nexuml_library.scenarios.evaluation.base¶
Evaluation scenario fragments.
merge_eval_specs(
eval_specs: list[EvaluationSpec],
) -> EvaluationSpec
Merge a list of evaluation specs into a single combined spec.
Returns:
| Name | Type | Description |
|---|---|---|
EvaluationSpec |
EvaluationSpec
|
Merged evaluation specification. |
reconstruction_evaluation(
reconstruction_feature_key: str = "features",
reconstruction_key: str = "reconstructed",
reconstruction_mask_key: str | None = None,
reconstruction_patch_size: int
| tuple[int, int]
| None = None,
reconstruction_label_keys: list[str] | None = None,
metrics: list[str] = ["mse", "mae"],
) -> EvaluationSpec
Evaluation spec for reconstruction metrics.
Returns:
| Name | Type | Description |
|---|---|---|
EvaluationSpec |
EvaluationSpec
|
Evaluation specification with reconstruction metrics. |
classification_evaluation(
label_key: str = "class",
feature_key: str = "latent",
max_samples: int = 2000,
metrics: list[str] = ["accuracy", "f1"],
) -> EvaluationSpec
Evaluation spec for classification metrics.
Returns:
| Name | Type | Description |
|---|---|---|
EvaluationSpec |
EvaluationSpec
|
Evaluation specification with classification metrics. |
regression_evaluation() -> EvaluationSpec
Evaluation spec for regression metrics.
Returns:
| Name | Type | Description |
|---|---|---|
EvaluationSpec |
EvaluationSpec
|
Evaluation specification with regression metrics. |