1. Create Project#

An Efemarai project groups together multiple datasets, models and operational domains that you use when working on a given ML problem. We recommend setting up a project per source code repository where you develop your ML solution.

To list the projects you have associated with your account, you can run

ef project list

The result should be empty, as you’ve just created your account.

Let’s create our first project.

Project structure#

Let’s assume this is a standard ML project structure. You have a code, data and weights folders with corresponding files within.

├── code
│   ├── intialize.sh
│   ├── model.py
│   ├── dataset_utils.py
│   └──
├── data
│   ├── train
│   ├── test
│   └── val
├── weights
│   ├── checkpoint-1.pt
│   └── checkopint-2.tf
└── Dockerfile

In order to use Efemarai with it, you have to create a base file that describes the project and functionality within. This file is called efemarai.yaml and should sit in the root of the folder tree.

Efemarai.yaml file#

This is how a minimalistic project can be defined efemarai.yaml:

  name: Traffic Sign Detection
  description: Implement multiple models for detecting traffic signs.

Once you have created the example yaml efemarai.yaml you can run the following command to create it!

efemarai project create efemarai.yaml

or if you’re using the standard name as above, you can simply type

efemarai project create .

After succesfully executing either of these commands, you will be able to see the project in the home page and in the CLI with

ef project list

Loaded project

For all parameters that can be set for the project, see the reference.

Next, let’s create our first dataset.