

All images are multiarch (x86_64 and ARM).See cuSpatial for an example and information on RAPIDS’ usage of Dev Containers.Development images are no longer being published, in the coming releases RAPIDS will roll out Dev Containers for development.To learn more about these changes, please see the RAPIDS Container README. RAPIDS 23.08 brings significant Docker changes. The installation method below may allow RAPIDS CUDA 12.0 packages to coexist with PyTorch CUDA 12.1 nightly packages if there is a hard-requirement of CUDA 12 but it is currently unsupported:Ĭonda create -solver=libmamba -n rapids-pytorch-cu12 -c rapidsai -c pytorch-nightly -c conda-forge -c nvidia rapids=23.08 cuda-version=12.0 pytorch pytorch-cuda=12.1 PyTorch currently only has nightly builds for CUDA 12.1, stable builds are limited to CUDA 11. For ARM support, please use CUDA 11.Īt the time of writing, there is no stable CUDA 12 release of PyTorch:

CONDA INSTALL JUPYTER LAB UPDATE
If the Conda installation is older than 22.11, please update your Conda or Miniconda to the latest version.If the Conda installation is version 22.11 or newer, run: conda install -n base conda-libmamba-solver.To resolve this error please follow one of these steps: Conda create: error: argument -solver: invalid choice: 'libmamba' (choose from 'classic')
