Features of DLAMI

The features of AWS Deep Learning AMIs (DLAMI) include preinstalled deep learning frameworks, GPU software, model servers, and model visualization tools.

Preinstalled frameworks

There are currently two primary flavors of DLAMI with other variations related to the operating system (OS) and software versions:

The Deep Learning AMI with Conda uses conda environments to isolate each framework, so you can switch between them at will and not worry about their dependencies conflicting. The Deep Learning AMI with Conda supports the following frameworks:

Preinstalled GPU software

Even if you use a CPU-only instance, the DLAMIs will have NVIDIA CUDA and NVIDIA cuDNN. The installed software is the same regardless of the instance type. Keep in mind that GPU-specific tools work only on an instance that has at least one GPU. For more information about instance types, see Choosing a DLAMI instance type.

For more information about CUDA, see CUDA Installations and Framework Bindings.

Model serving and visualization

Deep Learning AMI with Conda comes preinstalled with model servers for TensorFlow, as well as TensorBoard for model visualizations. For more information, see TensorFlow Serving.