Selective Generation on CNN (PRR Metrics)
0.15PRR (ROUGE-L)Perplexity
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PerplexityModel=Llama 8b v3.12025.02 | 0.15 | 0.242 | |
| SATRMD+MSPModel=Llama 8b v3.12025.02 | 0.135 | 0.016 | |
| SAPLMAModel=Llama 8b v3.12025.02 | 0.126 | 0.179 | |
| HUQ-SATRMDModel=Llama 8b v3.12025.02 | 0.126 | 0.048 | |
| Lexical Similarity ROUGE-LModel=Llama 8b v3.12025.02 | 0.123 | 0.122 | |
| SARModel=Llama 8b v3.12025.02 | 0.12 | 0.154 | |
| EigValLaplacian NLI Score Entail.Model=Llama 8b v3.12025.02 | 0.087 | 0.143 | |
| DegMat NLI Score Entail.Model=Llama 8b v3.12025.02 | 0.084 | 0.144 | |
| Semantic EntropyModel=Llama 8b v3.12025.02 | 0.077 | 0.096 | |
| SentenceSARModel=Llama 8b v3.12025.02 | 0.055 | 0.091 | |
| EigenScoreModel=Llama 8b v3.12025.02 | 0.04 | 0.045 | |
| Eccentricity NLI Score Entail.Model=Llama 8b v3.12025.02 | 0.028 | 0.05 | |
| Maximum Sequence ProbabilityModel=Llama 8b v3.12025.02 | 0.021 | 0.056 | |
| FactoscopeModel=Llama 8b v3.12025.02 | 0.007 | 0.004 |