agents.rag.multi_agent_rag.grading_components

Grading Components for RAG Workflows.

This module provides reusable grading agents for document relevance, answer quality, and hallucination detection.

Classes

AnswerGrade

Grade for generated answer quality.

CompositeGradingAgent

Combines multiple grading components for comprehensive evaluation.

DocumentGrade

Grade for a retrieved document.

HallucinationGrade

Grade for hallucination detection.

Functions

create_answer_grader([name])

Create an answer quality grading agent.

create_document_grader([name])

Create a document relevance grading agent.

create_hallucination_grader([name])

Create a hallucination detection agent.

create_priority_ranker([name])

Create a document priority ranking agent.

create_query_analyzer([name])

Create a query analysis agent.

Module Contents

class agents.rag.multi_agent_rag.grading_components.AnswerGrade(/, **data)

Bases: pydantic.BaseModel

Grade for generated answer quality.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:

data (Any)

class agents.rag.multi_agent_rag.grading_components.CompositeGradingAgent

Combines multiple grading components for comprehensive evaluation.

Init .

Returns:

Add return description]

Return type:

[TODO

async grade_rag_pipeline(query, documents, answer)

Perform comprehensive grading of entire RAG pipeline.

Parameters:
Return type:

dict[str, Any]

class agents.rag.multi_agent_rag.grading_components.DocumentGrade(/, **data)

Bases: pydantic.BaseModel

Grade for a retrieved document.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:

data (Any)

class agents.rag.multi_agent_rag.grading_components.HallucinationGrade(/, **data)

Bases: pydantic.BaseModel

Grade for hallucination detection.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:

data (Any)

agents.rag.multi_agent_rag.grading_components.create_answer_grader(name='answer_grader')

Create an answer quality grading agent.

Parameters:

name (str)

Return type:

haive.agents.simple.SimpleAgent

agents.rag.multi_agent_rag.grading_components.create_document_grader(name='document_grader')

Create a document relevance grading agent.

Parameters:

name (str)

Return type:

haive.agents.simple.SimpleAgent

agents.rag.multi_agent_rag.grading_components.create_hallucination_grader(name='hallucination_grader')

Create a hallucination detection agent.

Parameters:

name (str)

Return type:

haive.agents.simple.SimpleAgent

agents.rag.multi_agent_rag.grading_components.create_priority_ranker(name='priority_ranker')

Create a document priority ranking agent.

Parameters:

name (str)

Return type:

haive.agents.simple.SimpleAgent

agents.rag.multi_agent_rag.grading_components.create_query_analyzer(name='query_analyzer')

Create a query analysis agent.

Parameters:

name (str)

Return type:

haive.agents.simple.SimpleAgent