workbench 0.8.178__py3-none-any.whl → 0.8.180__py3-none-any.whl

workbench/utils/xgboost_model_utils.py

@@ -4,6 +4,7 @@ import logging
 import os
 import tempfile
 import tarfile
+import joblib
 import pickle
 import glob
 import awswrangler as wr
@@ -16,7 +17,6 @@ from typing import Dict, Any
 from sklearn.model_selection import KFold, StratifiedKFold
 from sklearn.metrics import (
     precision_recall_fscore_support,
-    confusion_matrix,
     mean_squared_error,
     mean_absolute_error,
     r2_score,
@@ -34,14 +34,12 @@ log = logging.getLogger("workbench")
 def xgboost_model_from_s3(model_artifact_uri: str):
     """
     Download and extract XGBoost model artifact from S3, then load the model into memory.
-    Handles both direct XGBoost model files and pickled models.
-    Ensures categorical feature support is enabled.
 
     Args:
         model_artifact_uri (str): S3 URI of the model artifact.
 
     Returns:
-        Loaded XGBoost model or None if unavailable.
+        Loaded XGBoost model (XGBClassifier, XGBRegressor, or Booster) or None if unavailable.
     """
 
     with tempfile.TemporaryDirectory() as tmpdir:
@@ -55,64 +53,86 @@ def xgboost_model_from_s3(model_artifact_uri: str):
 
         # Define model file patterns to search for (in order of preference)
         patterns = [
-            # Direct XGBoost model files
-            os.path.join(tmpdir, "xgboost-model"),
-            os.path.join(tmpdir, "xgb_model*.json"),
-            os.path.join(tmpdir, "model"),
-            os.path.join(tmpdir, "*.bin"),
+            # Joblib models (preferred - preserves everything)
+            os.path.join(tmpdir, "*model*.joblib"),
+            os.path.join(tmpdir, "xgb*.joblib"),
+            os.path.join(tmpdir, "**", "*model*.joblib"),
+            os.path.join(tmpdir, "**", "xgb*.joblib"),
+            # Pickle models (also preserves everything)
+            os.path.join(tmpdir, "*model*.pkl"),
+            os.path.join(tmpdir, "xgb*.pkl"),
+            os.path.join(tmpdir, "**", "*model*.pkl"),
+            os.path.join(tmpdir, "**", "xgb*.pkl"),
+            # JSON models (fallback - requires reconstruction)
+            os.path.join(tmpdir, "*model*.json"),
+            os.path.join(tmpdir, "xgb*.json"),
             os.path.join(tmpdir, "**", "*model*.json"),
-            os.path.join(tmpdir, "**", "rmse.json"),
-            # Pickled models
-            os.path.join(tmpdir, "*.pkl"),
-            os.path.join(tmpdir, "**", "*.pkl"),
-            os.path.join(tmpdir, "*.pickle"),
-            os.path.join(tmpdir, "**", "*.pickle"),
+            os.path.join(tmpdir, "**", "xgb*.json"),
         ]
 
         # Try each pattern
         for pattern in patterns:
-            # Use glob to find all matching files
             for model_path in glob.glob(pattern, recursive=True):
-                # Determine file type by extension
+                # Skip files that are clearly not XGBoost models
+                filename = os.path.basename(model_path).lower()
+                if any(skip in filename for skip in ["label_encoder", "scaler", "preprocessor", "transformer"]):
+                    log.debug(f"Skipping non-model file: {model_path}")
+                    continue
+
                 _, ext = os.path.splitext(model_path)
 
                 try:
-                    if ext.lower() in [".pkl", ".pickle"]:
-                        # Handle pickled models
+                    if ext == ".joblib":
+                        model = joblib.load(model_path)
+                        # Verify it's actually an XGBoost model
+                        if isinstance(model, (xgb.XGBClassifier, xgb.XGBRegressor, xgb.Booster)):
+                            log.important(f"Loaded XGBoost model from joblib: {model_path}")
+                            return model
+                        else:
+                            log.debug(f"Skipping non-XGBoost object from {model_path}: {type(model)}")
+
+                    elif ext in [".pkl", ".pickle"]:
                         with open(model_path, "rb") as f:
                             model = pickle.load(f)
-
-                        # Handle different model types
-                        if isinstance(model, xgb.Booster):
-                            log.important(f"Loaded XGBoost Booster from pickle: {model_path}")
+                        # Verify it's actually an XGBoost model
+                        if isinstance(model, (xgb.XGBClassifier, xgb.XGBRegressor, xgb.Booster)):
+                            log.important(f"Loaded XGBoost model from pickle: {model_path}")
                             return model
-                        elif hasattr(model, "get_booster"):
-                            log.important(f"Loaded XGBoost model from pipeline: {model_path}")
-                            booster = model.get_booster()
-                            return booster
-                    else:
-                        # Handle direct XGBoost model files
+                        else:
+                            log.debug(f"Skipping non-XGBoost object from {model_path}: {type(model)}")
+
+                    elif ext == ".json":
+                        # JSON files should be XGBoost models by definition
                         booster = xgb.Booster()
                         booster.load_model(model_path)
-                        log.important(f"Loaded XGBoost model directly: {model_path}")
+                        log.important(f"Loaded XGBoost booster from JSON: {model_path}")
                         return booster
+
                 except Exception as e:
-                    log.info(f"Failed to load model from {model_path}: {e}")
-                    continue  # Try the next file
+                    log.debug(f"Failed to load {model_path}: {e}")
+                    continue
 
