Update app.py

app.py

@@ -10,7 +10,7 @@ llm = Llama(
         repo_id=os.environ.get("REPO_ID", "mradermacher/Atlas-Chat-2B-GGUF"),
         filename=os.environ.get("MODEL_FILE", "Atlas-Chat-2B.Q8_0.gguf"),
     ),
-    n_ctx=4096,
+    n_ctx=2048,
 )
 
 # Training prompt format for Atlas-Chat style conversation
@@ -77,7 +77,7 @@ demo = gr.ChatInterface(
     undo_btn="Delete Previous",
     clear_btn="Clear",
     additional_inputs=[
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Slider(minimum=1, maximum=768, value=256, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
             minimum=0.1,