LLM-as-a-judge on Chatbot Arena
94.3CoverageHeuristic Selection
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Heuristic SelectionCascade=M → L → O, Target agreement level=0.852026.05 | 94.3 | 12.6 | |
| Heuristic SelectionCascade=L → Q → O, Target agreement level=0.852026.05 | 88.4 | 36.1 | |
| CSE + OursCascade=M → L → O, Target agreement level=0.852026.05 | 85.2 | 77.9 | |
| CSE + SimulatedCascade=M → L → O, Target agreement level=0.852026.05 | 84.8 | 74 | |
| CSE + VanillaCascade=M → L → O, Target agreement level=0.852026.05 | 84.8 | 74.4 | |
| CSE + PredictiveCascade=M → L → O, Target agreement level=0.852026.05 | 83.6 | 72.3 | |
| CSE + RandomCascade=M → L → O, Target agreement level=0.852026.05 | 83.4 | 70 | |
| CSE + OursCascade=L → Q → O, Target agreement level=0.852026.05 | 56.4 | 92.7 | |
| CSE + PredictiveCascade=L → Q → O, Target agreement level=0.852026.05 | 56.2 | 87.1 | |
| CSE + VanillaCascade=L → Q → O, Target agreement level=0.852026.05 | 52.4 | 90.8 | |
| CSE + SimulatedCascade=L → Q → O, Target agreement level=0.852026.05 | 51.8 | 90.2 | |
| CSE + RandomCascade=L → Q → O, Target agreement level=0.852026.05 | 49.5 | 89.6 |