Model Alignment on HH-RLHF 0-shot (test)
62.68Harmlessness BLEUChatGPT
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ChatGPTshot=0-shot2026.04 | 62.68 | 10.29 | 73.01 | 70.79 | 11.86 | 75.11 | 68.6 | 11.41 | 74.54 | |
| Mistral-Instructshot=0-shot2026.04 | 30.9 | 3.33 | 63.5 | 34.6 | 3.9 | 64.8 | 33.6 | 3.74 | 64.45 | |
| Llama3-Instructshot=0-shot2026.04 | 23 | 3.06 | 66.67 | 33.47 | 3.74 | 65.69 | 30.64 | 3.54 | 65.96 | |
| Llama3-Baseshot=0-shot2026.04 | 10.51 | 1.8 | 53.23 | 18.74 | 2.02 | 46.97 | 16.51 | 1.96 | 48.66 | |
| Mistral-Baseshot=0-shot2026.04 | 8.1 | 1.73 | 53.51 | 14.18 | 1.87 | 45.57 | 12.54 | 1.83 | 47.72 |