Trying different Llama settings

app.py

@@ -52,7 +52,8 @@ if not os.path.exists(model_path):
     print("Model downloaded successfully.")
 
 # Initialize the Llama model
-llm = Llama(model_path=model_path, n_ctx=2048, n_threads=4)
+llm = Llama(model_path=model_path, n_ctx=1024, n_threads=8, n_batch=512,
+    mlock=True, n_gpu_layers=-1)
 
 
 def search_problems(query, top_k=5):