NVIDIA CUDA Runtime

The `nvidia-cuda-runtime` package provides essential native CUDA runtime libraries (e.g., `libcudart.so`) required for Python applications to utilize NVIDIA GPUs. It ensures these libraries are discoverable within a Python environment, typically without requiring a full system-wide CUDA Toolkit installation. It is primarily a dependency for deep learning frameworks like PyTorch and TensorFlow, rather than offering direct Python APIs. The current version is 13.2.51, with releases tied to NVIDIA CUDA Toolkit updates.