Update README.md

README.md

@@ -64,10 +64,10 @@ tokenizer = AutoTokenizer.from_pretrained(model_id)
 # Text Generate Process
 instruction = input("instruction: ")
 inputs = input("input: ")
-user = f"# instruction:\n{instruction}\n\n# input:\n{inputs}"
+user = f"# instruction:\n{instruction}\n\n# input:\n{inputs}"  # The model uses this input template to quickly process various learning data and responses. (Therefore, if you do not use this template for input, the performance of Synapse-Model and the instructions of the persona may decline.)
 
 messages = [
-    {"role" : "system", "content" : ""}, # You can write a custom SystemPrompt here. However, it will not work properly (because SystemPrompt has not been learned).
+    {"role" : "system", "content" : ""},  # You can write a custom SystemPrompt here. However, it will not work properly (because SystemPrompt has not been learned).
     {"role" : "user", "content" : user}
 ]