Skip to content

Getting Started

You can create datasets through the SYNTH UI or SDK. The usage is the same for on-premise deployments and cloud use. See On-premise vs Cloud.

Requirements

Synth On-premise license or access to the Synth Private Beta

Creating a dataset

When you login the first page you see will allow you to create an organization. An organization is required to create datasets. Once your organization is created you will see the following: Create a dataset

Click the "Create Your First Dataset" button and you will have a choice between creating a synthetic dataset or a document ingestion pipeline (coming soon). Click on "Create synthetic dataset" and you will be routed to a page that looks like this: Create a synthetic dataset

Configure your dataset for your needs and follow the steps till the end to start the synthetic dataset generation. Once submitted and generation finishes, you will see a completed dataset in the datasets page: Datasets page with completed dt

You should then click into the dataset, and on the top right of the page there will be a download button: Datasets download btn

This will download the parquet file that you can use on your training runs!