Multivector Search with LanceDB
LanceDB's multivector support enables you to store and search multiple vector embeddings for a single item. This capability is particularly valuable when working with late interaction models like ColBERT and ColPaLi that generate multiple embeddings per document.
Multivector Support
Only in Python SDK
Multivector search is currently supported in our Python SDK.
Each item in your dataset can have a column containing multiple vectors, which LanceDB can efficiently index and search. When performing a search, you can query using either a single vector embedding or multiple vector embeddings.
LanceDB also integrates with ConteXtualized Token Retriever (XTR), an advanced retrieval model that prioritizes the most semantically important document tokens during search. This integration enhances the quality of search results by focusing on the most relevant token matches.
- Supported similarity metric: Cosine similarity only
- Vector value types: float16, float32, or float64
Computing Similarity
MaxSim (Maximum Similarity) is a key concept in late interaction models that:
- Computes the maximum similarity between each query embedding and all document embeddings
- Sums these maximum similarities to get the final relevance score
- Effectively captures fine-grained semantic matches between query and document tokens
The MaxSim calculation can be expressed as:
where sim
is the similarity function (e.g. cosine).
Distance Metric
For now, only cosine
metric is supported for multivector search.
The vector value type can be float16
, float32
or float64
.
Example: Multivector Search
1. Setup
Connect to LanceDB and import required libraries for data management.
2. Define Schema
Define a schema that specifies a multivector field. A multivector field is a nested list structure where each document contains multiple vectors. In this case, we'll create a schema with:
- An ID field as an integer (int64)
- A vector field that is a list of lists of float32 values
- The outer list represents multiple vectors per document
- Each inner list is a 256-dimensional vector
- Using float32 for memory efficiency while maintaining precision
3. Generate Multivectors
Generate sample data where each document contains multiple vector embeddings, which could represent different aspects or views of the same document.
In this example, we create 1024 documents where each document has 2 random vectors of dimension 256, simulating a real-world scenario where you might have multiple embeddings per item.
4. Create a Table
Create a table with the defined schema and sample data, which will store multiple vectors per document for similarity search.
5. Build an Index
Only cosine similarity is supported as the distance metric for multivector search operations.
For faster search, build the standard IVF_PQ
index over your vectors:
6. Query a Single Vector
When searching with a single query vector, it will be compared against all vectors in each document, and the similarity scores will be aggregated to find the most relevant documents.
7. Query Multiple Vectors
With multiple vector queries, LanceDB calculates similarity using late interaction - a technique that computes relevance by finding the best matching pairs between query and document vectors. This approach provides more nuanced matching while maintaining fast retrieval speeds.
What's Next?
If you still need more guidance, you can try the complete Multivector Search Notebook.
For advanced users, we prepared a Tutorial on ColPali and LanceDB.