-    # If no model found
     log.error("No XGBoost model found in the artifact.")
     return None
 
 
-def feature_importance(workbench_model, importance_type: str = "weight") -> Optional[List[Tuple[str, float]]]:
+def feature_importance(workbench_model, importance_type: str = "gain") -> Optional[List[Tuple[str, float]]]:
     """
     Get sorted feature importances from a Workbench Model object.
 
     Args:
         workbench_model: Workbench model object
-        importance_type: Type of feature importance.
-            Options: 'weight', 'gain', 'cover', 'total_gain', 'total_cover'
+        importance_type: Type of feature importance. Options:
+            - 'gain' (default): Average improvement in loss/objective when feature is used.
+                     Best for understanding predictive power of features.
+            - 'weight': Number of times a feature appears in trees (split count).
+                       Useful for understanding model complexity and feature usage frequency.
+            - 'cover': Average number of samples affected when feature is used.
+                      Shows the relative quantity of observations related to this feature.
+            - 'total_gain': Total improvement in loss/objective across all splits.
+                           Similar to 'gain' but not averaged (can be biased toward frequent features).
+            - 'total_cover': Total number of samples affected across all splits.
+                            Similar to 'cover' but not averaged.
 
     Returns:
         List of tuples (feature, importance) sorted by importance value (descending).
@@ -121,7 +141,8 @@ def feature_importance(workbench_model, importance_type: str = "weight") -> Opti
 
     Note:
         XGBoost's get_score() only returns features with non-zero importance.
-        This function ensures all model features are included in the output.
+        This function ensures all model features are included in the output,
+        adding zero values for features that weren't used in any tree splits.
     """
     model_artifact_uri = workbench_model.model_data_url()
     xgb_model = xgboost_model_from_s3(model_artifact_uri)
@@ -129,11 +150,18 @@ def feature_importance(workbench_model, importance_type: str = "weight") -> Opti
         log.error("No XGBoost model found in the artifact.")
         return None
 
-    # Get feature importances (only non-zero features)
-    importances = xgb_model.get_score(importance_type=importance_type)
+    # Check if we got a full sklearn model or just a booster (for backwards compatibility)
+    if hasattr(xgb_model, "get_booster"):
+        # Full sklearn model - get the booster for feature importance
+        booster = xgb_model.get_booster()
+        all_features = booster.feature_names
+    else:
+        # Already a booster (legacy JSON load)
+        booster = xgb_model
+        all_features = xgb_model.feature_names
 
-    # Get all feature names from the model
-    all_features = xgb_model.feature_names
+    # Get feature importances (only non-zero features)
+    importances = booster.get_score(importance_type=importance_type)
 
     # Create complete importance dict with zeros for missing features
     complete_importances = {feat: importances.get(feat, 0.0) for feat in all_features}
@@ -230,143 +258,132 @@ def leaf_stats(df: pd.DataFrame, target_col: str) -> pd.DataFrame:
     return result_df
 
 
-def cross_fold_inference(workbench_model: Any, nfolds: int = 5) -> Dict[str, Any]:
+def cross_fold_inference(workbench_model: Any, nfolds: int = 5) -> Tuple[Dict[str, Any], pd.DataFrame]:
     """
     Performs K-fold cross-validation with detailed metrics.
     Args:
         workbench_model: Workbench model object
         nfolds: Number of folds for cross-validation (default is 5)
     Returns:
-        Dictionary containing:
-            - folds: Dictionary of formatted strings for each fold
-            - summary_metrics: Summary metrics across folds
-            - overall_metrics: Overall metrics for all folds
+        Tuple of:
+            - Dictionary containing:
+                - folds: Dictionary of formatted strings for each fold
+                - summary_metrics: Summary metrics across folds
+            - DataFrame with columns: id, target, prediction (out-of-fold predictions for all samples)
     """
     from workbench.api import FeatureSet
 
