Binary Classification on AGIN-Nat. (test)
81.75Macro-F1SFT-Unified
Evaluation Results
| Method | Links | |
|---|---|---|
| SFT-Unifiedtraining=Unified multi-task SFT2026.03 | 81.75 | |
| MT-RL-Judgetraining=Multi-task RL2026.03 | 81.63 | |
| RL-Singletraining=Single-task RL2026.03 | 80.5 | |
| SFT-Singletraining=Single-task SFT2026.03 | 78.64 | |
| Off-the-shelf2026.03 | 67.99 |