Question Answering on TruthfulQA (vbal Metric)
0.685vbalBase
Evaluation Results
| Method | Links | |
|---|---|---|
| BaseModel=Tulu3.1-8B2026.04 | 0.685 | |
| BaseModel=OLMo2-7B2026.04 | 0.67 | |
| BaseModel=Llama3.1-8B2026.04 | 0.636 | |
| Oracle-rephraseModel=Llama3.1-8B2026.04 | 0.069 | |
| Oracle-rephraseModel=Tulu3.1-8B2026.04 | 0.038 | |
| SIMPL.Model=Llama3.1-8B2026.04 | 0.026 | |
| Oracle-rephraseModel=OLMo2-7B2026.04 | 0.026 | |
| STRUCT.Model=Llama3.1-8B2026.04 | 0.022 | |
| STRUCT.Model=Tulu3.1-8B2026.04 | 0.013 | |
| PROF.Model=Llama3.1-8B2026.04 | 0.009 | |
| Reflect-and-rephraseModel=Llama3.1-8B2026.04 | 0.007 | |
| SIMPL.Model=Tulu3.1-8B2026.04 | 0.002 | |
| PROF.Model=Tulu3.1-8B2026.04 | -0.004 | |
| PROF.Model=OLMo2-7B2026.04 | -0.011 | |
| SIMPL.Model=OLMo2-7B2026.04 | -0.011 | |
| STRUCT.Model=OLMo2-7B2026.04 | -0.014 | |
| Reflect-and-rephraseModel=OLMo2-7B2026.04 | -0.027 | |
| Reflect-and-rephraseModel=Tulu3.1-8B2026.04 | -0.069 |