> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.ovalbee.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.ovalbee.com/_mcp/server.

# Dataset

> Holds images or videos with their labels. Open it to see what is inside. Other nodes take items from here or put items here.

## Overview

Picks one OvalBee dataset for your flow and shows its assets and annotations on the node card. Pick a dataset in the selector, or type a new name to create one right away. The node takes the dataset's name. Until you pick one, it keeps its current name.

Put it at the start of a flow, so later nodes have a dataset to read from or write to through its **Dataset** input.

```mermaid
flowchart LR
  n1["Dataset"] --> n2["Train/validation split"]
```

## How it works

Pick a dataset in the **Dataset** selector. If you type a name that no dataset has yet, OvalBee creates that dataset right away and selects it. So the input never holds a dataset that doesn't exist.

Connect other nodes to it. Their dataset input then points to this dataset, so nothing has to run first. The node itself has no outputs.

Nothing runs. The node has no **Run** button and starts no job. When the flow reaches it, OvalBee checks that the dataset still exists and marks the node done right away.

## FAQ

#### Does the node change the dataset?

No. It never renames the dataset, changes its contents or changes its ontology.

#### Where do I pick the ontology?

Not here. The nodes that label or write into the dataset set its ontology, for example [Labeling](/node-library/nodes/label-labeling).

#### When is the dataset actually created?

When you configure the node, not when it runs. Typing a new name in the selector creates the dataset right away. So even a new flow already has a dataset you can open before the first run.

#### What does this node do during setup?

The setup assistant asks for **Dataset**. When setup starts the flow, this node finishes on its own and doesn't start the nodes after it.

#### Why does the node fail and say the dataset doesn't exist?

The dataset was deleted after you picked it, so OvalBee stops the node before anything else starts. Pick a dataset again in **Dataset**.

## Inputs

**`Dataset`** `dataset` — required

The dataset the rest of the flow reads from and writes to. Its name is shown on the node card. Key: `DATASET`.

---

## JSON config

Machine-readable node interface for automation and advanced usage.

#### Show JSON

```json
{
  "name": "Dataset",
  "description": "Holds images or videos with their labels. Open it to see what is inside. Other nodes take items from here or put items here.",
  "category": "Datasets",
  "namespace": "ovalbee",
  "templateKey": "dataset_ops/dataset",
  "version": "v2",
  "inputs": [
    {
      "key": "DATASET",
      "label": "Dataset",
      "type": "dataset",
      "description": "The dataset the rest of the flow reads from and writes to. Its name is shown on the node card.",
      "required": true,
      "default": null,
      "visibleWhen": null
    }
  ],
  "outputs": [],
  "runtime": {
    "type": "headless",
    "devices": [
      "cpu"
    ],
    "dockerImage": null
  },
  "automation": {
    "isolated": true,
    "requires_configuration": true
  },
  "widgets": {
    "widget": {
      "id": "asset-preview",
      "settings": {
        "datasetId": {
          "type": "variable",
          "value": "self.inputs.DATASET"
        }
      }
    }
  }
}
```