Uncertainty Quantification on Average of 6 Datasets
65PRRDissimilarity + beamsearch
Evaluation Results
| Method | Links | |
|---|---|---|
| Dissimilarity + beamsearchModel=Gemma 3 4B, Model Type=base2025.12 | 65 | |
| CocoaPPL + beamsearchModel=Gemma 3 4B, Model Type=base2025.12 | 64.9 | |
| DissimilarityModel=Gemma 3 4B, Model Type=base2025.12 | 63 | |
| CocoaPPLModel=Gemma 3 4B, Model Type=base2025.12 | 62.8 | |
| CocoaMSP + beamsearchModel=Gemma 3 4B, Model Type=base2025.12 | 61.5 | |
| Eccentricity + beamsearchModel=Gemma 3 4B, Model Type=base2025.12 | 60.3 | |
| EigVecDissimilarity + beamsearchModel=Gemma 3 4B, Model Type=base2025.12 | 59.8 | |
| CocoaMSPModel=Gemma 3 4B, Model Type=base2025.12 | 58.7 | |
| EccentricityModel=Gemma 3 4B, Model Type=base2025.12 | 56.3 | |
| EigVecDissimilarityModel=Gemma 3 4B, Model Type=base2025.12 | 56.1 | |
| Dissimilarity + beamsearchModel=Llama 3.1 8B, Model Type=base2025.12 | 54.3 | |
| CocoaPPL + beamsearchModel=Llama 3.1 8B, Model Type=base2025.12 | 53.6 | |
| PerplexityModel=Gemma 3 4B, Model Type=base2025.12 | 52.5 | |
| CocoaPPLModel=Llama 3.1 8B, Model Type=base2025.12 | 52.3 | |
| CocoaMSP + beamsearchModel=Llama 3.1 8B, Model Type=base2025.12 | 52.1 | |
| EigVecDissimilarity + beamsearchModel=Llama 3.1 8B, Model Type=base2025.12 | 51 | |
| DissimilarityModel=Llama 3.1 8B, Model Type=base2025.12 | 50.5 | |
| Eccentricity + beamsearchModel=Llama 3.1 8B, Model Type=base2025.12 | 50.5 | |
| CocoaMSPModel=Llama 3.1 8B, Model Type=base2025.12 | 50.5 | |
| CCPModel=Gemma 3 4B, Model Type=base2025.12 | 49.2 | |
| Dissimilarity + beamsearchModel=Qwen 3 8B, Model Type=base2025.12 | 47.8 | |
| DissimilarityModel=Qwen 3 8B, Model Type=base2025.12 | 47.7 | |
| MTEModel=Gemma 3 4B, Model Type=base2025.12 | 47.6 | |
| CocoaMSP + beamsearchModel=Qwen 3 8B, Model Type=base2025.12 | 47.3 | |
| MSPModel=Gemma 3 4B, Model Type=base2025.12 | 47.1 | |
| EigVecDissimilarityModel=Llama 3.1 8B, Model Type=base2025.12 | 46.3 | |
| CocoaMSPModel=Qwen 3 8B, Model Type=base2025.12 | 46.1 | |
| CocoaPPLModel=Qwen 3 8B, Model Type=base2025.12 | 46.1 | |
| CocoaPPL + beamsearchModel=Qwen 3 8B, Model Type=base2025.12 | 46.1 | |
| EccentricityModel=Llama 3.1 8B, Model Type=base2025.12 | 45.3 | |
| PerplexityModel=Llama 3.1 8B, Model Type=base2025.12 | 45.2 | |
| EigVecDissimilarity + beamsearchModel=Qwen 3 8B, Model Type=base2025.12 | 45 | |
| EigValLaplacianModel=Gemma 3 4B, Model Type=base2025.12 | 43.7 | |
| Lexical SimilarityModel=Gemma 3 4B, Model Type=base2025.12 | 42.6 | |
| EigValLaplacianModel=Llama 3.1 8B, Model Type=base2025.12 | 42.6 | |
| CocoaMSP + beamsearchModel=Llama 3.1 8B, Model Type=instruct2025.12 | 42.6 | |
| Lexical SimilarityModel=Qwen 3 8B, Model Type=base2025.12 | 42.5 | |
| EigVecDissimilarityModel=Qwen 3 8B, Model Type=base2025.12 | 42.5 | |
| MTEModel=Llama 3.1 8B, Model Type=base2025.12 | 42.2 | |
| NumSemSetsModel=Gemma 3 4B, Model Type=base2025.12 | 41.8 | |
| Dissimilarity + beamsearchModel=Llama 3.1 8B, Model Type=instruct2025.12 | 41.7 | |
| Semantic EntropyModel=Llama 3.1 8B, Model Type=base2025.12 | 41.4 | |
| EigVecDissimilarity + beamsearchModel=Llama 3.1 8B, Model Type=instruct2025.12 | 41.4 | |
| CocoaPPL + beamsearchModel=Llama 3.1 8B, Model Type=instruct2025.12 | 41.2 | |
| Lexical SimilarityModel=Llama 3.1 8B, Model Type=base2025.12 | 41.1 | |
| MSPModel=Llama 3.1 8B, Model Type=base2025.12 | 41 | |
| Eccentricity + beamsearchModel=Qwen 3 8B, Model Type=base2025.12 | 41 | |
| MTEModel=Qwen 3 8B, Model Type=base2025.12 | 40.7 | |
