Skip to content

RAG Systems and its Variations

RAGSystem

Bases: LlamaIndexBridgeMixin, _RAGSystem

RAG System with all available bridge functionality.

The RAGSystem is the main entry point for creating and managing retrieval-augmented generation systems.

Source code in src/fed_rag/core/rag_system/synchronous.py
class RAGSystem(LlamaIndexBridgeMixin, _RAGSystem):
    """RAG System with all available bridge functionality.

    The RAGSystem is the main entry point for creating and managing
    retrieval-augmented generation systems.
    """

    pass

AsyncRAGSystem

Bases: LlamaIndexBridgeMixin, _AsyncRAGSystem

Async RAG System with all available bridge functionality.

The RAGSystem is the main entry point for creating and managing retrieval-augmented generation systems.

Source code in src/fed_rag/core/rag_system/asynchronous.py
class AsyncRAGSystem(LlamaIndexBridgeMixin, _AsyncRAGSystem):
    """Async RAG System with all available bridge functionality.

    The RAGSystem is the main entry point for creating and managing
    retrieval-augmented generation systems.
    """

    pass

NoEncodeRAGSystem

Bases: _NoEncodeRAGSystem

NoEncode RAG System with all available bridge functionality.

The NoEncodeRAGSystem is the main entry point for creating and managing retrieval-augmented generation systems that skip encoding altogether, enabling direct natural language queries to knowledge sources like MCP servers, APIs, and databases.

Unlike traditional RAG systems that require separate retriever components and pre-computed embeddings, NoEncode RAG systems perform direct queries against NoEncode knowledge sources.

Source code in src/fed_rag/core/no_encode_rag_system/synchronous.py
class NoEncodeRAGSystem(_NoEncodeRAGSystem):
    """NoEncode RAG System with all available bridge functionality.

    The NoEncodeRAGSystem is the main entry point for creating and managing
    retrieval-augmented generation systems that skip encoding altogether,
    enabling direct natural language queries to knowledge sources like MCP
    servers, APIs, and databases.

    Unlike traditional RAG systems that require separate retriever components
    and pre-computed embeddings, NoEncode RAG systems perform direct queries
    against NoEncode knowledge sources.
    """

    pass

AsyncNoEncodeRAGSystem

Bases: _AsyncNoEncodeRAGSystem

Async NoEncode RAG System with all available bridge functionality.

The AsyncNoEncodeRAGSystem is the main entry point for creating and managing retrieval-augmented generation systems that skip encoding altogether, enabling direct natural language queries to knowledge sources like MCP servers, APIs, and databases.

Unlike traditional RAG systems that require separate retriever components and pre-computed embeddings, NoEncode RAG systems perform direct queries against NoEncode knowledge sources.

Source code in src/fed_rag/core/no_encode_rag_system/asynchronous.py
class AsyncNoEncodeRAGSystem(_AsyncNoEncodeRAGSystem):
    """Async NoEncode RAG System with all available bridge functionality.

    The AsyncNoEncodeRAGSystem is the main entry point for creating and managing
    retrieval-augmented generation systems that skip encoding altogether,
    enabling direct natural language queries to knowledge sources like MCP
    servers, APIs, and databases.

    Unlike traditional RAG systems that require separate retriever components
    and pre-computed embeddings, NoEncode RAG systems perform direct queries
    against NoEncode knowledge sources.
    """

    pass

Auxiliary types for RAG System

RAGConfig

Bases: BaseModel

Source code in src/fed_rag/data_structures/rag.py
class RAGConfig(BaseModel):
    top_k: int
    context_separator: str = "\n"