lenskit.graphs.lightgcn.LightGCNConfig
======================================

.. py:class:: lenskit.graphs.lightgcn.LightGCNConfig
   

   Bases: :py:obj:`lenskit.config.common.EmbeddingSizeMixin`, :py:obj:`pydantic.BaseModel`


   Configuration for :class:`LightGCNScorer`.

   :Stability: Experimental


   .. py:attribute:: embedding_size
      :type:  pydantic.PositiveInt
      :value: 16


      The dimension of the embedding space (number of latent features).  Seems
      to work best as a power of 2.


   .. py:attribute:: layer_count
      :type:  pydantic.PositiveInt
      :value: 2


      The number of layers to use.


   .. py:attribute:: layer_blend
      :type:  pydantic.PositiveFloat | list[pydantic.PositiveFloat] | None
      :value: None


      The blending coefficient(s) for layer blending.  This is equivalent to
      ``alpha`` in :class:`LightGCN`.


   .. py:attribute:: batch_size
      :type:  pydantic.PositiveInt
      :value: 4096


      The training batch size.


   .. py:attribute:: learning_rate
      :type:  pydantic.PositiveFloat
      :value: 0.01


      The learning rate for training.


   .. py:attribute:: epochs
      :type:  pydantic.PositiveInt
      :value: 10


      The number of training epochs.


   .. py:attribute:: regularization
      :type:  pydantic.PositiveFloat | None
      :value: 0.01


      The regularization strength.


   .. py:attribute:: loss
      :type:  Literal['logistic', 'pairwise']
      :value: 'pairwise'


      The loss to use for model training.

      ``pairwise``
          BPR pairwise ranking loss, using :meth:`LightGCN.recommend_loss`.

      ``logistic``
          Logistic link prediction loss, using :meth:`LightGCN.link_pred_loss`.


   .. py:method:: check_layer_blending()