Update app.py

app.py

@@ -114,13 +114,13 @@ async def load_model():
     print(f"Model downloaded to: {model_path}")
 
     print("Loading model into memory...")
-    llm = Llama(
+    llm = Llama(                                                                                                       
         model_path=model_path,
-        n_ctx=MAX_CONTEXT,
-        n_threads=4,           # Use more threads for better performance
-        n_batch=512,
+        n_ctx=MAX_CONTEXT,                                                                                             
+        n_threads=8,           # Changed from 4 to 8
+        n_batch=1024,          # Changed from 512 to 1024
         verbose=False,
-        n_gpu_layers=0         # CPU only (change if GPU available)
+        n_gpu_layers=0
     )
     print("Model loaded successfully!")