🎓Learn Coretex

Building blocks of Coretex.ai platform

When working with Coretex.ai there are couple of modules you should get used to. The diagram below shows those modules and how they interact with each other, so you can paint a mental picture of the complete information flow while moving around.

Interaction of different entities in Coretex.ai platform

The first thing to know is that as a Coretex user you belong to a specific Organization. You can think of this as a company, team, department, institution or any other organizational entity your account belongs to.

If you are registering as an individual user - don't worry - we will create a default organization with you as one and only member.

Organizing your work is made simple with the top most module called Project This module defines the scope of your data, runs and results, storing them under one logical group.

Think of a Project as a domain for your work - once you're about to start a new unit of work you first need to create a new Project for it, give it a name and indicate a Type you're attempting to solve. Everything you create in Coretex will reside inside of that Project, so when you share it with your teammates they will instantly get access to all of the content in that Project.

Once created, a Project can only change its name, description and visibility. Make sure you select the correct Type associated with a Project when creating it, since it will determine the type of Datasets you can add to it.

One Project can contain multiple Tasks, Datasets and Runs.

A Dataset is a named collection of data Samples of a certain type (images, IMU data, other categorized data) which you can upload to the platform or generate from scratch. While there are no limits to what kind of datasets you can work with, there is a set of supported formats we have built a UI around to streamline and simplify your data management. Please check the Types page for more details on supported data types.

To execute runs on your data, you first need to define a Task a module responsible for data processing. At its core, Task is a collection of code and configuration files describing how your data will be handled and how the results of runs will be stored . Run is a single execution of that code. Refer to the Execute Run section of the tutorial for more information on executing runs.

Coretex, among other things, manages how and when your scheduled Run requests will be fulfilled.

Successful completion of an Run generates several different results, depending on the Type used. There are three main modules representing results: Models, Artifacts and Metrics.

A model is a file generated from runs incorporating ML neural networks training, and it can be used to make predictions based on certain data input (i.e. Computer Vision).

In addition to models, runs results can be represented in the form of Artifacts, which range from result summaries to full run reports and any format specified by the Task used.

Metrics are information on run execution, including accuracy, loss, CPU and Memory usage, disk I/O etc.

These modules and their interactions simplify the iterative process of data gathering, data manipulation, experimenting and result analysis. Each module is explained in more details in the following pages.