dvc-cc 0.10.30

This connector is used to combine the work of CC (www.curious-containers.cc) and DVC (Open-source Version Control System for Machine Learning Projects).

DVC-CC is a wrapper for using the tool Data Version Control (DVC) to make it possible to use DVC to run your script in a cloud. To make this idea possible, we wrote a script that is part of a docker image that can:

1. download a git repository,
2. download all required files with your DVC storage server,
3. execute your script, and
4. push the results to GIT and to your DVC storage server.

To assign the right hardware for your need in the cloud, we use Curious Containers (CC). This Software runs on our cloud and manages the cloud.

Installation of DVC-CC

DVC-CC is written in python so you can easily install DVC-CC by using pip. We recommend that you install DVC-CC in a conda environment. You can use anaconda or miniconda. For windows user We recommend this website to install miniconda. Currently DVC-CC does not work under Windows!

You can create, and activate an environment with the following lines:

conda create --name dvc_cc python pip
conda activate dvc_cc

If conda activate dvc_cc does not work, try source activate dvc_cc.

Installation with pip

The following script will install the client on your computer:

pip install --upgrade dvc-cc

If you have problems on windows with "win32file", you need to install pywin32 with conda install -c anaconda pywin32.

Installation from source

If you want to install the latest version from source you can install it with poetry.

git clone https://github.com/deep-projects/dvc-cc.git
cd dvc-cc/dvc-cc
poetry build
pip install dvc_cc-?????.whl # replace ????? with the current version that you build in the previous step.

Get started

Install DVC-CC and take a look at this tutorial.

Tutorials

• Working with jupyter notebooks
• working with sshfs
• DVC-CC Settings
• Working with pure DVC syntax
• Using live output
• An old tutorial

Structure of this repository

Acknowledgements

The DVC-CC software is developed at CBMI (HTW Berlin - University of Applied Sciences). The work is supported by the German Federal Ministry of Education and Research (project deep.TEACHING, grant number 01IS17056 and project deep.HEALTH, grant number 13FH770IX6).