Factuality Evaluation on TruthfulQA (0-shot)
64.3Factuality Score (0-shot)AquilaChat2
Evaluation Results
| Method | Links | |
|---|---|---|
| AquilaChat2Size=34B2024.03 | 64.3 | |
| Yi-ChatSize=34B2024.03 | 62.4 | |
| Yi-Chat-8bits(GPTQ)Size=34B2024.03 | 61.8 | |
| Yi-Chat-4bits(AWQ)Size=34B2024.03 | 61.8 | |
| LLaMA2-ChatSize=70B2024.03 | 54 | |
| Qwen-ChatSize=14B2024.03 | 52.5 | |
| InternLM-ChatSize=20B2024.03 | 51.8 | |
| Yi-ChatSize=6B2024.03 | 50.6 | |
| Yi-Chat-4bits(AWQ)Size=6B2024.03 | 50.3 | |
| Yi-Chat-8bits(GPTQ)Size=6B2024.03 | 49.9 | |
| Baichuan2-ChatSize=13B2024.03 | 49 | |
| LLaMA2-ChatSize=13B2024.03 | 36.8 |