackbone=Qwen3-32B2026.05 | 30.7 | 46.04 | |
| IRCoTBackbone=Qwen3-32B2026.05 | 29.86 | 49.36 | |
| CoTBackbone=Gemma3-27B2026.05 | 28.51 | 48.01 | |
| ProbTreeBackbone=Gemma3-27B2026.05 | 28.06 | 49.96 | |
| BeamAggrBackbone=Gemma3-27B2026.05 | 27.64 | 45.7 | |
| FLAREBackbone=Gemma3-27B2026.05 | 27.45 | 44.27 | |
| ProbTreeBackbone=Qwen3-32B2026.05 | 27.25 | 50.67 | |
| CoTBackbone=Qwen3-32B2026.05 | 26.78 | 41.99 | |
| IRCoTBackbone=Gemma3-27B2026.05 | 22.98 | 33.95 | |
| Self-AskBackbone=Gemma3-27B2026.05 | 19.22 | 36.96 |