adjust thread numbers

app.py

@@ -13,9 +13,10 @@ hf_hub_download(
 def load_model():
     return Llama(
         model_path="models/qwen2.5-1.5b-instruct-q4_k_m.gguf",
-        n_ctx=2048,
-        n_threads=6,
-        n_batch=8,
+        n_ctx=1024,
+        n_threads=2,
+        n_threads_batch=2,
+        n_batch=4,
         n_gpu_layers=0,
         use_mlock=False,
         use_mmap=True,