Spaces:
Running on Zero
Running on Zero
| """Retrieval module for extensible search strategies. | |
| This module provides a pluggable architecture for different retrieval strategies: | |
| - vector: Semantic similarity search using embeddings (ChromaDB) | |
| - bm25: Lexical/keyword search using BM25 algorithm | |
| - bm25_vector: Hybrid search combining BM25 and vector search with fusion | |
| - page_index: (future) Vision-based document understanding (ColPali) | |
| - graph_vector: (future) Knowledge graph enhanced retrieval | |
| """ | |
| from .base import BaseRetrieverStrategy | |
| from .factory import RetrieverFactory | |
| from .fusion import FusionRetriever, reciprocal_rank_fusion | |
| __all__ = [ | |
| "BaseRetrieverStrategy", | |
| "RetrieverFactory", | |
| "FusionRetriever", | |
| "reciprocal_rank_fusion", | |
| ] | |