| EigValLaplacianModel=Qwen 3 8B, Model Type=base2025.12 | 40.6 | |
| CocoaMSPModel=Llama 3.1 8B, Model Type=instruct2025.12 | 40.4 | |
| CCPModel=Llama 3.1 8B, Model Type=base2025.12 | 40.1 | |
| Semantic EntropyModel=Gemma 3 4B, Model Type=base2025.12 | 40.1 | |
| Eccentricity + beamsearchModel=Llama 3.1 8B, Model Type=instruct2025.12 | 39.7 | |
| CocoaPPLModel=Llama 3.1 8B, Model Type=instruct2025.12 | 39.7 | |
| NumSemSetsModel=Llama 3.1 8B, Model Type=base2025.12 | 39.6 | |
| EccentricityModel=Qwen 3 8B, Model Type=base2025.12 | 39.6 | |
| SARModel=Gemma 3 4B, Model Type=base2025.12 | 38.6 | |
| SARModel=Llama 3.1 8B, Model Type=instruct2025.12 | 38.5 | |
| DissimilarityModel=Llama 3.1 8B, Model Type=instruct2025.12 | 37.9 | |
| MSPModel=Qwen 3 8B, Model Type=base2025.12 | 37.6 | |
| Semantic EntropyModel=Llama 3.1 8B, Model Type=instruct2025.12 | 37.6 | |
| EigVecDissimilarity + beamsearchModel=Qwen 3 8B, Model Type=instruct2025.12 | 37.6 | |
| PerplexityModel=Qwen 3 8B, Model Type=base2025.12 | 37.2 | |
| EigValLaplacianModel=Llama 3.1 8B, Model Type=instruct2025.12 | 37.1 | |
| EigVecDissimilarityModel=Llama 3.1 8B, Model Type=instruct2025.12 | 37 | |
| EccentricityModel=Llama 3.1 8B, Model Type=instruct2025.12 | 36.8 | |
| Lexical SimilarityModel=Llama 3.1 8B, Model Type=instruct2025.12 | 36.6 | |
| NumSemSetsModel=Qwen 3 8B, Model Type=base2025.12 | 36.5 | |
| MTEModel=Llama 3.1 8B, Model Type=instruct2025.12 | 36.4 | |
| CCPModel=Llama 3.1 8B, Model Type=instruct2025.12 | 36.4 | |
| SARModel=Qwen 3 8B, Model Type=base2025.12 | 36.3 | |
| CCPModel=Qwen 3 8B, Model Type=base2025.12 | 35.5 | |
| Dissimilarity + beamsearchModel=Qwen 3 8B, Model Type=instruct2025.12 | 35.5 | |
| SARModel=Llama 3.1 8B, Model Type=base2025.12 | 35.2 | |
| CocoaMSP + beamsearchModel=Qwen 3 8B, Model Type=instruct2025.12 | 34.7 | |
| Eccentricity + beamsearchModel=Qwen 3 8B, Model Type=instruct2025.12 | 34.5 | |
| CocoaMSP + beamsearchModel=Gemma 3 4B, Model Type=instruct2025.12 | 34.5 | |
| MSPModel=Llama 3.1 8B, Model Type=instruct2025.12 | 34.4 | |
| CocoaPPL + beamsearchModel=Gemma 3 4B, Model Type=instruct2025.12 | 33.9 | |
| CocoaPPL + beamsearchModel=Qwen 3 8B, Model Type=instruct2025.12 | 33.7 | |
| CocoaMSPModel=Qwen 3 8B, Model Type=instruct2025.12 | 33.4 | |
| CCPModel=Gemma 3 4B, Model Type=instruct2025.12 | 33.1 | |
| DissimilarityModel=Qwen 3 8B, Model Type=instruct2025.12 | 32.7 | |
| CocoaPPLModel=Qwen 3 8B, Model Type=instruct2025.12 | 32.7 | |
| PerplexityModel=Llama 3.1 8B, Model Type=instruct2025.12 | 32.3 | |
| Semantic EntropyModel=Qwen 3 8B, Model Type=base2025.12 | 31.9 | |
| NumSemSetsModel=Llama 3.1 8B, Model Type=instruct2025.12 | 31.9 | |
| MTEModel=Gemma 3 4B, Model Type=instruct2025.12 | 31.7 | |
| CocoaMSPModel=Gemma 3 4B, Model Type=instruct2025.12 | 31.4 | |
| CocoaPPLModel=Gemma 3 4B, Model Type=instruct2025.12 | 31.2 | |
| EigVecDissimilarity + beamsearchModel=Gemma 3 4B, Model Type=instruct2025.12 | 30.1 | |
| Semantic EntropyModel=Qwen 3 8B, Model Type=instruct2025.12 | 29.9 | |
| MTEModel=Qwen 3 8B, Model Type=instruct2025.12 | 29.7 | |
| Semantic EntropyModel=Gemma 3 4B, Model Type=instruct2025.12 | 29.3 | |
| MSPModel=Gemma 3 4B, Model Type=instruct2025.12 | 29.2 | |
| SARModel=Qwen 3 8B, Model Type=instruct2025.12 | 29.2 | |
| CCPModel=Qwen 3 8B, Model Type=instruct2025.12 | 29.1 | |
| MSPModel=Qwen 3 8B, Model Type=instruct2025.12 | 28.9 | |
| PerplexityModel=Gemma 3 4B, Model Type=instruct2025.12 | 28.8 | |
| Eccentricity + beamsearchModel=Gemma 3 4B, Model Type=instruct2025.12 | 28.5 |