Evaluation Datasets

Evaluation datasets store and manage all evaluation task data. They can be created from an empty dataset or by uploading a local file.

​

Creating an Evaluation Dataset

Go to the Evaluation Datasets page and click Create Evaluation Dataset, then complete the following steps:

1. Fill in the basic information: Enter the evaluation dataset name and description.
2. Select the dataset type: Choose the type: Agent / Workflow / Custom. When you select Agent or Workflow, you must also select the specific agent or workflow and its version.
3. Edit the table schema: The system parses the input and output fields of the selected agent / workflow, or uses the default custom fields. You can then edit the table schema—configuring column name, data type, whether the field is required, and description.
4. Import the dataset file: Import your prepared evaluation dataset file. You can also download the data template first and fill it in.
   Supported file extensions: .xls, .xlsx. Maximum file size: 20 MB.
5. Confirm the upload: After clicking Confirm, the uploaded file appears on the Evaluation Datasets page.
6. Publish the evaluation dataset: Once the Import Status changes to Import Successful, click Publish to publish the dataset.

   Draft evaluation datasets cannot be used for application evaluation. They must be published before use.

​

Managing Evaluation Datasets

Go to the Evaluation Datasets page and click View for the target dataset to open its details page. The details page supports the following operations:

• Version management: Manage dataset versions—the draft version and the 10 most recent versions are displayed.
• Content search: Search within the dataset by selecting All Fields or a specific dataset header field, then entering the search content.
• Edit headers: For headers that have already been added, only the name and description can be modified. You can also add new header fields.
• Add and export data: Add data one record at a time or import in batch, and export the current dataset data.