EvaluationDatasets
Datasets
Create and manage evaluation datasets and dataset runs from the dashboard, SDK, or API.
Datasets
Datasets store collections of input/expected-output pairs used to evaluate LLM pipelines. Each item has an input, an optional expectedOutput, and optional context and metadata.
You can manage datasets from:
- Dashboard UI — create datasets, add items (single or CSV import), run evaluations, and export results from Datasets in the left sidebar.
- SDK — TypeScript, Python, and Java SDKs for programmatic dataset management.
- REST API — see the Datasets API reference.
Manage Datasets
Create, list, rename, duplicate, and delete datasets.
Dataset Items
Add items individually, upload from CSV, archive, and export.
Build with Assist
Let the Assist AI agent generate new items and edit existing ones conversationally.
Dataset Runs
Create runs with prompts or APIs, upload CSV results, and compare runs.