torch_frame.nn.encoder.LinearEncoder

class LinearEncoder(out_channels: int | None = None, stats_list: list[dict[StatType, Any]] | None = None, stype: stype | None = None, post_module: torch.nn.Module | None = None, na_strategy: NAStrategy | None = None)

Bases: StypeEncoder

A linear function based encoder for numerical features. It applies linear layer torch.nn.Linear(1, out_channels) on each raw numerical feature and concatenates the output embeddings. Note that the implementation does this for all numerical features in a batched manner.

reset_parameters() → None

Initialize the parameters of post_module.

encode_forward(feat: Tensor, col_names: list[str] | None = None) → Tensor

The main forward function. Maps input feat from TensorFrame (shape [batch_size, num_cols]) into output x of shape [batch_size, num_cols, out_channels].