Update app.py

app.py

@@ -77,8 +77,10 @@ def respond(
                 model_path=f"models/{model}",
                 flash_attn=False,
                 n_gpu_layers=0,
-                n_batch=32,
-                n_ctx=8192,
+                n_batch=16,
+                n_ctx=2048,
+                n_threads=2,
+                use_mmap=True
             )
             llm_model = model
         provider = LlamaCppPythonProvider(llm)
@@ -158,9 +160,9 @@ demo = gr.ChatInterface(
         ),
         gr.Slider(
             minimum=512,
-            maximum=4096,
-            value=2048,
-            step=512,
+            maximum=2048,
+            value=1024,
+            step=1,
             label="Max Tokens",
             info="Maximum length of response (higher = longer replies)",
         ),