attention.anchor_attention.config
=================================

.. py:module:: attention.anchor_attention.config


Classes
-------

.. autoapisummary::

   attention.anchor_attention.config.MultiBranchAnchorAttentionConfig
   attention.anchor_attention.config.CrossAchorAttentionConfig
   attention.anchor_attention.config.JointAnchorAttentionConfig
   attention.anchor_attention.config.TokenSpec
   attention.anchor_attention.config.AttentionPattern
   attention.anchor_attention.config.MixedAttentionConfig


Module Contents
---------------

.. py:class:: MultiBranchAnchorAttentionConfig(/, **data)

   Bases: :py:obj:`emmi.schemas.modules.attention.AttentionConfig`


   Configuration for Multi-Branch Anchor Attention module.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.


   .. py:attribute:: branches
      :type:  list[str]
      :value: None



   .. py:attribute:: anchor_suffix
      :type:  str
      :value: None



.. py:class:: CrossAchorAttentionConfig(/, **data)

   Bases: :py:obj:`MultiBranchAnchorAttentionConfig`


   Configuration for Cross Anchor Attention module.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.


.. py:class:: JointAnchorAttentionConfig(/, **data)

   Bases: :py:obj:`MultiBranchAnchorAttentionConfig`


   Configuration for Joint Anchor Attention module.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.


.. py:class:: TokenSpec(/, **data)

   Bases: :py:obj:`pydantic.BaseModel`


   Specification for a token type in the attention mechanism.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.


   .. py:attribute:: name
      :type:  Literal['surface_anchors', 'volume_anchors', 'surface_queries', 'volume_queries']


   .. py:attribute:: size
      :type:  int
      :value: None



   .. py:method:: from_dict(token_dict)
      :classmethod:


      Create TokenSpec from dictionary with single key-value pair.



   .. py:method:: to_dict()

      Convert TokenSpec to dictionary.



.. py:class:: AttentionPattern(/, **data)

   Bases: :py:obj:`pydantic.BaseModel`


   Defines which tokens attend to which other tokens.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.


   .. py:attribute:: query_tokens
      :type:  collections.abc.Sequence[str]


   .. py:attribute:: key_value_tokens
      :type:  collections.abc.Sequence[str]


.. py:class:: MixedAttentionConfig(/, **data)

   Bases: :py:obj:`emmi.schemas.modules.attention.DotProductAttentionConfig`


   Configuration for Mixed Attention module.

   Create a new model by parsing and validating input data from keyword arguments.

   Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
   validated to form a valid model.

   `self` is explicitly positional-only to allow `self` as a field name.