diff --git a/nbs/nixtla_client.ipynb b/nbs/nixtla_client.ipynb
index 1d204229..7d0af32e 100644
--- a/nbs/nixtla_client.ipynb
+++ b/nbs/nixtla_client.ipynb
@@ -272,7 +272,6 @@
     "        self.weights_x: pd.DataFrame = None\n",
     "        self.freq: str = self.base_freq\n",
     "        self.drop_uid: bool = False\n",
-    "        self.x_cols: List[str]\n",
     "        self.input_size: int\n",
     "        self.model_horizon: int\n",
     "\n",
@@ -561,8 +560,7 @@
     "            X_df = df.drop(columns='y')\n",
     "            x_cols = X_df.drop(columns=['unique_id', 'ds']).columns.to_list()\n",
     "            X_df = self.preprocess_X_df(X_df)\n",
-    "        self.x_cols = x_cols\n",
-    "        return Y_df, X_df\n",
+    "        return Y_df, X_df, x_cols\n",
     "\n",
     "    def dataframes_to_dict(self, Y_df: pd.DataFrame, X_df: pd.DataFrame):\n",
     "        to_dict_args = {'orient': 'split'}\n",
@@ -604,7 +602,7 @@
     "    ):\n",
     "        df, X_df = self.transform_inputs(df=df, X_df=X_df)\n",
     "        main_logger.info('Preprocessing dataframes...')\n",
-    "        Y_df, X_df = self.preprocess_dataframes(df=df, X_df=X_df)\n",
+    "        Y_df, X_df, x_cols = self.preprocess_dataframes(df=df, X_df=X_df)\n",
     "        self.set_model_params()\n",
     "        if self.h > self.model_horizon:\n",
     "            main_logger.warning(\n",
@@ -659,7 +657,7 @@
     "        )\n",
     "        if 'weights_x' in response_timegpt:\n",
     "            self.weights_x = pd.DataFrame({\n",
-    "                'features': self.x_cols,\n",
+    "                'features': x_cols,\n",
     "                'weights': response_timegpt['weights_x'],\n",
     "            })\n",
     "        fcst_df = pd.DataFrame(**response_timegpt['forecast'])\n",
@@ -693,7 +691,7 @@
     "        # exogenous variables are passed after df \n",
     "        df, _ = self.transform_inputs(df=df, X_df=None)\n",
     "        main_logger.info('Preprocessing dataframes...')\n",
-    "        Y_df, X_df = self.preprocess_dataframes(df=df, X_df=None)\n",
+    "        Y_df, X_df, x_cols = self.preprocess_dataframes(df=df, X_df=None)\n",
     "        main_logger.info('Calling Anomaly Detector Endpoint...')\n",
     "        y, x = self.dataframes_to_dict(Y_df, X_df)\n",
     "        response_timegpt = self._call_api(\n",
@@ -709,7 +707,7 @@
     "        )\n",
     "        if 'weights_x' in response_timegpt:\n",
     "            self.weights_x = pd.DataFrame({\n",
-    "                'features': self.x_cols,\n",
+    "                'features': x_cols,\n",
     "                'weights': response_timegpt['weights_x'],\n",
     "            })\n",
     "        anomalies_df = pd.DataFrame(**response_timegpt['forecast'])\n",
diff --git a/nixtla/nixtla_client.py b/nixtla/nixtla_client.py
index 8d2e2707..212bf813 100644
--- a/nixtla/nixtla_client.py
+++ b/nixtla/nixtla_client.py
@@ -188,7 +188,6 @@ def __init__(
         self.weights_x: pd.DataFrame = None
         self.freq: str = self.base_freq
         self.drop_uid: bool = False
-        self.x_cols: List[str]
         self.input_size: int
         self.model_horizon: int
 
@@ -499,8 +498,7 @@ def preprocess_dataframes(
             X_df = df.drop(columns="y")
             x_cols = X_df.drop(columns=["unique_id", "ds"]).columns.to_list()
             X_df = self.preprocess_X_df(X_df)
-        self.x_cols = x_cols
-        return Y_df, X_df
+        return Y_df, X_df, x_cols
 
     def dataframes_to_dict(self, Y_df: pd.DataFrame, X_df: pd.DataFrame):
         to_dict_args = {"orient": "split"}
@@ -546,7 +544,7 @@ def forecast(
     ):
         df, X_df = self.transform_inputs(df=df, X_df=X_df)
         main_logger.info("Preprocessing dataframes...")
-        Y_df, X_df = self.preprocess_dataframes(df=df, X_df=X_df)
+        Y_df, X_df, x_cols = self.preprocess_dataframes(df=df, X_df=X_df)
         self.set_model_params()
         if self.h > self.model_horizon:
             main_logger.warning(
@@ -602,7 +600,7 @@ def forecast(
         if "weights_x" in response_timegpt:
             self.weights_x = pd.DataFrame(
                 {
-                    "features": self.x_cols,
+                    "features": x_cols,
                     "weights": response_timegpt["weights_x"],
                 }
             )
@@ -637,7 +635,7 @@ def detect_anomalies(self, df: pd.DataFrame):
         # exogenous variables are passed after df
         df, _ = self.transform_inputs(df=df, X_df=None)
         main_logger.info("Preprocessing dataframes...")
-        Y_df, X_df = self.preprocess_dataframes(df=df, X_df=None)
+        Y_df, X_df, x_cols = self.preprocess_dataframes(df=df, X_df=None)
         main_logger.info("Calling Anomaly Detector Endpoint...")
         y, x = self.dataframes_to_dict(Y_df, X_df)
         response_timegpt = self._call_api(
@@ -658,7 +656,7 @@ def detect_anomalies(self, df: pd.DataFrame):
         if "weights_x" in response_timegpt:
             self.weights_x = pd.DataFrame(
                 {
-                    "features": self.x_cols,
+                    "features": x_cols,
                     "weights": response_timegpt["weights_x"],
                 }
             )