Question Answering on Web Questions (PR-AUC)
72.2PR-AUCDissimilarity
Evaluation Results
| Method | Links | |
|---|---|---|
| DissimilarityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 72.2 | |
| DissimilarityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 71.5 | |
| CocoaPPLStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 69.4 | |
| CocoaMSPStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 67.9 | |
| CocoaPPLStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 67.8 | |
| EigVecDissimilarityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 66.8 | |
| CocoaMSPStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 66.6 | |
| EigVecDissimilarityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 66.1 | |
| EccentricityStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 65.3 | |
| Degree MatrixStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 64.2 | |
| EccentricityStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 63.3 | |
| Semantic EntropyStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 61.4 | |
| Degree MatrixStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 60.5 | |
| SARStrategy=beam-search, Backbone=Gemma 3 4B base2025.12 | 58.9 | |
| SARStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 57.1 | |
| Semantic EntropyStrategy=multinomial-sampling, Backbone=Gemma 3 4B base2025.12 | 50.5 |