Custom Space

Classes

CustomSpace(vector: superlinked.framework.common.schema.schema_object.FloatList | list[superlinked.framework.common.schema.schema_object.FloatList], length: int, aggregation: superlinked.framework.dsl.space.custom_space.CustomSpace.AggregationStrategy = AggregationStrategy.SUM_AND_NORMALIZE, description: str | None = None) : CustomSpace is the instrument of ingesting your own vectors into Superlinked. This way you can use your own vectors right away. What you need to know: (you can use numbering too) - vectors need to have the same length - vectors will be L2Norm normalized to ensure weighting makes sense - weighting can be performed (query-time) - you are going to need an FloatList typed SchemaField to supply your data - the FloatList field will be able to parse any Sequence[float | int] - you can leave the aggregation parameter as default, or switch it to using vector averaging during aggregation.

Initializes a CustomSpace for vector storage and manipulation within Superlinked.

This constructor sets up a space designed for custom vector ingestion, allowing users to specify how these
vectors are aggregated and normalized.

Args:
    vector (FloatList | list[FloatList]): The input vector(s) to be stored in the space.
      This can be a single FloatList SchemaField or a list of those.
    length (int): The fixed length that all vectors in this space must have. This ensures uniformity and
      consistency in vector operations.
    aggregation (AggregationStrategy, optional): The strategy to use for aggregating multiple vectors.
      This can be either `SUM_AND_NORMALIZE` for summing vectors and normalizing to unit length,
      or `VECTOR_AVERAGE` for averaging vectors during aggregation, but not performing other normalization.
      Defaults to `SUM_AND_NORMALIZE`.

Raises:
    InvalidAggregationStrategyException: If the specified aggregation strategy is not recognized. This ensures
      that only valid aggregation strategies are used.

### Ancestors (in MRO)

* superlinked.framework.dsl.space.space.Space
* abc.ABC

### Class variables

`AggregationStrategy`
:   Controls how the supplied vectors are aggregated and normalized under the hood. Choose the option most suitable
    for your custom vectors:
        - sum_and_normalize: during aggregation, vectors are summed up elementwise, and normalized using L2 norm
          of the vector to achieve unit vector length when needed.
        - vector_average: vectors are summed up elementwise in case of aggregation, and normalized using the
          number of the aggregated vectors to achieve < 1 length.
          This policy expects vectors that are roughly unit length. Use it for vectors that are incompatible with L2
          normalization.

### Instance variables

`annotation: str`
: