Implementation:Explodinggradients Ragas MetricValidators Module

Field	Value
source	Explodinggradients_Ragas (GitHub)
domains	Metrics, Validation
last_updated	2026-02-10 00:00 GMT

Overview

The validators module provides mixin classes and helper functions for validating metric result values against type-specific constraints such as discrete allowed values, numeric ranges, and ranked list lengths.

Description

The module defines an abstract BaseValidator with a validate_result_value method, and three concrete validator mixins: DiscreteValidator checks that a result value is among a list of allowed strings; NumericValidator verifies that a numeric result falls within a tuple range or Python range object; RankingValidator ensures that a list result has the expected length specified by an integer. Two helper functions are provided: get_validator_for_allowed_values returns the appropriate validator class based on the type of the allowed_values parameter, and get_metric_type_name returns a human-readable metric type name string.

Usage

These validators are used as mixins in metric class hierarchies. The allowed_values attribute is inherited from the metric base class and determines which validator behavior applies. Call validate_result_value() on a metric instance to check whether a result is valid.

Code Reference

Item	Detail
Source Location	src/ragas/metrics/validators.py L19-125
Classes	BaseValidator(ABC), DiscreteValidator, NumericValidator, RankingValidator
Functions	get_validator_for_allowed_values(allowed_values) -> Type[BaseValidator], get_metric_type_name(allowed_values) -> str
Import	from ragas.metrics.validators import NumericValidator, DiscreteValidator

I/O Contract

Inputs

Parameter	Type	Description
result_value (DiscreteValidator)	Any	Value to check against allowed_values: List[str]
result_value (NumericValidator)	Any	Value to check against allowed_values: Tuple[float, float] or range
result_value (RankingValidator)	Any	Value to check against allowed_values: int (expected list length)
allowed_values (helper)	AllowedValuesType	Union of List[str], Tuple[float, float], range, or int

Outputs

Output	Type	Description
validate_result_value()	Optional[str]	Error message string if validation fails, None if valid
get_validator_for_allowed_values()	Type[BaseValidator]	The appropriate validator class
get_metric_type_name()	str	One of "DiscreteMetric", "NumericMetric", "RankingMetric", or "CustomMetric"

Usage Examples

from ragas.metrics.validators import (
    DiscreteValidator, NumericValidator, RankingValidator,
    get_validator_for_allowed_values, get_metric_type_name,
)

# Determine validator type from allowed_values
validator_cls = get_validator_for_allowed_values(["yes", "no"])
print(validator_cls)  # <class 'DiscreteValidator'>

print(get_metric_type_name((0.0, 1.0)))  # "NumericMetric"
print(get_metric_type_name(3))            # "RankingMetric"

Related Pages

• RankingMetric_Class - Uses RankingValidator as a mixin
• MetricResult_Class - Values validated by these validators
• Validation_Module - Dataset-level validation utilities