Reasoning on ReClor OOD
87.8AccuracyGPT-4.5
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4.52026.06 | 87.8 | |
| IGA (Ours)2026.06 | 82.1 | |
| Pareto-GS2026.06 | 76.5 | |
| SAND-Mask2026.06 | 74.3 | |
| PCGrad2026.06 | 73.6 | |
| V-REx2026.06 | 73.1 | |
| IRM2026.06 | 72.7 | |
| ERM-SFT2026.06 | 70.3 | |
| CoT-Distill2026.06 | 69.8 | |
| LoRA-SFT2026.06 | 69.1 |