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

# Workspaces

A workspace provides a context for organising model development. Select it together with the dataset and configurations when you create a training job.

Use the same workspace for related training runs so that their models share a project context. The workspace identifies where a run belongs; the dataset and configurations determine what the run trains and how it runs.

## Before you begin

Sign in to the account where you will train the model. Check **Model Lab > Workspaces** for an existing workspace for the project. Reuse it when appropriate; a separate workspace is useful for a different modelling project.

## Create a workspace

1. Open **Model Lab > Workspaces** and select **Create**.

2. Enter a **Workspace name**, then select **Create**.

3. Verify that the workspace appears in the list before using it for model development.

![Workspace name and Create control](/_fern-files/_img/795dbe0f320aef1b48125d6f6c5f351f81a77d5d248eca24083589e093223270.webp)

## Review a workspace

1. Return to **Workspaces** and use **Search workspaces…** to find the name. Clear the search to see the full list, or use the pagination controls to move between pages.
2. Select the workspace name. In **Overview**, check the name and ID and review the **Audit trail**.
3. Use **Details** when you need the complete resource record. Copy the workspace ID when recording which workspace a training run uses.

The list shows **Created at** and **Updated at**. Refresh the list if a newly created workspace has not appeared. A matching name helps locate a workspace; use the ID to distinguish records when names are similar.

## Use a workspace for training

Select the saved workspace in the **Workspace** field when creating a training job. After the job completes, check the workspace reference on the output model. Creating a workspace does not create a dataset or start a training job.

## Delete a workspace

Select a workspace in the list and use the delete control, or open the workspace and select **Delete**. Review the selected workspace before confirming. Deletion is permanent and may affect associated models, jobs and configurations; check those dependencies before removing a workspace.

## Next steps

Prepare optional [Model constraints](/ui-guides/models/constraints).