Pairwise Discrimination on Management and Economics Research Pitch Pairs shared pairwise subset (test)
78.67Distance 1 AccuracySFT GPT-4.1
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SFT GPT-4.1task=label-free2026.03 | 78.67 | 89 | 92 | 84.33 | |
| GPT-5.2 Hightask=label-free2026.03 | 69.33 | 85 | 94 | 78.67 | |
| GPT-4.1 (baseline)task=label-free2026.03 | 69.33 | 79 | 90 | 76 | |
| Gemini 3.1 Protask=label-free2026.03 | 68.67 | 86 | 86 | 77.33 |