     # Load model
-    model_type = workbench_model.model_type.value
     model_artifact_uri = workbench_model.model_data_url()
-    loaded_booster = xgboost_model_from_s3(model_artifact_uri)
-    if loaded_booster is None:
+    loaded_model = xgboost_model_from_s3(model_artifact_uri)
+    if loaded_model is None:
         log.error("No XGBoost model found in the artifact.")
-        return {}
-    # Create the model wrapper
-    is_classifier = model_type == "classifier"
-    xgb_model = (
-        xgb.XGBClassifier(enable_categorical=True) if is_classifier else xgb.XGBRegressor(enable_categorical=True)
-    )
-    xgb_model._Booster = loaded_booster
+        return {}, pd.DataFrame()
+
+    # Check if we got a full sklearn model or need to create one
+    if isinstance(loaded_model, (xgb.XGBClassifier, xgb.XGBRegressor)):
+        xgb_model = loaded_model
+        is_classifier = isinstance(xgb_model, xgb.XGBClassifier)
+    elif isinstance(loaded_model, xgb.Booster):
+        # Legacy: got a booster, need to wrap it
+        log.warning("Deprecated: Loaded model is a Booster, wrapping in sklearn model.")
+        is_classifier = workbench_model.model_type.value == "classifier"
+        xgb_model = (
+            xgb.XGBClassifier(enable_categorical=True) if is_classifier else xgb.XGBRegressor(enable_categorical=True)
+        )
+        xgb_model._Booster = loaded_model
+    else:
+        log.error(f"Unexpected model type: {type(loaded_model)}")
+        return {}, pd.DataFrame()
+
     # Prepare data
     fs = FeatureSet(workbench_model.get_input())
     df = fs.view("training").pull_dataframe()
+
+    # Get id column - assuming FeatureSet has an id_column attribute or similar
+    id_col = fs.id_column
+    target_col = workbench_model.target()
     feature_cols = workbench_model.features()
+
     # Convert string features to categorical
     for col in feature_cols:
         if df[col].dtype in ["object", "string"]:
             df[col] = df[col].astype("category")
-    # Split X and y
-    X = df[workbench_model.features()]
-    y = df[workbench_model.target()]
 
-    # Encode target if it's a classification problem
+    X = df[feature_cols]
+    y = df[target_col]
+    ids = df[id_col]
+
+    # Encode target if classifier
     label_encoder = LabelEncoder() if is_classifier else None
     if label_encoder:
-        y = pd.Series(label_encoder.fit_transform(y), name=workbench_model.target())
+        y_encoded = label_encoder.fit_transform(y)
+        y_for_cv = pd.Series(y_encoded, index=y.index, name=target_col)
+    else:
+        y_for_cv = y
+
     # Prepare KFold
-    kfold = (
-        StratifiedKFold(n_splits=nfolds, shuffle=True, random_state=42)
-        if is_classifier
-        else KFold(n_splits=nfolds, shuffle=True, random_state=42)
-    )
+    kfold = (StratifiedKFold if is_classifier else KFold)(n_splits=nfolds, shuffle=True, random_state=42)
+
+    # Initialize results collection
+    fold_metrics = []
+    predictions_df = pd.DataFrame({id_col: ids, target_col: y})  # Keep original values
+    # Note: 'prediction' column will be created automatically with correct dtype
 
-    fold_results = []
-    all_predictions = []
-    all_actuals = []
-    for fold_idx, (train_idx, val_idx) in enumerate(kfold.split(X, y)):
+    # Perform cross-validation
+    for fold_idx, (train_idx, val_idx) in enumerate(kfold.split(X, y_for_cv), 1):
         X_train, X_val = X.iloc[train_idx], X.iloc[val_idx]
-        y_train, y_val = y.iloc[train_idx], y.iloc[val_idx]
+        y_train, y_val = y_for_cv.iloc[train_idx], y_for_cv.iloc[val_idx]
 
-        # Train the model
+        # Train and predict
         xgb_model.fit(X_train, y_train)
         preds = xgb_model.predict(X_val)
-        all_predictions.extend(preds)
-        all_actuals.extend(y_val)
 
-        # Calculate metrics for this fold
-        fold_metrics = {"fold": fold_idx + 1}
+        # Store predictions (decode if classifier)
+        val_indices = X_val.index
+        if is_classifier:
+            predictions_df.loc[val_indices, "prediction"] = label_encoder.inverse_transform(preds.astype(int))
+        else:
+            predictions_df.loc[val_indices, "prediction"] = preds
 
+        # Calculate fold metrics
         if is_classifier:
-            y_val_original = label_encoder.inverse_transform(y_val)
-            preds_original = label_encoder.inverse_transform(preds.astype(int))
-            scores = precision_recall_fscore_support(
-                y_val_original, preds_original, average="weighted", zero_division=0
+            y_val_orig = label_encoder.inverse_transform(y_val)
+            preds_orig = label_encoder.inverse_transform(preds.astype(int))
+            prec, rec, f1, _ = precision_recall_fscore_support(
+                y_val_orig, preds_orig, average="weighted", zero_division=0
             )
-            fold_metrics.update({"precision": float(scores[0]), "recall": float(scores[1]), "fscore": float(scores[2])})
+            fold_metrics.append({"fold": fold_idx, "precision": prec, "recall": rec, "fscore": f1})
         else:
-            fold_metrics.update(
+            fold_metrics.append(
                 {
-                    "rmse": float(np.sqrt(mean_squared_error(y_val, preds))),
-                    "mae": float(mean_absolute_error(y_val, preds)),
-                    "r2": float(r2_score(y_val, preds)),
+                    "fold": fold_idx,
+                    "rmse": np.sqrt(mean_squared_error(y_val, preds)),
+                    "mae": mean_absolute_error(y_val, preds),
+                    "r2": r2_score(y_val, preds),
                 }
             )
 
