> For the complete documentation index, see [llms.txt](https://docs.stackai.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.stackai.com/technical-reference/evaluations-and-observability/evaluation-datasets-and-input-mapping.md).

# Evaluation Datasets and Input Mapping

An evaluation dataset pairs inputs with the information needed to judge results. Map dataset fields to the project’s actual input and output contract.

{% embed url="<https://files.gitbook.com/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSQ1Sgy933krTyK0S4JpP%2Fuploads%2FR1fKiLkMkVbzYPlRuqhm%2FG081.mp4?alt=media&token=4dfec768-94a0-4ab0-9dc7-e5cc1b9b7621>" %}

## Before you begin

An accessible project and permission to create or update its evaluation dataset.

## Evaluation concepts

| Term               | Definition                                                                                                                                                                |
| ------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Evaluation dataset | A collection of input cases and any expected outputs needed for assessment. Its mappings must match the project inputs and selected output.                               |
| Input field        | A value collected from a user or client for execution. Map its displayed label to the correct workflow input key.                                                         |
| Expected output    | The reference value used by an evaluator when its method requires one. It represents the desired behavior, not necessarily a verbatim answer for every evaluation method. |
| Schema             | A definition of fields, types, required values, and supported constraints. Use it to validate the shape of data; it does not establish factual correctness.               |
| Batch row          | One input record processed through the selected project. Check the row’s status and result, not only the batch-level indicator.                                           |

## Prepare an evaluation dataset

1. Select the project inputs and output to evaluate.

![Select the workflow inputs and expected-output field for the dataset.](https://content.gitbook.com/content/SQ1Sgy933krTyK0S4JpP/blobs/19QkFuXqBZkdFA8ZF4Uu/assets/g081-g081-02-create.png)

Select the workflow inputs and expected-output field for the dataset.

2. Add rows with valid input values and expected output where required by the evaluator.
3. Review the mapping before generating results.

Each dataset row can run against the intended project fields.

## Behavior and limitations

A dataset field label is not necessarily a node ID. An input mapping that does not match the project can fail before scoring. Through the public API, append requests accept 1–100 rows and repeated appends can duplicate rows. Preserve row IDs when updating existing data.

## Related resources

[Create an evaluation dataset](https://docs.stackai.com/guides-and-quickstarts/evaluations-and-observability/create-an-evaluation-dataset.md)

[Evaluators, Scoring, and Pass Criteria](https://docs.stackai.com/technical-reference/evaluations-and-observability/evaluators-scoring-and-pass-criteria.md)


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.stackai.com/technical-reference/evaluations-and-observability/evaluation-datasets-and-input-mapping.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
