| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Preference Alignment | Koala | Wins (Count)196 | 14 | |
| Instruction Following | Koala low-resource | Win Rate (bn)58.3 | 7 | |
| Overrefusal Evaluation | Koala | Refusal Rate4.44 | 6 | |
| Camera-controlled video generation | Koala | RotErr0.637 | 6 | |
| Personalized LLM response generation | Koala (test) | Win Rate (Reward Model)88 | 3 | |
| Preference Alignment | Koala | Win Rate (Reward Model)77.75 | 3 |