Scalar Quantization (SQ)
Scalar quantization (SQ) is a vector compression technique that can reduce the size of a vector.
To use SQ, enable it in the collection definition, then add data to the collection.
Basic configuration
SQ must be enabled at collection creation time. You cannot enable SQ after you add data to a collection.
To enable SQ, set vector_index_config.
- Python Client v4
- Python Client v3
import weaviate.classes.config as wc
client.collections.create(
name="MyCollection",
vectorizer_config=wc.Configure.Vectorizer.text2vec_openai(),
vector_index_config=wc.Configure.VectorIndex.hnsw(
quantizer=wc.Configure.VectorIndex.Quantizer.sq()
),
)
class_definition = {
"class": "MyCollection",
"vectorizer": "text2vec-openai", # Can be any vectorizer
"vectorIndexType": "hnsw",
"vectorIndexConfig": {
"SQ": {
"enabled": True,
},
},
# Remainder not shown
}
client.schema.create_class(class_definition)
Custom configuration
To tune SQ, set these vectorIndexConfig parameters.
| Parameter | Type | Default | Details |
|---|---|---|---|
sq: enabled | boolean | false | Uses SQ when true. The Python client v4 does not use the enabled parameter. To enable SQ with the v4 client, set a quantizer in the collection definition. |
sq: rescoreLimit | integer | -1 | The minimum number of candidates to fetch before rescoring. |
sq: trainingLimit | integer | 100000 | The size of the training set to determine scalar bucket boundaries. |
sq: cache | boolean | false | Use the vector cache when true. |
vectorCacheMaxObjects | integer | 1e12 | Maximum number of objects in the memory cache. By default, this limit is set to one trillion (1e12) objects when a new collection is created. For sizing recommendations, see Vector cache considerations. |
- Python Client v4
- Python Client v3
import weaviate.classes.config as wc
client.collections.create(
name="MyCollection",
vectorizer_config=wc.Configure.Vectorizer.text2vec_openai(),
vector_index_config=wc.Configure.VectorIndex.hnsw(
distance_metric=wc.VectorDistances.COSINE,
vector_cache_max_objects=100000,
quantizer=wc.Configure.VectorIndex.Quantizer.sq(
rescore_limit=200,
training_limit=50000,
cache=True,
)
),
)
class_definition = {
"class": "MyCollection",
"vectorizer": "text2vec-openai", # Can be any vectorizer
"vectorIndexType": "hnsw",
"vectorIndexConfig": {
"SQ": {
"enabled": True,
"rescoreLimit": 200, # The minimum number of candidates to fetch before rescoring
"cache": True, # Default: False
},
"vectorCacheMaxObjects": 100000, # Cache size (used if `cache` enabled)
},
# Remainder not shown
}
client.schema.create_class(class_definition)
Multiple vector embeddings (named vectors)
v1.24Collections can have multiple named vectors. The vectors in a collection can have their own configurations, and compression must be enabled independently for each vector. Every vector is independent and can use PQ, BQ, SQ, or no compression.
Multi-vector embeddings (ColBERT, ColPali, etc.)
v1.30Multi-vector embeddings (implemented through models like ColBERT, ColPali, or ColQwen) represent each object or query using multiple vectors instead of a single vector. Just like with single vectors, multi-vectors support PQ, BQ, SQ, or no compression.
During the initial search phase, compressed vectors are used for efficiency. However, when computing the MaxSim operation, uncompressed vectors are utilized to ensure more precise similarity calculations. This approach balances the benefits of compression for search efficiency with the accuracy of uncompressed vectors during final scoring.
Related pages
Questions and feedback
If you have any questions or feedback, let us know in the user forum.