-        fold_results.append(fold_metrics)
-    # Calculate overall metrics
-    overall_metrics = {}
-    if is_classifier:
-        all_actuals_original = label_encoder.inverse_transform(all_actuals)
-        all_predictions_original = label_encoder.inverse_transform(all_predictions)
-        scores = precision_recall_fscore_support(
-            all_actuals_original, all_predictions_original, average="weighted", zero_division=0
-        )
-        overall_metrics.update(
-            {
-                "precision": float(scores[0]),
-                "recall": float(scores[1]),
-                "fscore": float(scores[2]),
-                "confusion_matrix": confusion_matrix(
-                    all_actuals_original, all_predictions_original, labels=label_encoder.classes_
-                ).tolist(),
-                "label_names": list(label_encoder.classes_),
-            }
-        )
-    else:
-        overall_metrics.update(
-            {
-                "rmse": float(np.sqrt(mean_squared_error(all_actuals, all_predictions))),
-                "mae": float(mean_absolute_error(all_actuals, all_predictions)),
-                "r2": float(r2_score(all_actuals, all_predictions)),
-            }
-        )
-    # Calculate summary metrics across folds
-    summary_metrics = {}
-    metrics_to_aggregate = ["precision", "recall", "fscore"] if is_classifier else ["rmse", "mae", "r2"]
-
-    for metric in metrics_to_aggregate:
-        values = [fold[metric] for fold in fold_results]
-        summary_metrics[metric] = f"{float(np.mean(values)):.3f} ±{float(np.std(values)):.3f}"
-    # Format fold results as strings (TBD section)
+    # Calculate summary metrics (mean ± std)
+    fold_df = pd.DataFrame(fold_metrics)
+    metric_names = ["precision", "recall", "fscore"] if is_classifier else ["rmse", "mae", "r2"]
+    summary_metrics = {metric: f"{fold_df[metric].mean():.3f} ±{fold_df[metric].std():.3f}" for metric in metric_names}
+
+    # Format fold results for display
     formatted_folds = {}
-    for fold_data in fold_results:
-        fold_key = f"Fold {fold_data['fold']}"
+    for _, row in fold_df.iterrows():
+        fold_key = f"Fold {int(row['fold'])}"
         if is_classifier:
             formatted_folds[fold_key] = (
-                f"precision: {fold_data['precision']:.3f}  "
-                f"recall: {fold_data['recall']:.3f}  "
-                f"fscore: {fold_data['fscore']:.3f}"
+                f"precision: {row['precision']:.3f}  " f"recall: {row['recall']:.3f}  " f"fscore: {row['fscore']:.3f}"
             )
         else:
-            formatted_folds[fold_key] = (
-                f"rmse: {fold_data['rmse']:.3f}  mae: {fold_data['mae']:.3f}  r2: {fold_data['r2']:.3f}"
-            )
-    # Return the results
-    return {
-        "summary_metrics": summary_metrics,
-        # "overall_metrics": overall_metrics,
-        "folds": formatted_folds,
-    }
+            formatted_folds[fold_key] = f"rmse: {row['rmse']:.3f}  " f"mae: {row['mae']:.3f}  " f"r2: {row['r2']:.3f}"
+
+    # Build return dictionary
+    metrics_dict = {"summary_metrics": summary_metrics, "folds": formatted_folds}
+
+    return metrics_dict, predictions_df
 
 
 if __name__ == "__main__":
@@ -384,6 +401,10 @@ if __name__ == "__main__":
     model_artifact_uri = model.model_data_url()
     xgb_model = xgboost_model_from_s3(model_artifact_uri)
 
+    # Verify enable_categorical is preserved (for debugging/confidence)
+    print(f"Model parameters: {xgb_model.get_params()}")
+    print(f"enable_categorical: {xgb_model.enable_categorical}")
+
     # Test with UQ Model
     uq_model = Model("aqsol-uq")
     _xgb_model = xgboost_model_from_s3(uq_model.model_data_url())
@@ -408,10 +429,12 @@ if __name__ == "__main__":
 
     print("\n=== CROSS FOLD REGRESSION EXAMPLE ===")
     model = Model("abalone-regression")
-    results = cross_fold_inference(model)
+    results, df = cross_fold_inference(model)
     pprint(results)
+    print(df.head())
 
     print("\n=== CROSS FOLD CLASSIFICATION EXAMPLE ===")
     model = Model("wine-classification")
-    results = cross_fold_inference(model)
+    results, df = cross_fold_inference(model)
     pprint(results)
+    print(df.head())

