Getting Started With Rapid

In this guide, we'll cover the four steps to launching a production batch. Each section links to an in-depth product guide.

Step 1: Project definition

Set up your project by creating the labeling task. You'll upload your data, create a taxonomy, and write instructions.

You'll need to:

  • Select project type
  • Upload at least a small subset of your data
  • Create the taxonomy for your project
  • Write instructions for the Taskers who will be labeling your data
View Project Setup Details

Step 2: Calibration batch

Before you send off all your data to the Taskers, you'll calibrate your project by running a Calibration Batch. This will allow a single tasker to test your project, and they'll give you feedback on the instructions and attempt to label a small subset of your data.

You'll need to:

  • Create a Calibration Batch for a small subset of your data
  • Audit the results of the Calibration Batch
View Calibration Batch Details

Step 3: Prepare launch

This is the time to refine your project! Review the results of the Calibration Batch & try to perfect your project so that the Taskers can better label your data. You can also run additional Calibration Batches to iterate and get more feedback.

The labels that you get out of your production batch ultimately depend on the quality of the project that you've built. Rapid gives you the data & tools to perfect your project.

You'll need to:

  • Create Quality Tasks
  • Review instruction feedback
  • Look over your project taxonomy, and instructions
  • Upload any additional data you'd want labeled
View Prepare for Launch Details

Step 4: Launch production batch

Once your project is in a good place, it's time for a Production Batch! You can expect higher quality from these batches.

You'll need to:

  • Select all the data you want labeled
  • Sit back and wait for the results to come in!
View Launch Guide
Updated a month ago