Skip to content

Releases: NVIDIA/data-science-stack

v2.11.0

23 Aug 18:31
6df0d57
Compare
Choose a tag to compare
  • 5d96f1f moving cuda up to 11.7
  • 52807cb moving cuda to 11.5
  • d3b8a7f fixing bug re: driver version
  • 0f6ebe2 fixing miniconda bug in Dockerfile
  • 0d9b49d moving to CUDA 11.5.0 in Dockerfile
  • 3a81e30 pinning
  • 69916d6 updating conda and notebook versions in the dockerfile
  • 4bbb535 updating notebooks version
  • f6900b2 overcoming bug in miniconda installation script
  • b67a323 updating conda
  • 175369b updating the environment
  • 3f5d77e updating docker
  • c50c357 move to driver 525
  • 7cf0f32 support ubuntu 22.04

v2.9.0

25 Jun 19:57
b8625cf
Compare
Choose a tag to compare
  • docs: Documented WSL v2 support via containers only due to CUDA 11.0 requirement in WSL v2
  • fix: typos in README files
  • fix: remove double quotes in cuda 11+ version reporing
  • fix: checking that conda is installed before run-notebook
  • fix: checking that the notebook exists in run-notebook
  • fix: temporarily removing nvdashboad from create conda env pending jupyter 3.0 compatibility
  • fix: quote termination in script
  • fix: not including nvdashboard in Jupyter Lab pending support for Jupyter 3.0
  • fix: docker file cuda version should be 11.2.2
  • feat: pin (#115)
  • feat: move nvdashboard to rapidsai channel (#118)
  • fix: docker file cuda version should be 11.2.2
  • feat: pin (#115)
  • feat: move nvdashboard to rapidsai channel (#118)
  • feat: remove chainer (#117)
  • feat: move to RAPIDS 21.06 (#112)
  • feat: move to pytorch 1.8 (#113)
  • feat: remove fastai (#110)
  • feat: tf2 support (#111)
  • feat: notebooks 21.06 (#114)
  • feat: back to conda solver(#116)
  • feat: #105 update driver to 465
  • feat: #106: update rapids version
  • fix: dev2.9 tag
  • feat: update rapids notebooks version (#107)
  • feat: dev 2.9 tag (#108)
  • Merge pull request #103 from hoanghphan/dev2.9
  • add object_detection example
  • benchmarks renamed to examples

v2.8.0

31 Mar 22:21
4d2edc0
Compare
Choose a tag to compare
  • update versions to 2.8.0
  • docs: automated install on RHEL 8.x
  • fix: aws cli purge must force
  • docs: readme for the pytorch lightning example
  • docs: added readme for the knn example
  • feat: rapids cuml benchmarks
  • fix: notebook version 0.18
  • fix: pinning pl ver for benchmarks
  • fix version comparison for driver_install
  • fix: purge ngc cli
  • fix: WSL v2 Ubuntu image does not have unzip
  • feat:documentation (#83)
  • feat: pins (#97)
  • feat: bump ppa driver to 460 (#96)
  • fix: aws cli
  • TF2 env (#43)
  • feat: AWS CLI (#65)
  • feat: kaggle docker image (#69)
  • feat: kaggle cli (#93)
  • feat: ngc cli (#72)
  • feat: notebooks 0.18 (#95)
  • feat: RAPIDS 0.18
  • feat: microbenchmarks (#78)
  • update unsupported message for install_driver
  • switch from rpmfusion.org to nvidia.com
  • add install_driver() for rhel8
  • fix dual detection, WSL is not mutually exclusive w/ /etc/os-release
  • fix ubuntu, be consistent w/ OS_RELEASE
  • fix issue where nvidia-docker relies on full distro version
  • fix compound VERSION_ID issue, rhel8.3 -> rhel8
  • fix redhat -> rhel to match os-release
  • remove redundant sourcing of /etc/os-release
  • use /etc/os-release to detect OS_FLAVOR and OS_RELEASE
  • refactor changes to .bashrc in install_cuda()
  • check for existence of pinned environment before trying to create a conda env from it #87
  • fix 'git clone' url
  • feat: upgrade the min driver and docker versions (#79)
  • feat: update STACK_VERSION to 2.8.0-dev (#82)

