Find answers from the community

Updated 6 months ago

```class VectorStoreQueryMode(str, Enum

At a glance

The post defines a VectorStoreQueryMode class with various modes for querying a vector store, such as DEFAULT, SPARSE, HYBRID, TEXT_SEARCH, SEMANTIC_HYBRID, and some learners like SVM, LOGISTIC_REGRESSION, and LINEAR_REGRESSION. In the comments, a community member notes that for the base SimpleVectorStore, the matching is done using similarity, but it may differ for third-party vector stores, and provides a link to the relevant code.

Useful resources

AAzathoth

Plain Text

class VectorStoreQueryMode(str, Enum):
    """Vector store query mode."""

    DEFAULT = "default"
    SPARSE = "sparse"
    HYBRID = "hybrid"
    TEXT_SEARCH = "text_search"
    SEMANTIC_HYBRID = "semantic_hybrid"

    # fit learners
    SVM = "svm"
    LOGISTIC_REGRESSION = "logistic_regression"
    LINEAR_REGRESSION = "linear_regression"

    # maximum marginal relevance
    MMR = "mmr"

1 comment

WWhiteFang_Jr

Yes for base SimpleVectorStore it matches using similarity. But it can differ in case of third party vector store.

https://github.com/run-llama/llama_index/blob/045582caf3564f5a549266c07d291f30b997425d/llama-index-core/llama_index/core/vector_stores/simple.py#L368

Add a reply