Commonsense Reasoning on OpenBookQA (test)
66.17AUC (%)MAXIMUM ENTROPY w/ STAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MAXIMUM ENTROPY w/ STARBase Model=LLaMA2-7B2024.03 | 66.17 | 7.47 | |
| PREDICTIVE ENTROPY w/ STARBase Model=LLaMA2-7B2024.03 | 64.86 | 3.88 | |
| RANDOMBase Model=LLaMA2-7B2024.03 | 63.44 | — | |
| MAXIMUM ENTROPYBase Model=LLaMA2-7B2024.03 | 63.36 | -0.22 | |
| PREDICTIVE ENTROPYBase Model=LLaMA2-7B2024.03 | 63.05 | -1.07 |