{workbench-0.8.178.dist-info → workbench-0.8.180.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: workbench
-Version: 0.8.178
+Version: 0.8.180
 Summary: Workbench: A Dashboard and Python API for creating and deploying AWS SageMaker Model Pipelines
 Author-email: SuperCowPowers LLC <support@supercowpowers.com>
 License-Expression: MIT

{workbench-0.8.178.dist-info → workbench-0.8.180.dist-info}/RECORD

@@ -31,7 +31,7 @@ workbench/api/__init__.py,sha256=kvrP70ypDOMdPGj_Eeftdh8J0lu_1qQVne6GXMkD4_E,102
 workbench/api/compound.py,sha256=kf5EaM5qjWwsZutcxqj9IC_MPnDV1uVHDMns9OA_GOo,2545
 workbench/api/data_source.py,sha256=Ngz36YZWxFfpJbmURhM1LQPYjh5kdpZNGo6_fCRePbA,8321
 workbench/api/df_store.py,sha256=Wybb3zO-jPpAi2Ns8Ks1-lagvXAaBlRpBZHhnnl3Lms,6131
-workbench/api/endpoint.py,sha256=RWGqxsCW_pMiENMb_XZlm2ZCldMS4suEBM3F5gT3hYI,3814
+workbench/api/endpoint.py,sha256=RDQs78OFRSp9GYGHHph0gLtAfF4p3We96bzdJ4-9bzw,3926
 workbench/api/feature_set.py,sha256=Yxei3tvWR4gSLcdJnNndux07dNeKNu1HKgsChJtHxEM,6633
 workbench/api/graph_store.py,sha256=LremJyPrQFgsHb7hxsctuCsoxx3p7TKtaY5qALHe6pc,4372
 workbench/api/meta.py,sha256=1_9989cPvf3hd3tA-83hLijOGNnhwXAF8aZF45adeDQ,8596
@@ -54,8 +54,8 @@ workbench/core/artifacts/cached_artifact_mixin.py,sha256=ngqFLZ4cQx_TFouXZgXZQsv
 workbench/core/artifacts/data_capture_core.py,sha256=q8f79rRTYiZ7T4IQRWXl8ZvPpcvZyNxYERwvo8o0OQc,14858
 workbench/core/artifacts/data_source_abstract.py,sha256=5IRCzFVK-17cd4NXPMRfx99vQAmQ0WHE5jcm5RfsVTg,10619
 workbench/core/artifacts/data_source_factory.py,sha256=YL_tA5fsgubbB3dPF6T4tO0rGgz-6oo3ge4i_YXVC-M,2380
-workbench/core/artifacts/endpoint_core.py,sha256=Q6wL0IpMgCkVssX-BvPwawgogQjq9klSaoBUZ6tEIuc,49146
-workbench/core/artifacts/feature_set_core.py,sha256=0wvW4VyZii0GmO6tBudoGEqZktLtb6spDyIkn7MkDcw,30292
+workbench/core/artifacts/endpoint_core.py,sha256=S9QHtXrQXkCTY-coaYSsAGJi60nZj3Aq9ruyVE1PUQs,49224
+workbench/core/artifacts/feature_set_core.py,sha256=7b1o_PzxtwaYC-W2zxlkltiO0fYULA8CVGWwHNmqgtI,31457
 workbench/core/artifacts/model_core.py,sha256=ECDwQ0qM5qb1yGJ07U70BVdfkrW9m7p9e6YJWib3uR0,50855
 workbench/core/artifacts/monitor_core.py,sha256=M307yz7tEzOEHgv-LmtVy9jKjSbM98fHW3ckmNYrwlU,27897
 workbench/core/cloud_platform/cloud_meta.py,sha256=-g4-LTC3D0PXb3VfaXdLR1ERijKuHdffeMK_zhD-koQ,8809
@@ -132,36 +132,36 @@ workbench/model_scripts/custom_models/chem_info/requirements.txt,sha256=7HBUzvNi
 workbench/model_scripts/custom_models/meta_endpoints/example.py,sha256=hzOAuLhIGB8vei-555ruNxpsE1GhuByHGjGB0zw8GSs,1726
 workbench/model_scripts/custom_models/network_security/Readme.md,sha256=Z2gtiu0hLHvEJ1x-_oFq3qJZcsK81sceBAGAGltpqQ8,222
 workbench/model_scripts/custom_models/proximity/Readme.md,sha256=RlMFAJZgAT2mCgDk-UwR_R0Y_NbCqeI5-8DUsxsbpWQ,289
-workbench/model_scripts/custom_models/proximity/feature_space_proximity.template,sha256=2c3eDu4sLP_bCTBAf_aIR1QdC7CpYDpXo8UU_2ZoLuE,4833
+workbench/model_scripts/custom_models/proximity/feature_space_proximity.template,sha256=eOllmqB20BWtTiV53dgpIqXKtgSbPFDW_zf8PvM3oF0,4813
