Implementation:Explodinggradients Ragas Collections ExampleMetric Class

Field	Value
source	Repo
domains	Metrics, Framework
last_updated	2026-02-10 00:00 GMT

Overview

ExampleMetric is a template v2 metric class that demonstrates how to create new metrics by extending BaseMetric with minimal boilerplate.

Description

ExampleMetric extends BaseMetric to serve as a reference implementation and developer guide for creating new v2 collection metrics. It illustrates the minimal pattern required:

1. Inherit from BaseMetric.
2. Define an __init__ method that calls super().__init__(name=...).
3. Override async def ascore(**kwargs) -> MetricResult with custom scoring logic.

The example scoring logic is intentionally trivial: it returns the length of the response divided by 100, capped at 1.0. This placeholder demonstrates the return type and value range expectations without requiring any external dependencies, LLM, or embeddings.

By inheriting from BaseMetric, ExampleMetric automatically gains batch processing (abatch_score), sync wrappers (score, batch_score), numeric validation, and type safety.

Usage

This class is primarily intended as a template for developers. Instantiate directly for testing, or copy and modify for creating new production metrics.

Code Reference

Property	Value
Source Location	src/ragas/metrics/collections/example_metric.py L1--47
Signature	class ExampleMetric(BaseMetric)
Import	from ragas.metrics.collections.example_metric import ExampleMetric

I/O Contract

Inputs

Parameter	Type	Required	Description
user_input	str	Yes	The original question (unused in example logic)
response	str	Yes	The response text to evaluate

Constructor Parameters

Parameter	Type	Default	Description
name	str	"example_metric"	Metric name

Outputs

Field	Type	Description
MetricResult.value	float	Score in range 0.0--1.0 (response length / 100, capped at 1.0)

Usage Examples

from ragas.metrics.collections.example_metric import ExampleMetric

# Basic usage
metric = ExampleMetric()
result = await metric.ascore(
    user_input="What is Python?",
    response="Python is a programming language."
)
print(f"Example Score: {result.value}")  # len("Python is...") / 100

# Sync usage
result = metric.score(
    user_input="Hello",
    response="A" * 150  # Longer than 100 chars
)
print(f"Score (capped): {result.value}")  # 1.0

# Batch evaluation
results = await metric.abatch_score([
    {"user_input": "Q1", "response": "Short"},
    {"user_input": "Q2", "response": "A much longer response text here"},
])
for r in results:
    print(r.value)

# Template for creating your own metric:
from ragas.metrics.collections.base import BaseMetric
from ragas.metrics.result import MetricResult

class MyMetric(BaseMetric):
    def __init__(self, name: str = "my_metric", **kwargs):
        super().__init__(name=name, **kwargs)

    async def ascore(self, user_input: str, response: str) -> MetricResult:
        # Your custom scoring logic here
        score = 0.5  # Replace with actual computation
        return MetricResult(value=score)

Related Pages

• Explodinggradients_Ragas_Collections_BaseMetric_Class -- Base class that ExampleMetric extends
• Explodinggradients_Ragas_Collections_BleuScore_Metric -- Real non-LLM metric following the same pattern
• Explodinggradients_Ragas_Collections_AnswerRelevancy_Metric -- Real LLM-based metric following the same pattern