Regressors

Base Regressor

class udao.model.regressors.base_regressor.BaseRegressor(net_params: Params)

Bases: Module, ABC

class Params(input_embedding_dim: int, input_features_dim: int, output_dim: int)

Bases: object

input_embedding_dim: int

Size of the embedding part of the input.

input_features_dim: int

Size of the tabular features.

output_dim: int

Size of the output tensor.

abstract forward(embedding: Tensor, inst_feat: Tensor) → Tensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

MLP Regressor

class udao.model.regressors.mlp.MLP(net_params: Params)

Bases: BaseRegressor

MLP to compute the final regression from the embedding and the tabular input features. :param net_params: For the parameters, see the MLP.Params dataclass. :type net_params: MLP.Params

class Params(input_embedding_dim: int, input_features_dim: int, output_dim: int, n_layers: int, hidden_dim: int, dropout: float, agg_dims: List[int] | None = None)

Bases: Params

agg_dims: List[int] | None = None

Dimensions of the aggregation layers in the MLP.

dropout: float

Probability of dropout.

hidden_dim: int

Size of the hidden layers outputs.

n_layers: int

Number of layers in the MLP

forward(embedding: Tensor, inst_feat: Tensor) → Tensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.