Update app.py

app.py

@@ -15,11 +15,14 @@ hf_hub_download(
     local_dir="./models"
 )
 
-def get_messages_formatter_type(model_name):
-    if "gemma" in model_name:
-        return MessagesFormatterType.GEMMA_2
-    else:
-        raise ValueError(f"Unsupported model: {model_name}")
+
+
+hf_hub_download(
+    repo_id="bartowski/gemma-2-27b-it-GGUF",
+    filename="gemma-2-27b-it-Q5_K_M.gguf",
+    local_dir="./models"
+)
+
 
 @spaces.GPU(duration=120)
 def respond(
@@ -33,7 +36,7 @@ def respond(
     top_k,
     repeat_penalty,
 ):
-    chat_template = get_messages_formatter_type(model)
+    chat_template = MessagesFormatterType.GEMMA_2
 
     llm = Llama(
         model_path=f"models/{model}",
@@ -90,9 +93,10 @@ demo = gr.ChatInterface(
     respond,
     additional_inputs=[
         gr.Dropdown([
-                'gemma-2-9b-it-Q5_K_M.gguf'
+                'gemma-2-9b-it-Q5_K_M.gguf',
+                'gemma-2-27b-it-Q5_K_M.gguf'
             ],
-            value="gemma-2-9b-it-Q5_K_M.gguf",
+            value="gemma-2-27b-it-Q5_K_M.gguf",
             label="Model"
         ),
         gr.Textbox(value="You are a helpful assistant.", label="System message"),
@@ -124,7 +128,7 @@ demo = gr.ChatInterface(
     undo_btn="Undo",
     clear_btn="Clear",
     submit_btn="Send",
-    description="Llama-cpp-agent: Chat with gemma-2-9b-it model",
+    description="Chat with Gemma 2 using llama.cpp",
     chatbot=gr.Chatbot(
         scale=1, 
         likeable=False,