Package dev.langchain4j.adaptiverag

Class HallucinationGrader

java.lang.Object

dev.langchain4j.adaptiverag.HallucinationGrader

All Implemented Interfaces:

Function<HallucinationGrader.Arguments,HallucinationGrader.Score>

public class HallucinationGrader
extends Object
implements Function<HallucinationGrader.Arguments,HallucinationGrader.Score>

Provides functionality to grade the groundedness of an LLM generation with respect to a set of facts. The class implements the Function interface, which accepts parameters of type HallucinationGrader.Arguments and returns a score indicating whether the generation is grounded or not.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	HallucinationGrader.Arguments	Represents a set of facts and an associated LLM generation.
static class	HallucinationGrader.Score	Binary score for hallucination present in generation answer.

Constructor Summary

Constructors

Constructor	Description
HallucinationGrader()

Method Summary

Modifier and Type	Method	Description
HallucinationGrader.Score	apply(HallucinationGrader.Arguments args)	Applies the provided arguments to a chat language model and returns the score.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface java.util.function.Function

andThen, compose

Constructor Details

HallucinationGrader

public HallucinationGrader()

Method Details

apply

public HallucinationGrader.Score apply(HallucinationGrader.Arguments args)

Applies the provided arguments to a chat language model and returns the score. Uses OpenAI's GPT-3.5-Turbo model with specified settings. Logs requests and responses, retries twice, and has a max token limit of 2000.

Specified by:

apply in interface Function<HallucinationGrader.Arguments,HallucinationGrader.Score>

Parameters:

args - The arguments to be applied.

Returns:

A Score object from the AI grader's invocation.