Human Evaluation on RCC-PVD (evaluation)
61Rank Preference RateRanking
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RankingData Protocol=Error Injection, Comparison Baseline=vs. DPO2026.03 | 61 | 39 | |
| RankingData Protocol=Error Injection, Comparison Baseline=vs. MPO2026.03 | 60 | 40 | |
| RankingData Protocol=Caption Chains, Comparison Baseline=vs. DPO2026.03 | 56 | 44 | |
| RankingData Protocol=Caption Chains, Comparison Baseline=vs. MPO2026.03 | 53 | 47 |