diff --git a/tests/models/bloom/test_tokenization_bloom.py b/tests/models/bloom/test_tokenization_bloom.py
index 068886d0d829bd..71318d9dd1744e 100644
--- a/tests/models/bloom/test_tokenization_bloom.py
+++ b/tests/models/bloom/test_tokenization_bloom.py
@@ -135,7 +135,7 @@ def test_encodings_from_xnli_dataset(self):
     @require_jinja
     def test_tokenization_for_chat(self):
         tokenizer = self.get_rust_tokenizer()
-        tokenizer.chat_template = "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}"
+        tokenizer.chat_template = "{% for message in messages %}" "{{ message.content }}{{ eos_token }}" "{% endfor %}"
         test_chats = [
             [{"role": "system", "content": "You are a helpful chatbot."}, {"role": "user", "content": "Hello!"}],
             [