Tool usage in multi-turn dialogue on ChatHuman 1.0 (test)
100Success Rate (Args)ChatHuman
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ChatHuman2024.05 | 100 | 95.9 | 92.7 | 95.5 | 96.2 | |
| Visual ChatGPT-4Base Model=GPT-42024.05 | 86 | 79.4 | 71.1 | 74.4 | 78.9 | |
| GPT4Tools2024.05 | 58.2 | 55.1 | 55.3 | 51.3 | 61.2 | |
| Visual ChatGPT-3.5Base Model=GPT-3.52024.05 | 43.8 | 20.3 | 16.2 | 17.3 | 69.1 |