This is a and fully integrated software stack that combines Theano, a numerical computation library for Python, with Python 3.6. It offers a stable and rigorously tested execution environment suitable for various tasks such as training, inference, or serving as an API service. The stack seamlessly integrates into continuous integration and deployment workflows, making it versatile for both short and long-running high-performance operations. It is fine-tuned for optimal performance when utilized on NVidia GPU.
This stack includes several essential components, such as CUDA, a parallel computing platform and API model, and cuDNN, a GPU-accelerated library of primitives designed for deep neural networks. Additionally, it comes with NVidia drivers and a Development preset, which includes program development and building tools like a C compiler and make.
Key versions included in this stack are Theano 1.0.0, Python 3.6.3, CUDA 9.0.176, cuDNN 7.0.5, NVidia drivers 384.98, Libc 2.22, OpenBLAS 0.2.20, NumPy 1.13.3, PyCUDA 2017.1.1, and SciPy 1.0.0.
