Binary Classification on ImageReward (test)
65.07Macro-F1RL-Single
Evaluation Results
| Method | Links | |
|---|---|---|
| RL-Singletraining=Single-task RL2026.03 | 65.07 | |
| MT-RL-Judgetraining=Multi-task RL2026.03 | 64.97 | |
| SFT-Singletraining=Single-task SFT2026.03 | 64.95 | |
| SFT-Unifiedtraining=Unified multi-task SFT2026.03 | 63.34 | |
| Off-the-shelf2026.03 | 55.07 |