Put all BaseMode.meta fields into @property methods
#1700
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…s-into-properties
…s-into-properties
rabah-khalek
reviewed
Jan 3, 2024
Contributor
There was a problem hiding this comment.
nitpick: can we also propagate the change to the other daughter classes?
python-client/giskard/models/huggingface.py
255: return [[p[label] for label in self.meta.classification_labels] for p in _predictions]
python-client/giskard/models/langchain.py
62: model_type=self.meta.model_type,
67: classification_threshold=self.meta.classification_threshold,
68: classification_labels=self.meta.classification_labels,
python-client/giskard/models/base/wrapper.py
176: if raw_predictions.shape[-1] != len(self.meta.classification_labels):
179: f"where `n_classes` is the number of classes in your model output ({len(self.meta.classification_labels)} in this case)."
and all of these occurrences?
python-client/tests/test_statistical.py
22: classification_label=model.meta.classification_labels[label],
46: classification_label=model.meta.classification_labels[label],
69: classification_label=model.meta.classification_labels[label],
python-client/tests/models/test_sklearn_wrapper.py
25: gsk_model.meta.model_type = "unknown"
python-client/tests/models/test_model_postprocess.py
14: model.meta = meta
python-client/tests/models/test_model_explanation.py
41: for label in model.meta.classification_labels:
python-client/tests/models/test_catboost_wrapper.py
26: assert model.meta.feature_names == ["feature_1", "feature_2"]
27: assert model.meta.classification_labels == ["target1", "target2"]
python-client/tests/scan/test_overconfidence_detector.py
27: model.meta.classification_labels = ["A", "B", "C"]
python-client/tests/scan/test_scanner.py
68: german_credit_model.meta.feature_names = None
python-client/tests/scan/test_underconfidence_detector.py
27: model.meta.classification_labels = ["A", "B", "C"]
python-client/tests/integrations/test_mlflow.py
20: model_type=mlflow_model_types[model.meta.model_type],
60: evaluator_config = {"model_config": {"classification_labels": model.meta.classification_labels}}
69: evaluator_config = {"model_config": {"classification_labels": model.meta.classification_labels}}
98: cl = model.meta.classification_labels
python-client/tests/test_metamorphic_direction.py
50: classification_label=model.meta.classification_labels[0],
69: classification_label=model.meta.classification_labels[0],
python-client/tests/test_upload.py
72: model_type=model.meta.model_type,
75: classification_labels=model.meta.classification_labels,
84: model_type=model.meta.model_type,
85: feature_names=model.meta.feature_names,
python-client/giskard/ml_worker/websocket/utils.py
271: if i.type == "Model" and i.model_meta is not None:
272: return ModelInput(i.name, i.model_meta.model_type)
python-client/giskard/llm/talk/talk.py
62: required_features = set() if self.model.meta.feature_names is None else set(self.model.meta.feature_names)
202: name=self.spec.model.meta.name,
203: model_type=self.spec.model.meta.model_type,
204: feature_names=self.spec.model.meta.feature_names,
205: classification_threshold=self.spec.model.meta.classification_threshold,
206: classification_labels=self.spec.model.meta.classification_labels,
python-client/giskard/core/model_validation.py
39: model_type = model.meta.model_type
47: validate_classification_labels(model.meta.classification_labels, model_type)
50: validate_classification_threshold_label(model.meta.classification_labels, model.meta.classification_threshold)
52: assert model.meta.feature_names is None or isinstance(
53: model.meta.feature_names, list
58: validate_features(feature_names=model.meta.feature_names, validate_df=validate_ds.df)
67: model.meta.name, model.meta.classification_labels, target_values, validate_ds.target
73: if model.meta.model_type == SupportedModelTypes.CLASSIFICATION and validate_ds.target is not None:
89: features = model.meta.feature_names if model.meta.feature_names is not None else validation_ds.df.columns
125: validate_prediction_output(validation_ds, model.meta.model_type, prediction.raw)
127: validate_classification_prediction(model.meta.classification_labels, prediction.raw)
153: with open(f + "/giskard-model-meta.yaml") as yaml_f:
314: num_classes = len(model.meta.classification_labels)
python-client/giskard/ml_worker/websocket/listener.py
255: labels = {k: v for k, v in enumerate(model.meta.classification_labels)}
257: if len(model.meta.classification_labels) > 2 or model.meta.classification_threshold is None:
265: diff = prediction_results.all_predictions.iloc[:, 1] - model.meta.classification_threshold
419: if model.meta.feature_names:
420: input_df = input_df[model.meta.feature_names]
422: map_features_weight = dict(zip(model.meta.classification_labels, list_weights))
python-client/giskard/ml_worker/testing/utils.py
29: and isinstance(model.meta.classification_labels[0], numbers.Number)
41: f'{model.meta.classification_labels} and not the target: "{target}". '
46: model.meta.model_type != SupportedModelTypes.CLASSIFICATION
47: or classification_label in model.meta.classification_labels
48: ), f'"{classification_label}" is not part of model labels: {model.meta.classification_labels}'
python-client/giskard/utils/analytics_collector.py
59: [anonymize(n) for n in model.meta.feature_names]
60: if model.meta.feature_names
66: "model_type": model.meta.model_type.value,
python-client/giskard/models/model_explanation.py
71: features_names = model.meta.feature_names
200: feature_names = model.meta.feature_names or list(dataset.df.columns.drop(dataset.target, errors="ignore"))
204: return ShapResult(shap_explanations, feature_types, model.meta.model_type, only_highest_proba)
229: feature_names = model.meta.feature_names or list(dataset.df.columns.drop(dataset.target, errors="ignore"))
237: class_names=model.meta.classification_labels,
240: raise ValueError(f"Prediction task is not supported: {model.meta.model_type}")
python-client/giskard/models/_precooked.py
50: model.meta.model_type,
51: model.meta.feature_names,
52: model.meta.classification_labels,
python-client/giskard/models/base/model.py
31:META_FILENAME = "giskard-model-meta.yaml"
382: client.save_model_meta(project_key, self.id, self.meta, platform.python_version(), get_size(f))
409: meta_response = client.load_model_meta(project_key, model_id)
python-client/giskard/testing/tests/statistic.py
153: raise ValueError(f"Prediction task is not supported: {model.meta.model_type}")
238: if isinstance(model.meta.classification_labels[0], numbers.Number):
244: if positive_outcome not in list(model.meta.classification_labels):
246: f"The positive outcome chosen {positive_outcome} is not part of the dataset target values {list(model.meta.classification_labels)}."
