Evidence-Based QA on CHATREPORT (test)
61.22Attributability ScoreGPT-4
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4Evaluation Mode=Zero-shot2024.02 | 61.22 | |
| GPT-3.5Evaluation Mode=Zero-shot2024.02 | 46.73 | |
| Llama-2-13b-chatEvaluation Mode=Zero-shot2024.02 | 10.54 | |
| Zephyr-7b-βEvaluation Mode=Zero-shot2024.02 | 5.22 |