feat: Enforce right values in y_train to avoid out of index error

torchFastText/torchFastText.py

@@ -311,6 +311,12 @@ def build(
             self.num_classes = len(
                 np.unique(y_train)
             )  # Be sure that y_train contains all the classes !
+
+            if np.max(y_train) >= self.num_classes:
+                raise ValueError(
+                    f"y_train must contain values between 0 and {self.num_classes - 1}. Make sure that np.max(y_train) == len(np.unique(y_train))-1."
+                )
+
         else:
             if self.num_classes is None:
                 raise ValueError(