Tables

A Table is perhaps the most important object you will work with in 3LC. It represents a collection of rows belonging to a dataset, and is used as a means to work with your data both in the Dashboard and the Python package. It allows you to visualize, analyze and modify your data in the Dashboard, and to consume those changes directly in your Python code through tlc.Tables, in place of for example PyTorch Datasets.

This section covers the essential aspects of working with Tables. You’ll learn how data is described through Schemas, how to create Tables and how to consume their data as rows and samples. We’ll explore how Tables can be thought of as revisions of your data, enabling you to track changes in a Git-like fashion. Further topics like bulk data handling, caching and data export are also covered.

In the Dashboard, double click a row in the TABLES tab to open and view that Table:

On opening a Table, the row data contained within the Table is shown in the bottom panel. The current Table’s name is shown in the upper left corner.