Skip to main content

​
Datasets

Datasets contain input/output pairs used for training, evaluation, and testing AI models.

Input Data

Store structured input data that will be processed by your AI models.

Expected Output

Define expected outputs for evaluation and training purposes.

Metadata

Add contextual information and labels to enhance data organization.

Version Control

Track dataset changes and maintain data lineage across iterations.

​
Creating Datasets

Choose from two available options when working with datasets:

​
New Dataset

New dataset creation How to use: Upload a new dataset file using the drag & drop interface. Required:
  • A JSON or CSV file containing your dataset
  • Drag & drop the file into the upload area or click to select from your computer
  • Only .json and .csv file formats are accepted

​
Existing Dataset

Existing dataset selection How to use: Select from previously created datasets in your project. Required:
  • Enable the “Use Existing Dataset” toggle
  • At least one dataset must already exist in your project
  • Select the desired dataset from the table by checking the checkbox next to the Dataset ID

Create Custom Metrics

Build evaluation functions to measure dataset performance

Optimize with Tune

Use datasets to automatically improve prompt performance
⌘I