 workbench/model_scripts/custom_models/proximity/generated_model_script.py,sha256=RdbKbXtrSNYQJvB-oLcRHpJ6w0TM7zbmMfuocHb7GM0,7967
 workbench/model_scripts/custom_models/proximity/proximity.py,sha256=zqmNlX70LnWXr5fdtFFQppSNTLjlOciQVrjGr-g9jRE,13716
 workbench/model_scripts/custom_models/proximity/requirements.txt,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workbench/model_scripts/custom_models/uq_models/Readme.md,sha256=UVpL-lvtTrLqwBeQFinLhd_uNrEw4JUlggIdUSDrd-w,188
-workbench/model_scripts/custom_models/uq_models/bayesian_ridge.template,sha256=U4LIlpp8Rbu3apyzPR7-55lvlutpTsCro_PUvQ5pklY,6457
-workbench/model_scripts/custom_models/uq_models/ensemble_xgb.template,sha256=0IJnSBACQ556ldEiPqR7yPCOOLJs1hQhHmPBvB2d9tY,13491
-workbench/model_scripts/custom_models/uq_models/gaussian_process.template,sha256=QbDUfkiPCwJ-c-4Twgu4utZuYZaAyeW_3T1IP-_tutw,6683
-workbench/model_scripts/custom_models/uq_models/generated_model_script.py,sha256=DUH80Y-We_-3OomUNjvBdRPrNQLQb3zlSsKZIPiglU4,22402
-workbench/model_scripts/custom_models/uq_models/mapie.template,sha256=SHP1Sd-nWMVF5sgB9Ski6C4IkQlm4g0EqpnJT1GfHl4,18204
-workbench/model_scripts/custom_models/uq_models/meta_uq.template,sha256=eawh0Fp3DhbdCXzWN6KloczT5ZS_ou4ayW65yUTTE4o,14109
-workbench/model_scripts/custom_models/uq_models/ngboost.template,sha256=9-O6P-SW50ul5Wl6es2DMWXSbrwOg7HWsdc8Qdln0MM,8278
+workbench/model_scripts/custom_models/uq_models/bayesian_ridge.template,sha256=ca3CaAk6HVuNv1HnPgABTzRY3oDrRxomjgD4V1ZDwoc,6448
+workbench/model_scripts/custom_models/uq_models/ensemble_xgb.template,sha256=xlKLHeLQkScONnrlbAGIsrCm2wwsvcfv4Vdrw4nlc_8,13457
+workbench/model_scripts/custom_models/uq_models/gaussian_process.template,sha256=3nMlCi8nEbc4N-MQTzjfIcljfDQkUmWeLBfmd18m5fg,6632
+workbench/model_scripts/custom_models/uq_models/generated_model_script.py,sha256=hE54OgIsR20-5RBjqqjUp5Z_aMzrg1Ykhwx4J3kG8Pw,19401
+workbench/model_scripts/custom_models/uq_models/mapie.template,sha256=FEBYxM-gVzwHWYga11DZf3ORUqmCrT0eQk8apixwx_E,19000
+workbench/model_scripts/custom_models/uq_models/meta_uq.template,sha256=XTfhODRaHlI1jZGo9pSe-TqNsk2_nuSw0xMO2fKzDv8,14011
+workbench/model_scripts/custom_models/uq_models/ngboost.template,sha256=v1rviYTJGJnQRGgAyveXhOQlS-WFCTlc2vdnWq6HIXk,8241
 workbench/model_scripts/custom_models/uq_models/proximity.py,sha256=zqmNlX70LnWXr5fdtFFQppSNTLjlOciQVrjGr-g9jRE,13716
 workbench/model_scripts/custom_models/uq_models/requirements.txt,sha256=fw7T7t_YJAXK3T6Ysbesxh_Agx_tv0oYx72cEBTqRDY,98
 workbench/model_scripts/custom_script_example/custom_model_script.py,sha256=T8aydawgRVAdSlDimoWpXxG2YuWWQkbcjBVjAeSG2_0,6408
 workbench/model_scripts/custom_script_example/requirements.txt,sha256=jWlGc7HH7vqyukTm38LN4EyDi8jDUPEay4n45z-30uc,104
-workbench/model_scripts/ensemble_xgb/ensemble_xgb.template,sha256=s8tPPk_q6UqA2nAzknD8viA-kN7f62Rim2XwMKcqHKc,10399
+workbench/model_scripts/ensemble_xgb/ensemble_xgb.template,sha256=pWmuo-EVz0owvkRI-h9mUTYt1-ouyD-_yyQu6SQbYZ4,10350
 workbench/model_scripts/ensemble_xgb/generated_model_script.py,sha256=dsjUGm22xI1ThGn97HPKtooyEPK-HOQnf5chnZ7-MXk,10675
