Current API Reference

Top-level package

mandol.__all__ currently exports:

  • MemoryUnit

  • MemorySpace

  • MemorySpaceRegistry

  • TowerSpace

  • SemanticMap

  • SemanticGraph

  • cluster_nodes

  • ClusterMethod

  • __version__

Core helpers

mandol.core also provides:

  • create_semantic_map(text_model=None, image_model=None, embedding_dim=None, preset=None, **kwargs)

  • create_semantic_graph(semantic_map_instance=None, preset=None, **kwargs)

  • create_memory_unit(uid, content, content_type="text", metadata=None, **kwargs)

  • create_memory_space(name, faiss_index_type=None)

  • get_default_core_config()

  • get_model_preset(preset_name)

  • get_core_component_status()

  • validate_embedding_dim(embedding_dim)

  • get_recommended_faiss_index(num_vectors, embedding_dim)

Retrieval package

mandol.retrieval exports:

  • BaseRetriever

  • RetrievalInterface

  • MultiRetrievalInterface

  • RetrievalMethod

  • RetrievalResult

  • parse_retrieval_methods

  • parse_weights

  • MultiRetriever

  • BM25Retriever

  • SPLADERetriever

  • CosineRetrieverAdapter

  • GraphContext

  • PathInfo

  • GraphContextExpander

  • RerankerManager

  • QueryBundle

  • ScoreFusion

Three-tower retrieval

mandol.triple_retrieval exports:

  • TripleTowerRetriever

  • TripleTowerConfig

  • TripleTowerSearchMode

  • RerankStrategy

  • TripleTowerResult

  • HierarchicalTowerResult

  • GraphTowerResult

  • EpisodicTowerResult

  • SecondStageRerankResult

  • TowerRetrievalStats

  • create_triple_tower_retriever

Other retrieval-facing packages

mandol.hierarchical exposes hierarchical retrieval types and clustering helpers for already-built hierarchical memory spaces.

mandol.entity_relation exposes EntityRelationGraphRetriever, GraphRetrievalConfig and create_graph_retriever.

mandol.episodic exposes EpisodicMemoryRetriever, EpisodicRetrievalConfig and create_episodic_retriever.

High-level memory construction

mandol.auto_builder exposes the maintained high-level construction path:

  • MemoryOrchestrator and OrchestratorConfig

  • HighLevelMemoryBuilder and HighLevelMemoryBuildConfig

  • build_high_level_memory

  • HierarchicalAutoBuilder, EntityRelationAutoBuilder and EpisodicAutoBuilder with their configuration and result types

  • PipelineStrategy and STYLE_STRATEGIES

Routing and quantification

mandol.memory_router exposes LocomoTowerRouter, LongMemEvalTowerRouter and their routing configuration and category types. These policies are used by the paper’s router + quantification workflows.

mandol.quantification exposes SemanticQuantifier, QueryExpander, confidence-aware pruning, cascade pruning and the associated factory functions.

Providers, storage and clustering

mandol.llm provides LLMClient, LocalLLMClient, provider/model registries and client factories. mandol.storage provides RocksDBPayloadStore and TieredStorageManager. mandol.cluster provides the Leiden and DBSCAN clusterer interfaces used by the builders.