Question Answering with Clarification on AmbigQA (dev)
15.81RewardSGP-Oracle
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SGP-OracleBase Model=Gemma-2-9b2025.12 | 15.81 | 23.13 | 35.78 | 11.08 | |
| SGPBase Model=Gemma-2-9b2025.12 | 6.63 | 19.94 | 40.25 | 18.22 | |
| AnswerBase Model=Gemma-2-9b2025.12 | 1.32 | 14.33 | 0 | 0 | |
| ClarifyBase Model=Gemma-2-9b2025.12 | -9.59 | 16.3 | 100 | 0 | |
| Prompted-COTBase Model=Gemma-2-9b2025.12 | -20.11 | 16.44 | 53.93 | 50.48 | |
| PromptedBase Model=Gemma-2-9b2025.12 | -34.49 | 16.16 | 66.83 | 63.88 | |
| Multi-answer (MA)Base Model=Gemma-2-9b2025.12 | -46.48 | 15.67 | 0 | 100 | |
| Clarify-and-MABase Model=Gemma-2-9b2025.12 | -50.61 | 16.94 | 100 | 100 |