Dockerhub intelpython update

[peters__kurt]

It seems the dockerhub version of Intelpython hasn’t been changed in over Six months and is “stuck” at 2020.2. Does anyone know when/if it will be updated? I tried myself, by using their dockerfile, but conda fails solving for dependencies for some of the extra items I install In The docker myself: jupyter_core, tensorflow-mkl, scikit-learn, xgboost, daal4py, keras.

I then do a:
Conda install -c conda-forge -y jupyterlab
Any thoughts?

Regards,
Kurt

[RashmiP_Intel · Accepted solution]

Hi,

Thanks for reaching to Intel® Distribution for Python* forum.

The latest version of intelpython is 2021.2.

We had also observed that intelpython images in dockerhub contain 2020.2 version and we are following up with concerned team regarding the same. Meanwhile you can use intel/oneapi-aitoolkit image it contains preinstalled latest version of intelpython(2021.2).

Command for pulling docker image:

docker pull intel/oneapi-aitoolkit

For more information about aitoolkit visit: https://software.intel.com/content/www/us/en/develop/tools/oneapi/ai-analytics-toolkit.html#gs.16omv2

In above docker, 2021.2/base environments, scikit-learn, xgboost, daal4py these libraries are preinstalled. Other you have to install manually.

You can install packages using below command:

conda install -c intel <package-name>

You can also change version of intelpython in dockerfile(pls refer dockerfile I have attached).

I tried to reproduce issue you mentioned regarding installation of libraries, but I can build a docker using dockerfile and not facing any issues in installing dependencies you mentioned.

I am attaching the dockerfile here.

Please try it and if you are still facing issues please share the detailed error.

Regards,

Rashmi

[RashmiP_Intel]

Hi,

We haven't heard back from you. Could you please give us an update?

Thanks.

[RashmiP_Intel]

Hi,

Thanks for accepting our solution. If you need any additional information, please submit a new question as this thread will no longer be monitored.

Regards,

Rashmi