Question Answering on HotpotQA (Helpfulness, Fluency, Accuracy)
4.82HelpfulnessClaude
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Claude2024.08 | 4.82 | 4.99 | 1.26 | 58 | |
| GPT42024.08 | 4.78 | 4.96 | 1.12 | 66 | |
| TextDavinci2024.08 | 4.72 | 4.87 | 1.22 | 45 | |
| GPT3.52024.08 | 4.72 | 4.95 | 1.49 | 63 | |
| TextBabbage2024.08 | 4.7 | 4.88 | 1.74 | 37 | |
| Llama2checkpoint=Llama-2-7B2024.08 | 4.7 | 4.95 | 1.32 | 55 | |
| Zephyrcheckpoint=Zephyr-alpha2024.08 | 4.64 | 4.88 | 1.01 | 40 | |
| Davinci2024.08 | 4.27 | 4.52 | 1.68 | 32 |