diff --git a/llmfoundry/utils/builders.py b/llmfoundry/utils/builders.py
index 071fb98ed8..104ad1764c 100644
--- a/llmfoundry/utils/builders.py
+++ b/llmfoundry/utils/builders.py
@@ -184,7 +184,7 @@ def build_tokenizer(
             int(1e30),
         )
 
-        return tokenizer
+    return tokenizer
 
 
 def build_icl_evaluators(
diff --git a/tests/test_builders.py b/tests/test_builders.py
new file mode 100644
index 0000000000..adff8e55ee
--- /dev/null
+++ b/tests/test_builders.py
@@ -0,0 +1,31 @@
+# Copyright 2022 MosaicML LLM Foundry authors
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+from transformers import PreTrainedTokenizerBase
+
+from llmfoundry.tokenizers.tiktoken import TiktokenTokenizerWrapper
+from llmfoundry.utils.builders import build_tokenizer
+
+
+@pytest.mark.parametrize('tokenizer_name,tokenizer_kwargs', [
+    ('tiktoken', {
+        'model_name': 'gpt-4'
+    }),
+    ('EleutherAI/gpt-neo-125M', {
+        'model_max_length': 10
+    }),
+    ('mosaicml/mpt-7b', {
+        'model_max_length': 20
+    }),
+])
+def test_tokenizer_builder(tokenizer_name: str, tokenizer_kwargs: dict):
+    tokenizer = build_tokenizer(tokenizer_name, tokenizer_kwargs)
+
+    if tokenizer_name == 'tiktoken':
+        assert isinstance(tokenizer, TiktokenTokenizerWrapper)
+        assert tokenizer.model_name == tokenizer_kwargs['model_name']
+    else:
+        assert tokenizer.model_max_length == tokenizer_kwargs[
+            'model_max_length']
+        assert isinstance(tokenizer, PreTrainedTokenizerBase)