Content Moderation on Toxic-Chat Out-of-Distribution
60.8Pornography ScoreCARO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CARObackbone=Qwen2.5-7B-Instruct2026.04 | 60.8 | 42.2 | 97.7 | 95 | |
| Qwen2.5-7B-Instructmodel_status=Base Model2026.04 | 59 | 17.6 | 96.9 | 93.3 |
Loading the SOTA2 catalog…