Update app.py

app.py

@@ -45,6 +45,16 @@ def format_prompt_gemma(message,chatbot,sytem_prompt):
         messages, tokenize=False, add_generation_prompt=True, return_tensors="pt")
     return newPrompt
 
+def format_prompt_nous(message,chatbot,sytem_prompt):
+    fullPrompt =f"<|im_start|>system/n{sytem_prompt}<|im_end|>/n"
+    for user_message, bot_message in chatbot:
+        fullPrompt +=f"<|im_start|>user/n{user_message}<|im_end|>/n"
+        fullPrompt +=f"<|im_start|>assistant/n{bot_message}<|im_end|>/n"
+    fullPrompt +=f"<|im_start|>user/n{message}<|im_end|>/n"
+    fullPrompt +=f"<|im_start|>assistant"
+    return newPrompt
+    
+
 def inference(message, history, systemPrompt=SYSTEM_PROMPT+SYSTEM_PROMPT_PLUS, temperature=0.9, maxTokens=512, topP=0.9, repPenalty=1.1):
     # Updating the settings for the generation
     client_settings = dict(
@@ -59,7 +69,7 @@ def inference(message, history, systemPrompt=SYSTEM_PROMPT+SYSTEM_PROMPT_PLUS, t
         seed=random.randint(0, 999999999),
     )
     # Generating the response by passing the prompt in right format plus the client settings
-    stream = client.text_generation(format_prompt(message, history, systemPrompt),
+    stream = client.text_generation(format_prompt_nous(message, history, systemPrompt),
                                     **client_settings)
     # Reading the stream
     partial_response = ""