diff --git a/scope/xgb.py b/scope/xgb.py
new file mode 100644
index 00000000..19f81b19
--- /dev/null
+++ b/scope/xgb.py
@@ -0,0 +1,129 @@
+import datetime
+import os
+import pathlib
+from .models import AbstractClassifier
+
+#!git clone https://github.com/DTrimarchi10/confusion_matrix.git
+import numpy as np
+from xgboost import XGBClassifier
+import xgboost as xgb
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score, confusion_matrix, matthews_corrcoef, mean_absolute_error
+from sklearn.metrics import f1_score, ConfusionMatrixDisplay
+#plot_confusion_matrix, is deprecated, so import ConfusionMatrixDisplay instead
+from sklearn.metrics import roc_curve, auc, precision_recall_curve, classification_report, average_precision_score
+import pandas as pd
+
+import seaborn as sns
+import matplotlib.pyplot as plt
+
+import os
+import sys
+import copy
+import json
+
+    
+
+class XGB(AbstractClassifier):
+    """Baseline model with a statically-defined architecture"""
+    def __init__(
+        self,
+        **kwargs,
+    )
+
+
+#    def call(self, inputs, **kwargs):
+#        features_input = inputs.get("features")
+#        dtrain = xgb.DMatrix(X_train, label=y_train)
+#        dtest = xgb.DMatrix(X_test, label=y_test)
+#
+#        return features_input, dtrain, dtest
+#
+    def summary(self, **kwargs):
+
+        model = xgb.train(
+            params,
+            dtrain,
+            num_boost_round=num_boost_round,
+            evals=[(dtest, "Test")],
+            early_stopping_rounds=10
+        )
+        return model.summary()
+
+
+    
+    def setup(
+        self,
+        'max_depth':6,
+        'min_child_weight': 1,
+        'eta':.1,
+        'subsample': 0.7,
+        'colsample_bytree': 0.7,
+        'objective': 'binary:logistic',
+        'eval_metric': 'auc',
+        **kwargs,
+    ):
+      #  removed for now:
+      #  'gpu_id': 0,
+      #  'tree_method': 'gpu_hist',
+
+    params = {'max_depth': max_depth,
+    'min_child_weight': min_child_weight,
+    'eta': eta,
+    'subsample': subsample,
+    'colsample_bytree': colsample_bytree,
+    'objective': objective,
+  #  'gpu_id': gpu_id,
+  #  'tree_method': tree_method,
+    'eval_metric': eval_metric,
+    }
+    
+    self.meta['params'] = params
+
+    self.model.compile(
+        params =self.meta["params"],
+        train = self.dtrain,
+        test = self.dtest,
+        )
+
+    def train(self,X_train, **kwargs):
+        dtrain = xgb.DMatrix(X_train, label=y_train)
+
+        return dtrain
+
+    def test(self, X_test, **kwargs);
+    
+        dtest = xgb.DMatrix(X_test, label=y_test)
+
+        return dtest
+
+              
+    def evaluate(self, test_dataset, **kwargs):
+        return self.model.evaluate(test_dataset, **kwargs)
+
+    def predict(self, x, **kwargs):
+        return self.model.predict(x, **kwargs)
+
+    
+    def load(self, path_model, **kwargs):
+        self.model.load_weights(path_model, **kwargs)
+
+    def save(
+        self,
+        tag: str,
+        output_path: str = "./",
+        output_format: str = "h5",
+    ):
+
+        if output_format not in ("h5",):
+            raise ValueError("unknown output format")
+
+        path = pathlib.Path(output_path)
+        if not path.exists():
+            path.mkdir(parents=True, exist_ok=True)
+
+        output_name = self.name if not tag else f"{self.name}.{tag}"
+        if not output_name.endswith('.h5'):
+            output_name += '.h5'
+        self.model.save(path / output_name, save_format=output_format)
+