CrossEncoderConfig

class lightning_ir.cross_encoder.cross_encoder_config.CrossEncoderConfig(query_length: int | None = 32, doc_length: int | None = 512, pooling_strategy: 'first' | 'mean' | 'max' | 'sum' = 'first', linear_bias: bool = False, **kwargs)[source]

Bases: LightningIRConfig

__init__(query_length: int | None = 32, doc_length: int | None = 512, pooling_strategy: 'first' | 'mean' | 'max' | 'sum' = 'first', linear_bias: bool = False, **kwargs)[source]

Configuration class for a cross-encoder model

Parameters:

  • query_length (int | None) – Maximum number of tokens per query. If None does not truncate. Defaults to 32.

  • doc_length (int | None) – Maximum number of tokens per document. If None does not truncate. Defaults to 512.

  • pooling_strategy (Literal['first', 'mean', 'max', 'sum']) – Pooling strategy to aggregate the contextualized embeddings into a single vector for computing a relevance score. Defaults to “first”.

  • linear_bias (bool) – Whether to use a bias in the prediction linear layer. Defaults to False.

Methods

__init__([query_length, doc_length, ...])

Configuration class for a cross-encoder model

Attributes

model_type

Model type for cross-encoder models.
model_type: str = 'cross-encoder'

Model type for cross-encoder models.