Question Answering on TriviaQA (PR-AUC)
0.766PR-AUCDissimilarity
Evaluation Results
| Method | Links | |
|---|---|---|
| DissimilarityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.766 | |
| DissimilarityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.755 | |
| EigVecDissimilarityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.753 | |
| CocoaPPLStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.748 | |
| CocoaMSPStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.747 | |
| EccentricityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.739 | |
| CocoaPPLStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.739 | |
| EigVecDissimilarityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.738 | |
| CocoaMSPStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.738 | |
| EccentricityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.714 | |
| Semantic EntropyStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.685 | |
| Degree MatrixStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.682 | |
| Degree MatrixStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.673 | |
| SARStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 0.671 | |
| SARStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.656 | |
| Semantic EntropyStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 0.622 |