neuraltrain.models.linear.LinearModel

class neuraltrain.models.linear.LinearModel(n_in_channels: int, n_outputs: int, reduction: str = 'mean', subject_layers_config: SubjectLayers | None = None)[source][source]

nn.Module implementation of Linear.

forward(x, subject_id=None)[source][source]

Forward pass: reduce time, then project.

Parameters:
  • x (Tensor) – Input of shape (B, C, T) or (B, C).

  • subject_id (Tensor or None) – Per-example subject indices, shape (B,).