Quality Estimation on GSM8K (test)
0.642BCE LossSigmoid Head
Evaluation Results
| Method | Links | |
|---|---|---|
| Sigmoid HeadEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 0.642 | |
| Boosted ProbEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 0.843 | |
| Softmax HeadEvaluation Mode=Eval Self, Backbone Model=Olmo, Language=en2026.01 | 12.802 |