 workbench/model_scripts/ensemble_xgb/requirements.txt,sha256=jWlGc7HH7vqyukTm38LN4EyDi8jDUPEay4n45z-30uc,104
 workbench/model_scripts/pytorch_model/generated_model_script.py,sha256=Mr1IMQJE_ML899qjzhjkrP521IjvcAvqU0pk--FB7KY,22356
-workbench/model_scripts/pytorch_model/pytorch.template,sha256=3jM3RUH68r75eH9Wayz6YTXZ7qpuDnaJCKKcHD_oKqA,22054
+workbench/model_scripts/pytorch_model/pytorch.template,sha256=_gRp6DH294FLxF21UpSTq7s9RFfrLjViKvjXQ4yDfBQ,21999
 workbench/model_scripts/pytorch_model/requirements.txt,sha256=ICS5nW0wix44EJO2tJszJSaUrSvhSfdedn6FcRInGx4,181
-workbench/model_scripts/quant_regression/quant_regression.template,sha256=AQihffV68qI6CG9qztA0jGunDWoijb3eeDWNG5tiIGc,9818
+workbench/model_scripts/quant_regression/quant_regression.template,sha256=2F25lZ7m_VafHvuGrC__R3uB1NzKgZu94eWJS9sWpYg,9783
 workbench/model_scripts/quant_regression/requirements.txt,sha256=jWlGc7HH7vqyukTm38LN4EyDi8jDUPEay4n45z-30uc,104
 workbench/model_scripts/scikit_learn/generated_model_script.py,sha256=c73ZpJBlU5k13Nx-ZDkLXu7da40CYyhwjwwmuPq6uLg,12870
 workbench/model_scripts/scikit_learn/requirements.txt,sha256=aVvwiJ3LgBUhM_PyFlb2gHXu_kpGPho3ANBzlOkfcvs,107
-workbench/model_scripts/scikit_learn/scikit_learn.template,sha256=d4pgeZYFezUQsB-7iIsjsUgB1FM6d27651wpfDdXmI0,12640
-workbench/model_scripts/xgb_model/generated_model_script.py,sha256=BPhr2gfJQC1C26knsyktfLGL7Jp0YBKCIQjplCuHUg0,22218
+workbench/model_scripts/scikit_learn/scikit_learn.template,sha256=QQvqx-eX9ZTbYmyupq6R6vIQwosmsmY_MRBPaHyfjdk,12586
+workbench/model_scripts/xgb_model/generated_model_script.py,sha256=Tbn7EMXxZZO8rDdKQ5fYCbpltACsMXNvuusLL9p-U5c,22319
 workbench/model_scripts/xgb_model/requirements.txt,sha256=jWlGc7HH7vqyukTm38LN4EyDi8jDUPEay4n45z-30uc,104
-workbench/model_scripts/xgb_model/xgb_model.template,sha256=HViJRsMWn393hP8VJRS45UQBzUVBhwR5sKc8Ern-9f4,17963
+workbench/model_scripts/xgb_model/xgb_model.template,sha256=9wed0Ii6KZpEiVj6fHGDkwpxPkLIhcSIDls7Ij0HUF0,17874
 workbench/repl/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workbench/repl/workbench_shell.py,sha256=Vhg4BQr2r4D4ymekrVtOFi0MaRvaH4V2UgcWRgvN_3U,22122
 workbench/resources/open_source_api.key,sha256=3S0OTblsmC0msUPdE_dbBmI83xJNmYscuwLJ57JmuOc,433
@@ -219,7 +219,7 @@ workbench/utils/lambda_utils.py,sha256=7GhGRPyXn9o-toWb9HBGSnI8-DhK9YRkwhCSk_mNK
 workbench/utils/license_manager.py,sha256=sDuhk1mZZqUbFmnuFXehyGnui_ALxrmYBg7gYwoo7ho,6975
 workbench/utils/log_utils.py,sha256=7n1NJXO_jUX82e6LWAQug6oPo3wiPDBYsqk9gsYab_A,3167
 workbench/utils/markdown_utils.py,sha256=4lEqzgG4EVmLcvvKKNUwNxVCySLQKJTJmWDiaDroI1w,8306
-workbench/utils/model_utils.py,sha256=97yqEEeGLV8KSDt_RTGexcUEK1wU_UnmLj-cfuryPOs,12779
+workbench/utils/model_utils.py,sha256=b71URPIsbkmFy2XEglftO47lQGIl5CpahjCbb2CnTpM,12801
 workbench/utils/monitor_utils.py,sha256=kVaJ7BgUXs3VPMFYfLC03wkIV4Dq-pEhoXS0wkJFxCc,7858
 workbench/utils/pandas_utils.py,sha256=uTUx-d1KYfjbS9PMQp2_9FogCV7xVZR6XLzU5YAGmfs,39371
 workbench/utils/performance_utils.py,sha256=WDNvz-bOdC99cDuXl0urAV4DJ7alk_V3yzKPwvqgST4,1329
