Constitutional AI Alignment on HH-RLHF (test)
4.596Likert Score RankingReflect
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ReflectBackbone Model=GPT-4.1-mini2026.01 | 4.596 | 0.02 | 0.372 | |
| ReflectBackbone Model=Claude-3.5-Haiku2026.01 | 4.406 | 0.06 | 0.462 | |
| CCBaseBackbone Model=GPT-4.1-mini2026.01 | 4.224 | 0.1036 | — | |
| ReflectBackbone Model=Mistral-7B2026.01 | 4.179 | 0.0766 | 0.218 | |
| CCBaseBackbone Model=Mistral-7B2026.01 | 3.961 | 0.1384 | — | |
| CCBaseBackbone Model=Claude-3.5-Haiku2026.01 | 3.944 | 0.1686 | — |