Parameters for constructing histogram index batches. More...

#include <data.h>

Public Member Functions
	BatchParam ()=default
	Exact or others that don't need histogram.

	BatchParam (bst_bin_t max_bin, double sparse_thresh)
	Used by the hist tree method.

	BatchParam (bst_bin_t max_bin, common::Span< float const > hessian, bool regenerate)
	Used by the approx tree method.

bool	ParamNotEqual (BatchParam const &other) const

bool	Initialized () const

BatchParam	MakeCache () const
	Make a copy of self for DMatrix to describe how its existing index was generated.

Data Fields
bst_bin_t	max_bin {0}
	Maximum number of bins per feature for histograms.

common::Span< float const >	hess
	Hessian, used for sketching with future approx implementation.

bool	regen {false}
	Whether should we force DMatrix to regenerate the batch.

bool	forbid_regen {false}
	Forbid regenerating the gradient index.

double	sparse_thresh {std::numeric_limits<double>::quiet_NaN()}
	Parameter used to generate column matrix for hist.

Detailed Description

Parameters for constructing histogram index batches.

Constructor & Destructor Documentation

xgboost::BatchParam::BatchParam	(	bst_bin_t	max_bin,
		common::Span< float const >	hessian,
		bool	regenerate
	)

inline

Used by the approx tree method.

Get batch with sketch weighted by hessian. The batch will be regenerated if the span is changed, so caller should keep the span for each iteration.

bool xgboost::BatchParam::forbid_regen {false}

Forbid regenerating the gradient index.

Used for internal validation.

bool xgboost::BatchParam::regen {false}

Whether should we force DMatrix to regenerate the batch.

Only used for GHistIndex.

The documentation for this struct was generated from the following file: