Factual Hallucination on TruthfulQA
41.74MC1 ScoreDeLask
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DeLaskModel=LLaMA3-8B, Decoding=DeLask2026.05 | 41.74 | 58.76 | |
| RegularModel=LLaMA3-8B, Decoding=Regular2026.05 | 39.72 | 57.1 | |
| UADModel=LLaMA3-8B, Decoding=UAD2026.05 | 39.59 | 57.44 | |
| DOLAModel=LLaMA3-8B, Decoding=DOLA2026.05 | 38.43 | 56.38 | |
| ITIModel=LLaMA3-8B, Decoding=ITI2026.05 | 37.14 | 57.05 | |
| ADModel=LLaMA3-8B, Decoding=AD2026.05 | 35.45 | 55.74 | |
| DeLaskModel=LLaMA2-7B, Decoding=DeLask2026.05 | 35.02 | 52.27 | |
| ADModel=LLaMA2-7B, Decoding=AD2026.05 | 34.14 | 52.03 | |
| DOLAModel=LLaMA2-7B, Decoding=DOLA2026.05 | 33.71 | 53.17 | |
| RegularModel=LLaMA2-7B, Decoding=Regular2026.05 | 33.51 | 51.12 | |
| ITIModel=LLaMA2-7B, Decoding=ITI2026.05 | 33.39 | 52.03 | |
| UADModel=LLaMA2-7B, Decoding=UAD2026.05 | 32.01 | 52.47 |