SamplingConfig Class
ml_debugger.data_filter.config.sampling_config.SamplingConfig
Bases: BaseModel
Sampling configuration for query result selection.
Attributes:
| Name | Type | Description |
|---|---|---|
method |
str
|
Sampling method. - "random": Random sampling from candidates within [min_value, max_value] range. - "class_balanced": Per-class quota random sampling. |
min_per_class |
int
|
Target number of samples per class (method="class_balanced" only). This is a target quota, not a guarantee. Classes with no candidates are skipped, and budget exhaustion may limit per-class counts. |
seed |
Optional[int]
|
Random seed for reproducibility (method="random" only). Ignored when method="class_balanced" (deterministic by design). |
min_value |
Optional[float]
|
Lower bound (inclusive) on rank_value for candidate filtering. Required (at least one of min_value/max_value) when method="random". Optional for method="class_balanced". |
max_value |
Optional[float]
|
Upper bound (inclusive) on rank_value for candidate filtering. Required (at least one of min_value/max_value) when method="random". Optional for method="class_balanced". |
validate_dict(data)
classmethod
Validate dict and return result with guide message.