v2.7.0

03 Feb 20:51
3eb6f24
Compare
Choose a tag to compare

feat: update versions to 2.7.0
fix: pinning jupyterlab to 2.2.9 in the env because of nvdashboard dependency
fix: permissions on the data-science-stack script
fix: upgrade command requires bash
fix: backup old versions of envs and docker images on upgrade
fix: pin the conda envi
doc: updated readme for 2.7.0 (#73)
feat: self-upgrade (#57)
feat: support for repo2docker (#54)
feat: beta support for WSL (#28)
feat: autocomplete (#53)
feat: Proper return codes (#56)
fix: Pull 0.17 notebooks for RAPIDS 0.17
feat: 2.6.0 -> 2.7.0-dev

v2.6.0

18 Dec 23:16
ba575ba
Compare
Choose a tag to compare

feat: 2.6.0-dev -> 2.6.0
feat: Add adlfs, dask-glm, dask-ml, and jupyterlab-nvdashboard to default environment
feat: Respect NOTEBOOKS_DIR if set in the user's environment (#49)
feat: Update default stacks to RAPIDS 0.17
feat: Start 2.6.0-dev branch

v2.5.1

11 Dec 02:04
Compare
Choose a tag to compare

feat: 2.5.0 -> 2.5.1
docs: container data mounting, OS Support, typos

v2.5.0

21 Nov 00:42
d2fb31c
Compare
Choose a tag to compare

Changes since v2.4.0 (latest first)

feat: 2.5.0-dev -> 2.5.0
feat: disable disk swap when installing Kubernes, and ask to turn on in purge
fix: logic around purge-driver y/n prompt always purges
feat: Kubernetes stack installer
docs: Add FAQ entry about connectivity and firewalls
feat: Command to run all notebooks in a directory
feat: Update to CUDA 11
feat: OS flavor and release logic, only run on supported OSs
fix: driver detection when driver broken
fix: Red Hat 8 docker-install, add enable and start of daemon
feat: additional stop-on-error around install commands
docs: Specify in README that Red Hat Workstation is the supported flavor
feat: After build-* commands display message about run- commands
feat: log output to data-science-stack.log
feat: Update default stacks to RAPIDS 0.16
feat: Update to 455 driver
fix: purge-driver bug on Ubuntu 20.04
feat: Start 2.5.0-dev branch

v2.4.0

20 Aug 23:55
Compare
Choose a tag to compare

Since v2.3.1 (latest first)

  • feat: 2.4.0-dev -> 2.4.0
  • fix: correct, explain, or reword various output and help messages
  • feat: latest RHEL7.x and 8.x support
  • fix: add jq to setup-system
  • fix: conda version comparison fixed for fresh install, rm miniconda.sh temp file
  • fix: remove docker-adduser command, replaced with setup-user
  • docs: Document how to remove environments and containers
  • docs: Add section on validating stack and notebook testing
  • fix: yum failures not treated as fatal
  • feat: Support for Ubuntu 20.04, update to 450 drivers
  • feat: Update default stacks to RAPIDS 0.14.1 and re-pin stacks
  • docs: document upgrade process and effects
  • docs: Document creation of custom stacks and debug process
  • docs: Document how to report issues and feature requests
  • docs: Add troubleshooting section with common errors
  • feat: Add commands to remove the container and conda environment
  • feat: don't require setup-user as a seperate step for initial user
  • feat: Additional diagnostics
  • feat: Have version command check GitHub for new releases
  • feat: If older Conda exists, ask user to upgrade instead of new install
  • feat: Update default stacks to RAPIDS 0.14, version to 2.4.0-dev

v2.3.1

23 Apr 23:08
Compare
Choose a tag to compare

feat: display working dir and script when telling user to reboot or logout
fix: detect missing nvidia-smi

v2.3.0

15 Apr 07:11
Compare
Choose a tag to compare

Since 2.2.2:
feat: 2.3 environment files and pinned yamls
fix: remove ipyvolume from labextensions
docs: Update README for 2.3
feat: use channels from .env file in pinned yaml
docs: add list command to help, and display default
feat: on Ubuntu detect .run file was used and guide user to solution
fix: change how notebook repo is cloned and submodules handled.
feat: update versions to 2.3.0 and RAPIDS 0.13.
feat: add list command
fix: display default stack in help message
feat: log command used to pin, output to -pinned.yaml
docs: add CONTRIBUTING.md