LLM Selection on Open LLM Leaderboard Legal Document Review scenario K = 42 models
69.9Main Objective Success RatePrefUCB
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| PrefUCB2026.06 | 69.9 | 46 | 80.1 | |
| ScalarizedUCBr.r.2026.06 | 62.7 | 58.5 | 68.1 | |
| GT-PF2026.06 | 55.7 | 57.1 | 59.3 | |
| ParetoUCB2026.06 | 52.2 | 53.2 | 56.8 |