Quantization settings

target_border

Command-line: --target-border

Description

If set, defines the border for converting target values to 0 and 1.

Depending on the specified value:

• $target\_value \le border\_value$ the target is converted to 0
• $target\_value > border\_value$ the target is converted to 1

Type

float

Default value

Supported processing units

CPU and GPU

border_count

Command-line: -x, --border-count

Alias: max_bin

Description

The number of splits for numerical features. Allowed values are integers from 1 to 65535 inclusively.

Type

int

Default value

The default value depends on the processing unit type and other parameters:

• CPU: 254
• GPU in PairLogitPairwise and YetiRankPairwise modes: 32
• GPU in all other modes: 128

Supported processing units

CPU and GPU

feature_border_type

Command-line: --feature-border-type

Description

The quantization mode for numerical features.

Possible values:

• Median
• Uniform
• UniformAndQuantiles
• MaxLogSum
• MinEntropy
• GreedyLogSum

Type

string

Default value

GreedyLogSum

Supported processing units

CPU and GPU

per_float_feature_quantization

Command-line: --per-float-feature-quantization

Description

The quantization description for the specified feature or list of features.

Description format for a single feature:

FeatureId[:border_count=BorderCount][:nan_mode=BorderType][:border_type=border_selection_method]

Examples:

•   per_float_feature_quantization='0:border_count=1024'
  

  In this example, the feature indexed 0 has 1024 borders.

•   per_float_feature_quantization=['0:border_count=1024', '1:border_count=1024']
  

  In this example, features indexed 0 and 1 have 1024 borders.

Type

• string
• list of strings

Default value

Supported processing units

CPU and GPU