tabensemb.model.AbstractNN.get_hidden_state#

method

static AbstractNN.get_hidden_state(required_model, x, derived_tensors, model_name=None) → Tensor | None[source]#

The input of the last layer of a deep learning model, i.e. the hidden representation, whose dimension is (batch_size, required_model.hidden_rep_dim). The definition can be different for different models, depending on the different implementations of AbstractWrapper for different model bases.

Parameters:

required_model: A required model specified in AbstractModel.required_models() and extracted by AbstractModel._get_required_models().
x: See _forward().
derived_tensors: See _forward().
model_name: The name of the required model. It is necessary if the model comes from the same model base.

Returns:

torch.Tensor: The input of the last layer of a deep learning model.