Update app.py

app.py

@@ -3,7 +3,7 @@ from huggingface_hub import InferenceClient
 
 client = InferenceClient("Aksh1t/mistral-7b-oig-unsloth-merged")
 
-# Define the custom chat template
+# Custom chat template
 custom_template = {
     "chat": {
         "prompt": "The following is a conversation with an AI assistant. The assistant is helpful, creative, clever, and very friendly.\n\nHuman: {input}\nAI:",
@@ -11,6 +11,25 @@ custom_template = {
     }
 }
 
+def format_messages(message, history):
+    formatted_messages = []
+    
+    # Add system message if present
+    if system_message:
+        formatted_messages.append({"role": "system", "content": system_message})
+    
+    # Add history messages
+    for val in history:
+        if val[0]:
+            formatted_messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            formatted_messages.append({"role": "assistant", "content": val[1]})
+    
+    # Add current user message
+    formatted_messages.append({"role": "user", "content": message})
+    
+    return formatted_messages
+
 def respond(
     message,
     history: list[tuple[str, str]],
@@ -19,26 +38,16 @@ def respond(
     temperature,
     top_p,
 ):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
+    formatted_messages = format_messages(message, history)
     response = ""
 
-    # Call chat_completion with custom template and other parameters
+    # Call chat_completion with formatted messages
     for message in client.chat_completion(
-        messages,
+        formatted_messages,
         max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
-        template=custom_template  # Pass custom template here
     ):
         token = message.choices[0].delta.content