Improve CUDA enabled images docs and add a policy (#2097)

README.md

@@ -101,6 +101,7 @@ more information is available in the [documentation](https://jupyter-docker-stac
 - Single-platform images have either `aarch64-` or `x86_64-` tag prefixes, for example, `quay.io/jupyter/base-notebook:aarch64-python-3.11.6`
 - Starting from `2022-09-21`, we create multi-platform images (except `tensorflow-notebook`)
 - Starting from `2023-06-01`, we create a multi-platform `tensorflow-notebook` image as well
+- Starting from `2024-02-24`, we create CUDA enabled variant of `pytorch-notebook` image for `x86_64` platform
 
 ## Using old images
 

docs/using/selecting.md

@@ -16,6 +16,14 @@ This section provides details about the first.
 The Jupyter team maintains a set of Docker image definitions in the <https://github.com/jupyter/docker-stacks> GitHub repository.
 The following sections describe these images, including their contents, relationships, and versioning strategy.
 
+## CUDA enabled variant
+
+We provide CUDA accelerated version of `pytorch-notebook` image.
+Prepend a CUDA version prefix (like `cuda12-`) to the image tag to allow PyTorch operations to use compatible NVIDIA GPUs for accelerated computation.
+We only build images for 2 last major versions of CUDA.
+
+For example, you can use an image `quay.io/jupyter/pytorch-notebook:cuda12-python-3.11.8`
+
 ### jupyter/docker-stacks-foundation
 
 [Source on GitHub](https://github.com/jupyter/docker-stacks/tree/main/images/docker-stacks-foundation) |
@@ -197,8 +205,6 @@ It contains:
 - Everything in `jupyter/scipy-notebook` and its ancestor images
 - [pytorch](https://pytorch.org/) machine learning library
 
-> **GPU Acceleration:** Append a CUDA version prefix (`cuda11-` for CUDA 11 or `cuda12-` for CUDA 12) to the image tag to allow PyTorch operations to use compatible NVIDIA GPUs for accelerated computation.
-
 ### jupyter/datascience-notebook
 
 [Source on GitHub](https://github.com/jupyter/docker-stacks/tree/main/images/datascience-notebook) |