nexuml_library.scenarios.asd.synthetic_linear_ae¶
Composed scenarios combining data + model + training + evaluation.
synthetic_linear_ae_reconstruction(
feature_shape: tuple[int, ...] = (128,),
num_samples: int = 1000,
hidden_dims: list[int] | None = None,
latent_dim: int = 8,
lr: float = 0.001,
batch_size: int = 64,
max_epochs: int = 10,
) -> ScenarioSpec
Synthetic vector reconstruction with linear autoencoder.
Returns:
| Name | Type | Description |
|---|---|---|
ScenarioSpec |
ScenarioSpec
|
Assembled scenario with pipeline, training, data and evaluation. |
synthetic_linear_ae_multiclass(
feature_shape: tuple[int, ...] = (128,),
num_samples: int = 1000,
num_classes: int = 5,
hidden_dims: list[int] | None = None,
latent_dim: int = 8,
lr: float = 0.001,
batch_size: int = 64,
max_epochs: int = 10,
) -> ScenarioSpec
Synthetic vector reconstruction + multiclass classification.
Returns:
| Name | Type | Description |
|---|---|---|
ScenarioSpec |
ScenarioSpec
|
Assembled scenario with pipeline, training, data and evaluation. |
synthetic_linear_ae_multilabel(
feature_shape: tuple[int, ...] = (128,),
num_samples: int = 1000,
num_classes: int = 5,
hidden_dims: list[int] | None = None,
latent_dim: int = 8,
lr: float = 0.001,
batch_size: int = 64,
max_epochs: int = 10,
) -> ScenarioSpec
Synthetic vector reconstruction + multilabel classification.
Returns:
| Name | Type | Description |
|---|---|---|
ScenarioSpec |
ScenarioSpec
|
Assembled scenario with pipeline, training, data and evaluation. |
synthetic_linear_ae_regression(
feature_shape: tuple[int, ...] = (128,),
num_samples: int = 1000,
num_outputs: int = 3,
hidden_dims: list[int] | None = None,
latent_dim: int = 8,
lr: float = 0.001,
batch_size: int = 64,
max_epochs: int = 10,
) -> ScenarioSpec
Synthetic vector reconstruction + regression.
Returns:
| Name | Type | Description |
|---|---|---|
ScenarioSpec |
ScenarioSpec
|
Assembled scenario with pipeline, training, data and evaluation. |