Running RAPIDS
RAPIDS is an NVIDIA toolset for preparing and executing end-to-end data science and analytics pipelines on GPUs. It's implemented as a Python library.
To use RAPIDS on Neu.ro, you can run it in conjunction with Jupyter.

Quick start

We created a preconfigured project that you can download and use for running RAPIDS.
Just copy the repository and run the following command in the CLI:
1
$ neuro-flow run jupyter
Copied!
This will open Jupyter in your browser, and you can start working with RAPIDS.

Running RAPIDS in your own project

To run RAPIDS in your own Neu.ro project, you will need to add the following live action description in its .neuro/live.yml file:
1
kind: live
2
title: <action_title>
3
4
jobs:
5
jupyter:
6
title: "<job_title>"
7
image: rapidsai/rapidsai:cuda10.1-runtime-ubuntu18.04-py3.8
8
preset: gpu-v100-small
9
http_port: 8888
10
http_auth: false
11
browse: True
12
detach: True
13
life_span: 3h
Copied!
Now, when you run neuro-flow run jupyter in the CLI, an instance of Jupyter will be opened in your browser, containing all necessary RAPIDS notebooks from the rapidsai/rapidsai:cuda10.1-runtime-ubuntu18.04-py3.8 image.