Response Generation on Open Assistant 953 prompts (test)
1,294Elo RatingGPT-4
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4Judge=GPT-4, # Prompts=9532023.05 | 1,294 | 1 | 1 | |
| ChatGPT-3.5 TurboJudge=GPT-4, # Prompts=9532023.05 | 1,015 | 2 | 5 | |
| Guanaco-65BJudge=GPT-4, # Prompts=9532023.05 | 1,008 | 3 | 2 | |
| Guanaco-33BJudge=GPT-4, # Prompts=9532023.05 | 1,002 | 4 | 4 | |
| Vicuna-13BJudge=GPT-4, # Prompts=9532023.05 | 936 | 5 | 5 | |
| Guanaco-13BJudge=GPT-4, # Prompts=9532023.05 | 885 | 6 | 6 | |
| Guanaco-7BJudge=GPT-4, # Prompts=9532023.05 | 860 | 7 | 7 |