@@ -231,7 +231,7 @@ workbench/utils/redis_cache.py,sha256=39LFSWmOlNNcah02D3sBnmibc-DPeKC3SNq71K4HaB
 workbench/utils/repl_utils.py,sha256=rWOMv2HiEIp8ZL6Ps6DlwiJlGr-pOhv9OZQhm3aR-1A,4668
 workbench/utils/resource_utils.py,sha256=EM4SrMmRUQnG80aR5M7hmzw86hYdP_S7fRPuqhpDSVo,1435
 workbench/utils/s3_utils.py,sha256=Xme_o_cftC_jWnw6R9YKS6-6C11zaCBAoQDlY3dZb5o,7337
-workbench/utils/shap_utils.py,sha256=gbrwha7hu64xmMGqr__kQXrMXwJpKHrYghh9RYP62OE,12346
+workbench/utils/shap_utils.py,sha256=dtjSIwSyvYSaQjjvIp5A9LGS7pr-5Vt907rvVKOrqNY,12651
 workbench/utils/shapley_values.py,sha256=3DvQz4HIPnxW42idgtuQ5vtzU-oF4_lToaWzLRjU-E4,3673
 workbench/utils/symbols.py,sha256=PioF1yAQyOabw7kLg8nhvaZBPFe7ABkpfpPPE0qz_2k,1265
 workbench/utils/test_data_generator.py,sha256=gqRXL7IUKG4wVfO1onflY3wg7vLkgx402_Zy3iqY7NU,11921
@@ -242,7 +242,7 @@ workbench/utils/workbench_cache.py,sha256=IQchxB81iR4eVggHBxUJdXxUCRkqWz1jKe5gxN
 workbench/utils/workbench_event_bridge.py,sha256=z1GmXOB-Qs7VOgC6Hjnp2DI9nSEWepaSXejACxTIR7o,4150
 workbench/utils/workbench_logging.py,sha256=WCuMWhQwibrvcGAyj96h2wowh6dH7zNlDJ7sWUzdCeI,10263
 workbench/utils/workbench_sqs.py,sha256=RwM80z7YWwdtMaCKh7KWF8v38f7eBRU7kyC7ZhTRuI0,2072
-workbench/utils/xgboost_model_utils.py,sha256=NNcALcBNOveqkIJiG7Wh7DS0O95RlGE3GZJbdSB8XWY,15571
+workbench/utils/xgboost_model_utils.py,sha256=wSUrs9VlftaTZ-cWZMEeHY6TmcLvxwrKk4S4lr7kWWw,17482
 workbench/utils/chem_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 workbench/utils/chem_utils/fingerprints.py,sha256=Qvs8jaUwguWUq3Q3j695MY0t0Wk3BvroW-oWBwalMUo,5255
 workbench/utils/chem_utils/misc.py,sha256=Nevf8_opu-uIPrv_1_0ubuFVVo2_fGUkMoLAHB3XAeo,7372
@@ -287,9 +287,9 @@ workbench/web_interface/page_views/main_page.py,sha256=X4-KyGTKLAdxR-Zk2niuLJB2Y
 workbench/web_interface/page_views/models_page_view.py,sha256=M0bdC7bAzLyIaE2jviY12FF4abdMFZmg6sFuOY_LaGI,2650
 workbench/web_interface/page_views/page_view.py,sha256=Gh6YnpOGlUejx-bHZAf5pzqoQ1H1R0OSwOpGhOBO06w,455
 workbench/web_interface/page_views/pipelines_page_view.py,sha256=v2pxrIbsHBcYiblfius3JK766NZ7ciD2yPx0t3E5IJo,2656
-workbench-0.8.178.dist-info/licenses/LICENSE,sha256=z4QMMPlLJkZjU8VOKqJkZiQZCEZ--saIU2Z8-p3aVc0,1080
-workbench-0.8.178.dist-info/METADATA,sha256=kS1snm2EjzaXVrpsg3TX28OmXqYDdZD1K7kQ0lXhNg8,9210
-workbench-0.8.178.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-workbench-0.8.178.dist-info/entry_points.txt,sha256=zPFPruY9uayk8-wsKrhfnIyIB6jvZOW_ibyllEIsLWo,356
-workbench-0.8.178.dist-info/top_level.txt,sha256=Dhy72zTxaA_o_yRkPZx5zw-fwumnjGaeGf0hBN3jc_w,10
-workbench-0.8.178.dist-info/RECORD,,
+workbench-0.8.180.dist-info/licenses/LICENSE,sha256=z4QMMPlLJkZjU8VOKqJkZiQZCEZ--saIU2Z8-p3aVc0,1080
+workbench-0.8.180.dist-info/METADATA,sha256=p3RyXhe_7ljNODfTkP6bW44_TQhwIriZnyJsJ41Xtck,9210
+workbench-0.8.180.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+workbench-0.8.180.dist-info/entry_points.txt,sha256=zPFPruY9uayk8-wsKrhfnIyIB6jvZOW_ibyllEIsLWo,356
+workbench-0.8.180.dist-info/top_level.txt,sha256=Dhy72zTxaA_o_yRkPZx5zw-fwumnjGaeGf0hBN3jc_w,10
+workbench-0.8.180.dist-info/RECORD,,