Relevance Correlation Evaluation on HJM (All)
0.516Spearman's RhoBehavior-Grounded LLM judge
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Behavior-Grounded LLM judgeJudge Type=Behavior-Grounded2026.07 | 0.516 | 0.476 | |
| Plain LLM judgeJudge Type=Plain2026.07 | 0.45 | 0.413 |
Loading the SOTA2 catalog…