Question Answering on Open QA
45.7Exact Match (EM)TUR-DPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TUR-DPOEvaluation Protocol=Human eval2026.04 | 45.7 | 45.7 | |
| PPOEvaluation Protocol=Human eval2026.04 | 45.4 | 45.4 | |
| TUR-DPOEvaluation Protocol=LLM judge2026.04 | 45.1 | 45.1 | |
| DPOEvaluation Protocol=Human eval2026.04 | 45 | 45 | |
| IPOEvaluation Protocol=LLM judge2026.04 | 42.5 | 42.5 | |
| DPOEvaluation Protocol=LLM judge2026.04 | 41.8 | 41.8 |