Share via


Microsoft.Extensions.AI.Evaluation Namespace

Contains types and abstractions for evaluating AI-generated responses, including metrics, evaluators, and extensions for interpreting and processing evaluation results.

Classes

BooleanMetric

An EvaluationMetric containing a Boolean value that can be used to represent an outcome that can have one of two possible values (such as yes v/s no, or pass v/s fail).

ChatConfiguration

Specifies the IChatClient that should be used when evaluation is performed using an AI model.

ChatMessageExtensions

Extension methods for ChatMessage.

ChatResponseExtensions

Extension methods for ChatResponse.

CompositeEvaluator

An IEvaluator that composes other IEvaluators to execute multiple (concurrent) evaluations on a supplied response.

EvaluationContext

An abstract base class that models additional contextual information (beyond that which is available in the conversation history) or other data that an IEvaluator may need to accurately evaluate supplied responses.

EvaluationDiagnostic

Represents a diagnostic (such as a warning, error or informational message) that applies to the result represented in an EvaluationMetric.

EvaluationMetric

A base class that represents the result of an evaluation.

EvaluationMetric<T>

An base class that represents the result of an evaluation containing a value of type T.

EvaluationMetricExtensions

Extension methods for EvaluationMetric.

EvaluationMetricInterpretation

Specifies how the result represented in an associated EvaluationMetric should be interpreted.

EvaluationResult

A collection of one or more EvaluationMetrics that represent the result of an evaluation.

EvaluationResultExtensions

Extension methods for EvaluationResult.

EvaluatorExtensions

Extension methods for IEvaluator.

NumericMetric

An EvaluationMetric containing a numeric value.

StringMetric

An EvaluationMetric containing a String value.

Interfaces

IEvaluator

Evaluates responses produced by an AI model.

Enums

EvaluationDiagnosticSeverity

An enumeration that identifies the set of possible values for Severity.

EvaluationRating

An enumeration that identifies the set of possible ways in which an EvaluationMetric can be interpreted.