Overview

CSV format is a lightweight option to train state machines. It relies only on Input, Output, and Reference data.

This format is ideal for data labelling, routing, and other simple models that don't require generative outputs or complex input patterns.

Use this section to understand the CSV schema, how rows relate to each other, and how values are inferred.

Download templates

Download CSV template Download XLSX template

CSV shape

Every CSV uses the following three columns (all are required in the header row):

• Input
• Output
• Reference

A minimum of 10 rows are required. More is better.

You may add extra columns such as Notes to aid development. Only items in Input, Output, and Reference are used to train the model.

Where to go next

• Review the Columns section for column-by-column details.
• Use the Examples page to store full example CSVs.