SgdNonCalibratedTrainer.Options Class
Definition
Important
Some information relates to prerelease product that may be substantially modified before it’s released. Microsoft makes no warranties, express or implied, with respect to the information provided here.
Options for the SgdNonCalibratedTrainer as used in SgdNonCalibrated(Options).
public sealed class SgdNonCalibratedTrainer.Options : Microsoft.ML.Trainers.SgdBinaryTrainerBase<Microsoft.ML.Trainers.LinearBinaryModelParameters>.OptionsBase
type SgdNonCalibratedTrainer.Options = class
inherit SgdBinaryTrainerBase<LinearBinaryModelParameters>.OptionsBase
Public NotInheritable Class SgdNonCalibratedTrainer.Options
Inherits SgdBinaryTrainerBase(Of LinearBinaryModelParameters).OptionsBase
- Inheritance
-
SgdNonCalibratedTrainer.Options
Constructors
SgdNonCalibratedTrainer.Options() |
Fields
CheckFrequency |
Determines the frequency of checking for convergence in terms of number of iterations. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
ConvergenceTolerance |
The convergence tolerance. If the exponential moving average of loss reductions falls below this tolerance, the algorithm is deemed to have converged and will stop. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
ExampleWeightColumnName |
Column to use for example weight. (Inherited from TrainerInputBaseWithWeight) |
FeatureColumnName |
Column to use for features. (Inherited from TrainerInputBase) |
L2Regularization |
The L2 weight for regularization. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
LabelColumnName |
Column to use for labels. (Inherited from TrainerInputBaseWithLabel) |
LearningRate |
The initial learning rate used by SGD. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
LossFunction |
The loss function to use. Default is LogLoss. |
NumberOfIterations |
The maximum number of passes through the training dataset. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
NumberOfThreads |
The degree of lock-free parallelism used by SGD. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
PositiveInstanceWeight |
The weight to be applied to the positive class. This is useful for training with imbalanced data. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |
Shuffle |
Determines whether to shuffle data for each training iteration. (Inherited from SgdBinaryTrainerBase<TModel>.OptionsBase) |