Multilinguality on TydiQA (F1)
70.8F1 ScoreGPT-4
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4Shots=1-shot, Config=GP2023.06 | 70.8 | |
| TÜLU-1.1 13BShots=1-shot, Config=GP2023.06 | 58.8 | |
| Human mix. 65BShots=1-shot, Config=GP2023.06 | 58.3 | |
| LLaMa 65BShots=1-shot, Config=GP2023.06 | 56.8 | |
| TÜLU 65BShots=1-shot, Config=GP2023.06 | 56.6 | |
| LLaMa-2 13BShots=1-shot, Config=GP2023.06 | 56.5 | |
| LLaMa 30BShots=1-shot, Config=GP2023.06 | 55.3 | |
| TÜLU-1.1 7BShots=1-shot, Config=GP2023.06 | 52.8 | |
| TÜLU 30BShots=1-shot, Config=GP2023.06 | 51.9 | |
| ChatGPTShots=1-shot, Config=GP2023.06 | 51.9 | |
| LLaMa-2 7BShots=1-shot, Config=GP2023.06 | 51.2 | |
| TÜLU 13BShots=1-shot, Config=GP2023.06 | 45.6 | |
| TÜLU 7BShots=1-shot, Config=GP2023.06 | 43.5 | |
| LLaMa 13BShots=1-shot, Config=GP2023.06 | 43.2 | |
| LLaMa 7BShots=1-shot, Config=GP2023.06 | 38.4 | |
| ShareGPT 65BShots=1-shot, Config=GP2023.06 | 31.6 |