python-client/giskard/integrations/mlflow/giskard_evaluator_utils.py
34: module = pyfunc_model.metadata.flavors["python_function"]["loader_module"]
37: return flavor.load_model(pyfunc_model.metadata.get_model_info().model_uri)
python-client/giskard/testing/tests/performance.py
51: is_binary_classification = len(model.meta.classification_labels) == 2
55: metric = score_fn(targets, prediction, pos_label=model.meta.classification_labels[1])
215: if len(model.meta.classification_labels) == 2:
python-client/giskard/testing/tests/metamorphic.py
186: passed_idx, failed_idx = _compare_prediction(results_df, model.meta.model_type, direction, output_sensitivity)
432: _, failed_idx = _compare_prediction(result_df, model.meta.model_type, direction, None)
python-client/giskard/testing/tests/drift.py
796: if classification_label is not None and isinstance(model.meta.classification_labels[0], numbers.Number):
892: if classification_label is not None and isinstance(model.meta.classification_labels[0], numbers.Number):
python-client/giskard/testing/tests/calibration.py
19: label2id = {label: n for n, label in enumerate(model.meta.classification_labels)}
31: n = len(model.meta.classification_labels)
python-client/giskard/push/perturbation.py
261: if model.meta.model_type == SupportedModelTypes.CLASSIFICATION:
270: elif model.meta.model_type == SupportedModelTypes.REGRESSION:
python-client/giskard/push/contribution.py
40: if model.meta.model_type == SupportedModelTypes.CLASSIFICATION:
48: elif model.meta.model_type == SupportedModelTypes.REGRESSION:
80: model_type=model.meta.model_type,
110: if model.meta.model_type == SupportedModelTypes.CLASSIFICATION:
124: model_type=model.meta.model_type,
257: if model.meta.model_type == SupportedModelTypes.CLASSIFICATION:
258: return explain(model, ds, df.iloc[0])["explanations"][model.meta.classification_labels[0]]
259: elif model.meta.model_type == SupportedModelTypes.REGRESSION:
python-client/giskard/scanner/scanner.py
77: num_features = len(model.meta.feature_names) if model.meta.feature_names else len(dataset.columns)
91: detectors = self.get_detectors(tags=[model.meta.model_type.value])
188: raise ValueError(f"Dataset must be provided for {model.meta.model_type.value} models.")
python-client/giskard/scanner/issues.py
105: return self.model.meta.feature_names or self.dataset.columns.tolist()
python-client/giskard/scanner/robustness/base_detector.py
40: if model.meta.feature_names:
41: features = [f for f in features if f in model.meta.feature_names]
python-client/giskard/scanner/correlation/spurious_correlation_detector.py
27: features = model.meta.feature_names or dataset.columns.drop(dataset.target, errors="ignore")
python-client/giskard/scanner/common/examples.py
50: num_labels_to_print = min(len(issue.model.meta.classification_labels), int(with_prediction))
55: "\n".join([f"{issue.model.meta.classification_labels[i]} (p = {ps[i]:.2f})" for i in label_idx])
python-client/giskard/scanner/common/loss_based_detector.py
41: max_data_size = self.MAX_DATASET_SIZE // len(model.meta.feature_names or dataset.columns)
78: features = model.meta.feature_names or dataset.columns.drop(dataset.target, errors="ignore")
python-client/giskard/scanner/performance/metrics.py
81: labels=model.meta.classification_labels,
82: pos_label=model.meta.classification_labels[1],
86: return metric_fn(y_true, y_pred, labels=model.meta.classification_labels, average="micro")
97: neg = model.meta.classification_labels[0]
111: return (y_pred == model.meta.classification_labels[1]).sum()
123: return (y_true == model.meta.classification_labels[1]).sum()
140: y_true, y_score, multi_class="ovo", labels=model.meta.classification_labels
python-client/giskard/scanner/performance/performance_bias_detector.py
86: metrics.log_loss([true_label], [probs], labels=model.meta.classification_labels)
…s-into-properties
kevinmessiaen
force-pushed
the
feature/gsk-1957-put-all-the-modelmeta-fields-into-properties
branch
from
7cf09e0 to
a5c636b
Compare
|
rabah-khalek
deleted the
feature/gsk-1957-put-all-the-modelmeta-fields-into-properties
branch
andreybavt
added a commit
that referenced
this pull request
Jan 8, 2024
* Put all `BaseMode.meta` fields into `@property` methods * Moved model.meta.property to model.property * Moved model.meta.property to model.property * Import reorganization --------- Co-authored-by: Andrey Avtomonov <andreybavt@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Put all
BaseMode.metafields into@propertymethodsRelated Issue
GSK-1957
Type of Change
Checklist
CODE_OF_CONDUCT.mddocument.CONTRIBUTING.mdguide.make codestyle.