diff --git a/tokenizers/src/models/bpe/serialization.rs b/tokenizers/src/models/bpe/serialization.rs
index ca2590f1..152586ef 100644
--- a/tokenizers/src/models/bpe/serialization.rs
+++ b/tokenizers/src/models/bpe/serialization.rs
@@ -151,18 +151,27 @@ mod test {
 
     #[test]
     fn test_serialization() {
-        let vocab: Vocab = [("<unk>".into(), 0), ("a".into(), 1), ("b".into(), 2)]
-            .iter()
-            .cloned()
-            .collect();
+        let vocab: Vocab = [
+            ("<unk>".into(), 0),
+            ("a".into(), 1),
+            ("b".into(), 2),
+            ("ab".into(), 3),
+        ]
+        .iter()
+        .cloned()
+        .collect();
         let bpe = BpeBuilder::default()
-            .vocab_and_merges(vocab, vec![])
+            .vocab_and_merges(vocab, vec![("a".to_string(), "b".to_string())])
             .unk_token("<unk>".to_string())
             .ignore_merges(true)
             .build()
             .unwrap();
 
         let data = serde_json::to_string(&bpe).unwrap();
+        assert_eq!(
+            data,
+            r#"{"type":"BPE","dropout":null,"unk_token":"<unk>","continuing_subword_prefix":null,"end_of_word_suffix":null,"fuse_unk":false,"byte_fallback":false,"ignore_merges":true,"vocab":{"<unk>":0,"a":1,"b":2,"ab":3},"merges":["a b"]}"#
+        );
         let reconstructed = serde_json::from_str(&data).unwrap();
 
         assert_eq!(bpe, reconstructed);
diff --git a/tokenizers/src/models/mod.rs b/tokenizers/src/models/mod.rs
index bb7cebc4..9dd3dc96 100644
--- a/tokenizers/src/models/mod.rs
+++ b/tokenizers/src/models/mod.rs
@@ -204,6 +204,7 @@ impl_enum_from!(WordLevelTrainer, TrainerWrapper, WordLevelTrainer);
 #[cfg(test)]
 mod tests {
     use super::*;
+    use crate::models::bpe::{BpeBuilder, Vocab};
 
     #[test]
     fn trainer_wrapper_train_model_wrapper() {
@@ -224,4 +225,45 @@ mod tests {
         let serialized = serde_json::to_string(&ordered).unwrap();
         assert_eq!(serialized, "{\"Hi\":0,\"There\":2}");
     }
+
+    #[test]
+    fn serialization() {
+        let vocab: Vocab = [
+            ("<unk>".into(), 0),
+            ("a".into(), 1),
+            ("b".into(), 2),
+            ("ab".into(), 3),
+        ]
+        .iter()
+        .cloned()
+        .collect();
+        let bpe = BpeBuilder::default()
+            .vocab_and_merges(vocab, vec![("a".to_string(), "b".to_string())])
+            .unk_token("<unk>".to_string())
+            .ignore_merges(true)
+            .build()
+            .unwrap();
+
+        let model = ModelWrapper::BPE(bpe);
+
+        let data = serde_json::to_string(&model).unwrap();
+        assert_eq!(
+            data,
+            r#"{"type":"BPE","dropout":null,"unk_token":"<unk>","continuing_subword_prefix":null,"end_of_word_suffix":null,"fuse_unk":false,"byte_fallback":false,"ignore_merges":true,"vocab":{"<unk>":0,"a":1,"b":2,"ab":3},"merges":["a b"]}"#
+        );
+        let reconstructed = serde_json::from_str(&data).unwrap();
+
+        assert_eq!(model, reconstructed);
+
+        let invalid = r#"{"type":"BPE","dropout":null,"unk_token":"<unk>","continuing_subword_prefix":null,"end_of_word_suffix":null,"fuse_unk":false,"byte_fallback":false,"ignore_merges":true,"vocab":{"<unk>":0,"a":1,"b":2,"ab":3},"merges":["a b c"]}"#;
+        let reconstructed: std::result::Result<ModelWrapper, serde_json::Error> =
+            serde_json::from_str(&invalid);
+        match reconstructed {
+            Err(err) => assert_eq!(
+                err.to_string(),
+                "data did not match any variant of untagged enum ModelWrapper"
+            ),
+            _ => panic!("Expected an error here"),
+        }
+    }
 }