Explanation Generation on ArMeme (test)
74.1BSProposed (+ Self-Sup.)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Proposed (+ Self-Sup.)Training Protocol=Self-supervised GRPO, Inference Strategy=Reasoning-based explanation generation2026.06 | 74.1 | 44.2 | |
| Proposed (Sup.)Training Protocol=Supervised GRPO, Inference Strategy=Reasoning-based explanation generation2026.06 | 74 | 44.1 | |
| [43]Context=Prior work2026.06 | 70 | 35 |