Selective Generation on MMLU (PRR Accuracy)
81.6PRR AccuracySATRMD+MSP
Evaluation Results
| Method | Links | |
|---|---|---|
| SATRMD+MSPModel=Llama 8b v3.12025.02 | 81.6 | |
| FactoscopeModel=Llama 8b v3.12025.02 | 72.7 | |
| HUQ-SATRMDModel=Llama 8b v3.12025.02 | 60.9 | |
| SAPLMAModel=Llama 8b v3.12025.02 | 48.1 | |
| Maximum Sequence ProbabilityModel=Llama 8b v3.12025.02 | 40.5 | |
| EigValLaplacian NLI Score Entail.Model=Llama 8b v3.12025.02 | 34.4 | |
| SentenceSARModel=Llama 8b v3.12025.02 | 34.3 | |
| Eccentricity NLI Score Entail.Model=Llama 8b v3.12025.02 | 31.2 | |
| Lexical Similarity ROUGE-LModel=Llama 8b v3.12025.02 | 31.1 | |
| PerplexityModel=Llama 8b v3.12025.02 | 30.8 | |
| SARModel=Llama 8b v3.12025.02 | 28.4 | |
| DegMat NLI Score Entail.Model=Llama 8b v3.12025.02 | 22.4 | |
| Semantic EntropyModel=Llama 8b v3.12025.02 | 22 | |
| EigenScoreModel=Llama 8b v3.12025.02 | 19.6 |