Update README.md

README.md

@@ -9,6 +9,9 @@ This repository includes a fast tokenizer for [google/gemma-7b](https://huggingf
 
 No new tokens were added during that process to ensure that the original model's embedding doesn't need to be modified. 
 
+
+_Note: It is important to note that this tokenizer is not 100% ChatML compliant, since it seems [google/gemma-7b](https://huggingface.co/google/gemma-7b), always requires the original `<bos>` token to be part of the input. This means the chat template is `<bos>` + `chatml` + `<eos>`_
+
 ```python
 from transformers import AutoTokenizer
 
@@ -21,7 +24,7 @@ messages = [
 ]
 
 chatml = tokenizer.apply_chat_template(messages, add_generation_prompt=False, tokenize=False)
-
+print(chatml)
 # <bos><|im_start|>system
 # You are Gemma.<|im_end|>
 # <|im_start|>user