1921 & GSK-1851 & GSK-1901 - save langchain chain without ml flow to reduce compatibility

giskard/models/automodel.py

@@ -5,9 +5,9 @@
 
 import pandas as pd
 
-from ..core.core import ModelType, SupportedModelTypes
 from .base.serialization import CloudpickleSerializableModel
 from .function import PredictionFunctionModel
+from ..core.core import ModelType, SupportedModelTypes
 
 logger = logging.getLogger(__name__)
 
@@ -128,8 +128,8 @@ def __new__(
                     giskard_cls = CloudpickleSerializableModel
                 # if save_model and load_model are overriden, replace them, if not, these equalities will be identities.
                 possibly_overriden_cls = cls
-                possibly_overriden_cls.save_model = giskard_cls.save_model
-                possibly_overriden_cls.load_model = giskard_cls.load_model
+                # possibly_overriden_cls.save_model = giskard_cls.save_model
+                # possibly_overriden_cls.load_model = giskard_cls.load_model
                 possibly_overriden_cls.should_save_model_class = True
             elif giskard_cls:
                 input_type = "'prediction_function'" if giskard_cls == PredictionFunctionModel else "'model'"

tests/models/test_model.py

@@ -2,6 +2,7 @@
 from pathlib import Path
 
 import numpy as np
+import pandas as pd
 
 from giskard import Model
 from giskard.core.core import SupportedModelTypes
@@ -57,3 +58,36 @@ def prediction_fn(df):
 
         assert np.all(np.equal(predictions.raw, second_predictions.raw))
         assert nb_of_prediction_calls[0] == 1
+
+
+def test_model_save_and_load_not_overriden():
+    def model_fn(df):
+        return [True] * len(df)
+
+    call_count = dict({"save": 0, "load": 0})
+
+    class MyCustomModel(Model):
+        def save_model(self, path):
+            call_count["save"] = call_count["save"] + 1
+            Path(path).joinpath("custom_data").touch()
+
+        @classmethod
+        def load_model(cls, path, **kwargs):
+            call_count["load"] = call_count["load"] + 1
+
+            def model(x):
+                return [True] * len(x)
+
+            return model
+
+        def model_predict(self, df: pd.DataFrame):
+            return self.model(df)
+
+    with tempfile.TemporaryDirectory() as tmpdirname:
+        gsk_model = MyCustomModel(model_fn, model_type="regression")
+
+        gsk_model.save(tmpdirname)
+        assert call_count["save"] == 1
+
+        MyCustomModel.load(tmpdirname)
+        assert call_count["load"] == 1