Searcher

class lightning_ir.retrieve.base.searcher.Searcher(index_dir: Path | str, search_config: SearchConfig, module: BiEncoderModule, use_gpu: bool = True)[source]

Bases: ABC

Base class for searchers in the Lightning IR framework.

__init__(index_dir: Path | str, search_config: SearchConfig, module: BiEncoderModule, use_gpu: bool = True) → None[source]

Initialize the Searcher.

Parameters:

  • index_dir (Path | str) – The directory containing the index files.

  • search_config (SearchConfig) – The configuration for the search.

  • module (BiEncoderModule) – The bi-encoder module to use for scoring.

  • use_gpu (bool) – Whether to use GPU for computations. Defaults to True.

Raises:

ValueError – If the document lengths do not match the index.

Methods

__init__(index_dir, search_config, module[, ...])

Initialize the Searcher.

search(output)

Search for documents based on the output of the bi-encoder model.

to_gpu()

Move the searcher to the GPU if available.
abstractmethod search(output: BiEncoderOutput) → Tuple[PackedTensor, List[List[str]]][source]

Search for documents based on the output of the bi-encoder model.

Parameters:

output (BiEncoderOutput) – The output from the bi-encoder model containing query and document embeddings.

Returns:

The top-k scores and corresponding document IDs.

Return type:

Tuple[PackedTensor, List[List[str]]]

to_gpu() → None[source]

Move the searcher to the GPU if available.