diff --git a/ragatouille/RAGTrainer.py b/ragatouille/RAGTrainer.py
index 15ebbc8..8da5ba6 100644
--- a/ragatouille/RAGTrainer.py
+++ b/ragatouille/RAGTrainer.py
@@ -95,7 +95,7 @@ def prepare_training_data(
             data_out_path: Union[str, Path] - Path to the directory where the data has been exported.
         """
         if all_documents is not None:
-            self.collection += all_documents
+            self.collection += [doc for doc in all_documents if isinstance(doc, str)]
 
         self.data_dir = Path(data_out_path)
         if len(raw_data[0]) == 2:
@@ -107,21 +107,22 @@ def prepare_training_data(
         else:
             raise ValueError("Raw data must be a list of pairs or triplets of strings.")
 
-        if type(raw_data[0][1]) == str:
-            self.collection += [x[1] for x in raw_data]
-        else:
-            for x in raw_data:
-                for txt in x[1]:
-                    self.collection.append(txt)
-        if data_type == "triplets":
-            if type(raw_data[0][2]) == str:
-                self.collection += [x[2] for x in raw_data]
+        self.queries = set()
+        for x in raw_data:
+            if isinstance(x[0], str):
+                self.queries.add(x[0])
             else:
-                for x in raw_data:
-                    for txt in x[2]:
-                        self.collection.append(txt)
-
-        self.queries = set([x[0] for x in raw_data])
+                raise ValueError("Queries must be a strings.")
+            if isinstance(x[1], str):
+                self.collection.append(x[1])
+            elif isinstance(x[1], list):
+                self.collection += [txt for txt in x[1] if isinstance(txt, str)]
+
+            if len(x) == 3:  # For triplets
+                if isinstance(x[2], str):
+                    self.collection.append(x[2])
+                elif isinstance(x[2], list):
+                    self.collection += [txt for txt in x[2] if isinstance(txt, str)]
         self.collection = list(set(self.collection))
         seeded_shuffle(self.collection)