Toro

RankingConfig

Common LightGBM ranking settings.

Record fields

NameDescription
LearningRateShrinkage rate, or None to use ML.NET auto-tuning.
MinimumExampleCountPerLeafMinimum examples per leaf, or None to use ML.NET auto-tuning.
NumberOfIterationsNumber of boosting iterations.
NumberOfLeavesMaximum number of leaves, or None to use ML.NET auto-tuning.
SeedRandom seed. The default is Some 0 for reproducibility.

RankingConfig module

Constructors for LightGBM ranking settings.

FunctionDescription
create ()Create settings from the ML.NET LightGBM defaults with a deterministic seed.

create

create ()

Create settings from the ML.NET LightGBM defaults with a deterministic seed.

Parameters

  • () : unit

Returns RankingConfig

RankingModel

A fitted LightGBM ranking model.

Instance members

NameDescription
this.FeatureCountNumber of input features expected by the model.
this.TransformerUnderlying ML.NET model.

Ranking

LightGBM learning-to-rank operations.

FunctionDescription
evaluate dataset modelEvaluate a fitted ranker with ML.NET ranking metrics.
fit config datasetFit a LightGBM ranker with common F# settings.
fitWithOptions seed configure datasetFit a LightGBM ranker after configuring the complete ML.NET options object.
load pathLoad a LightGBM ranking model from the ML.NET zip format.
predict features modelScore a feature matrix. Scores are returned as a detached CPU float32 tensor.
save path modelSave a fitted model in the ML.NET zip format.

evaluate

evaluate dataset model

Evaluate a fitted ranker with ML.NET ranking metrics.

Parameters

  • dataset : RankingDataset
  • model : RankingModel

Returns RankingMetrics

fit

fit config dataset

Fit a LightGBM ranker with common F# settings.

Parameters

  • config : RankingConfig
  • dataset : RankingDataset

Returns RankingModel

fitWithOptions

fitWithOptions seed configure dataset

Fit a LightGBM ranker after configuring the complete ML.NET options object.

Parameters

  • seed : int option
  • configure : Options -> unit
  • dataset : RankingDataset

Returns RankingModel

load

load path

Load a LightGBM ranking model from the ML.NET zip format.

Parameters

  • path : string

Returns RankingModel

predict

predict features model

Score a feature matrix. Scores are returned as a detached CPU float32 tensor.

Parameters

  • features : Tensor
  • model : RankingModel

Returns Tensor

save

save path model

Save a fitted model in the ML.NET zip format.

Parameters

  • path : string
  • model : RankingModel