Human Evaluation of Explanations on DRAMA-Reasoning (val)
74.7Usefulness (UniDrive Win %)UniDrive
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| UniDrive2026.06 | 74.7 | — | — | 76.8 | — | — | 82 | — | — | |
| Tie / No Preference2026.06 | — | — | 5.3 | — | — | 4.4 | — | — | 5.1 | |
| Video-LLAMA2026.06 | — | 20 | — | — | 18.8 | — | — | 12.9 | — |