Expand description
Core types for vector storage.
This module provides the fundamental types for vector embedding storage:
§Entity-based embeddings
Embedding- A validated dense vector embedding with dimension checksSparseEmbedding- A sparse vector embedding (index, value pairs)MultiVectorEmbedding- A multi-vector embedding (ColBERT-style token embeddings)BinaryEmbedding- A binary vector embedding (bit-packed for Hamming distance)EmbeddingName- A named embedding space identifierEmbeddingSpace- Metadata about an embedding space (dimension, distance metric)SparseEmbeddingSpace- Metadata about a sparse embedding spaceMultiVectorEmbeddingSpace- Metadata about a multi-vector embedding spaceBinaryEmbeddingSpace- Metadata about a binary embedding space
§Separated vector storage
VectorData- Vector data supporting dense, sparse, multi-vector, and binary formatsVectorRef- Reference to a vector attached to an entity
§Point collections (Qdrant-style)
Collection- A named collection of points with a defined schemaCollectionName- A validated collection nameCollectionSchema- Schema defining the vectors in a collectionVectorConfig- Configuration for a named vector (type, dimension)VectorName- A validated vector namePayload- JSON payload attached to a pointNamedVector- A vector value (dense, sparse, or multi-vector)
Structs§
- Binary
Embedding - A binary vector embedding represented as bit-packed
u64values. - Binary
Embedding Space - Metadata about a binary embedding space.
- Collection
- Metadata about a collection.
- Collection
Name - A validated collection name.
- Collection
Schema - Schema for a collection defining which vectors are expected.
- Embedding
- A vector embedding with dimension validation.
- Embedding
Name - A name for an embedding space.
- Embedding
Space - Metadata about an embedding space.
- Multi
Vector Embedding - A multi-vector embedding with validation.
- Multi
Vector Embedding Space - Metadata about a multi-vector embedding space.
- Payload
- A point’s payload (JSON data).
- Sparse
Embedding - A sparse vector embedding with validation.
- Sparse
Embedding Space - Metadata about a sparse embedding space.
- Vector
Config - Configuration for a vector in a collection schema.
- Vector
Name - A validated vector name.
- Vector
Ref - Reference to a vector attached to an entity.
Enums§
- Named
Vector - A named vector value that can be dense, sparse, or multi-vector.
- Vector
Data - Vector data supporting different types.
- Vector
Type - Type of vector in a collection schema.