Infinity in LogLossReduction metric on classification models?

Question

Hello! I'm training classification models in c# with ML.NET, but always I have infinity in LogLossReduction metric across different algorithms. Below you will find the metrics I have per algorithm, any idea how to fix this? ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.NaiveBayesMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 34.53877639128314 LogLossReduction: ∞ TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.LbfgsMaximumEntropyMulticlassTrainer::: MacroAccuracy: 1 MicroAccuracy: 1 LogLoss: 0.6302328855811895 LogLossReduction: ∞ TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.SdcaNonCalibratedMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 0 LogLossReduction: NaN TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.SdcaMaximumEntropyMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 1.2416904381844802 LogLossReduction: ∞ TopKAccuracy: 0===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.NaiveBayesMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 34.53877639128314 LogLossReduction: ∞ TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.LbfgsMaximumEntropyMulticlassTrainer::: MacroAccuracy: 1 MicroAccuracy: 1 LogLoss: 0.6302328855811895 LogLossReduction: ∞ TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.SdcaNonCalibratedMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 0 LogLossReduction: NaN TopKAccuracy: 0 ===== Evaluating Model's accuracy with Test data ===== Microsoft.ML.Trainers.SdcaMaximumEntropyMulticlassTrainer::: MacroAccuracy: 0.7758620689655172 MicroAccuracy: 0.7758620689655172 LogLoss: 1.2416904381844802 LogLossReduction: ∞ TopKAccuracy: 0 Thank you, Heidi