Preference Reconstruction on Consequences
75.23Preference AccuracyDICAI
Evaluation results
| Method | Links | |
|---|---|---|
| DICAIJudge LLM=GPT-52026.06 | 75.23 | |
| DICAIJudge Model=GPT-4o2026.06 | 73.21 | |
| ICAIJudge LLM=GPT-52026.06 | 67.8 | |
| ICAIJudge Model=GPT-4o2026.06 | 61.2 | |
| AutoRubricJudge LLM=GPT-52026.06 | 56.77 | |
| CoTJudge LLM=GPT-52026.06 | 54.56 | |
| Self-RefineJudge LLM=GPT-52026.06 | 52.51 | |
| AutoRubricJudge Model=GPT-4o2026.06 | 52.44 | |
| ToTJudge LLM=GPT-52026.06 | 52.27 | |
| Self-RefineJudge Model=GPT-4o2026.06 | 51.1 | |
| CoT-SCJudge LLM=GPT-52026.06 | 50.33 | |
| CoTJudge Model=GPT-4o2026.06 | 48.29 | |
| ToTJudge Model=GPT-4o2026.06 | 47.4 | |
| CoT-SCJudge Model=GPT-4o2026.06 | 47.24 |