Quality Estimation on PAWS-X (test)
0.325BCESigmoid Head
Evaluation Results
| Method | Links | |
|---|---|---|
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=fr2026.01 | 0.325 | |
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=es2026.01 | 0.399 | |
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=de2026.01 | 0.422 | |
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=zh2026.01 | 0.622 | |
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 0.883 | |
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=en2026.01 | 1.056 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Tower, Language=fr2026.01 | 1.256 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Tower, Language=es2026.01 | 1.362 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Tower, Language=de2026.01 | 1.645 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=fr2026.01 | 1.849 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Tower, Language=en2026.01 | 1.925 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=zh2026.01 | 2.172 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=es2026.01 | 2.309 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Olmo, Language=en2026.01 | 2.832 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Tower, Language=zh2026.01 | 2.925 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=de2026.01 | 3.151 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 3.769 | |
| Sigmoid HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=en2026.01 | 4.546 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Tower, Language=fr2026.01 | 6.297 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Tower, Language=en2026.01 | 6.858 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Tower, Language=es2026.01 | 7.324 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Tower, Language=de2026.01 | 8.726 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=fr2026.01 | 8.847 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=de2026.01 | 9.025 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Olmo, Language=en2026.01 | 9.096 | |
| Boosted ProbEvaluation Mode=Eval Others, Backbone Model=Tower, Language=zh2026.01 | 9.257 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=es2026.01 | 9.334 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=fr2026.01 | 9.977 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=en2026.01 | 10.835 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=es2026.01 | 10.868 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=en2026.01 | 11.668 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=de2026.01 | 11.756 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Tower, Language=zh2026.01 | 12.362 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Tower, Language=zh2026.01 | 12.745 | |
| Softmax HeadEvaluation Mode=Eval Others, Backbone Model=Olmo, Language=en2026.01 | 13.364 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 14.684 |