Update app.py

app.py

@@ -14,6 +14,9 @@ def run_command(command):
     return output.decode('utf-8')
 
 def install_packages():
+
+    run_command("apt install nvidia-cuda-toolkit")
+    
     # Clone the repository with submodules
     run_command("git clone --recurse-submodules https://github.com/abetlen/llama-cpp-python.git")
     
@@ -32,10 +35,12 @@ def install_packages():
     run_command("pip install -e .[all]")
     
     # Clear the local build cache
+    run_command("make GGML_CUDA=1")
+
     run_command("make clean")
     
     # Reinstall the package
-    run_command('CMAKE_ARGS="-DGGML_CUDA=on" pip install -e .')
+    run_command('pip install -e .')
 
     # Install llama-cpp-agent
     run_command("pip install llama-cpp-agent")