Question Answering on MMLU (Vanilla and Absolute Accuracy)
65.98Vanilla AccuracyGPT-3.5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-3.5Size=>20B2024.03 | 65.98 | 52.09 | |
| MeanLearnSize=8B2024.03 | 62.35 | 46.76 | |
| LLaMA-3Size=8B2024.03 | 61.86 | 47.31 | |
| PaLM-2Size=>20B2024.03 | 58.73 | 44.39 | |
| MeanLearnSize=7B2024.03 | 52.98 | 37.24 | |
| LLaMA-2Size=13B2024.03 | 51.87 | 35.72 | |
| MeanLearnSize=13B2024.03 | 51.68 | 35.21 | |
| Orca-2Size=7B2024.03 | 50.11 | 33.68 | |
| VicunaSize=7B2024.03 | 49.4 | 32.5 | |
| VicunaSize=13B2024.03 | 49.34 | 32.21 | |
| WizardLMSize=13B2024.03 | 49.34 | 29.47 | |
| Orca-2Size=13B2024.03 | 47.27 | 31.11 | |
| LLaMA-2Size=7B2024.03 | 41.65 | 25.48 | |
| WizardLMSize=7B2024.03 | 23.26 | 11.55 |