Dashboard

Create and manage Pages in the Dashboard, and add Charts to visualize evaluation results.

1. Manage Pages

①Page types

The Dashboard shows evaluation results as charts. Open it from Evaluation Dashboard in the left menu. A Page groups multiple charts on one screen. The Page list is split into two groups.

Group	Pages it contains	What you can do
`Default Page`	`Run Report`	View only. You can't create, delete, or rename it.
`Custom Page`	Pages you create	Create, rename, and delete Pages Add charts

When you open the Dashboard, Run Report appears first. Eval Report covers it in detail.
The top of a Custom Page shows its creation date (Created at).
The number of charts appears in Contained Chart.

1-1Create a Page
1. Click + New Dashboard in the left list.
2. The new Page is added to the list and opens right away.
- Names are assigned automatically, such as Page 1 and Page 2.
- You can create only Custom Page Pages. See Page types.
1-2Rename a Page
1. Click the edit icon next to the Page name.
2. Enter a new name in Page Name.
3. Click Save.
1-3Delete a Page
1. Click Edit Page.
2. Click Delete Page.
3. Click Delete in the confirmation dialog.
- You can delete a Page only when there are 2 or more Pages.
- The charts in the Page are deleted too, and you can't undo this.

2. Add a Chart

①Chart creation flow

Click + Add Chart to open the New Chart dialog. Four steps appear in order at the top of the dialog.

Step	What you set
`AI Input`	Describe the analysis you want in a sentence to fill in the remaining settings at once.
`Evaluation Scope`	The analysis scope for the chart
`Run / Version`	The Runs and versions for the chart
`Chart Configuration`	Chart type, axes, title, and description

This document shows how to click Manual Setup and choose each step yourself.
You can also review and edit settings filled in by AI Input in Chart Configuration.
Example sentence: Compare account transfer execution metrics by version in the latest Run

②Analysis scope

The analysis scope (Evaluation Scope) is the range of evaluation results the chart draws. Select it by drilling down the Criteria structure one level at a time.

Criteria › Task Group › Sub Group › Task › Metric

A Task Group is an evaluation area, such as General Performance, Policy & UX, or Safety.
Modules appear as separate Task Group entries, such as Router Module.
At the lowest level you drill down to, select items with the checkboxes. The selected items become the chart's scope.
(For example, if you select 3 Tasks at the Task level, the chart includes only the results of those 3 Tasks.)
By default, all of the Application's Task Group entries are selected.

Evaluation Core Concept explains the Criteria structure.

③Run and Version

A Run is a record of one evaluation run with selected Tasks, Metrics, and Versions. Runs appear by number, such as Run #142.
A Version is a version of the evaluated Application or Module (e.g., Agent v2.4, Router v4).
A/B Test Version shows only the versions evaluated in the selected Runs.
Versions are grouped by analysis scope (e.g., Select Application Version, Select Router Version).
If you don't select any Run, the version list doesn't appear.

④Chart types

You can choose from four types in Chart Type.

Type	Use	Example
`Bar Chart`	Compare scores across Tasks or Metrics.	Compare account transfer metrics by average score per version
`Line Chart`	See how scores change across Runs.	Show the Safety average score per Run, split by version
`Radar Chart`	See the balance of scores across multiple items at a glance.	Compare the score balance across evaluation items between versions
`Heatmap Chart`	Use cell colors to find where scores are low.	Average score by evaluation item and version. Yellower or redder cells are lower

You can draw a Radar Chart only when there are 3 or more X-axis items.

⑤Axes and measures

A chart is drawn with the three settings below.

Setting	What you set	Available values
`Measure`	The value plotted on the vertical axis (Y-axis)	`Score Average`, `Incorrect Count`, `Data Count`, `Score Min`, `Score Max`
`X-Axis`	The items laid out on the horizontal axis	`Evaluation Scope`, `Run`, `Version`, Metadata Key
`Group By`	The basis for splitting within one item	Same list as `X-Axis`. To skip splitting, choose `None`

The Measure values are, in order: average score, number of incorrect data items, number of data items, minimum score, and maximum score.
(For example, set X-Axis to Run and Group By to Version to split each Run's scores by version.)
A Metadata Key is a Metadata column in the Benchmark Set (e.g., difficulty, channel).
The Dashboard shows the latest Metadata values.
Click the button next to X-Axis and Group By to swap the two values.

2-1Select the analysis scope
1. Click + Add Chart. If this is your first time, read Chart creation flow first.
2. In the first step, click Manual Setup.
3. Select the scope to analyze, starting from the Criteria. See Analysis scope.
4. Use the checkboxes to select the items to include in the chart.
5. Click Next.
Click › next to an item to drill down to Sub Group, Task, and Metric.
2-2Select Runs and Versions
1. Select the Runs to analyze in Select Run. See Run and Version.
2. Select the versions to compare in A/B Test Version.
3. Click Next.
By default, all Runs and versions are selected. To remove one, click ✕.
2-3Select a chart type
1. Choose a Chart Type. Chart types describes the use of each type.
2. Choose Measure, X-Axis, and Group By. See Axes and measures.
3. Check the chart in Preview.
If no data matches the conditions, Preview shows a message.
2-4Enter Chart details
1. Enter a Chart Title and Description.
2. Click Add Chart. The chart is added to the Page.
Description can be up to 255 characters.

Frequently asked questions

Why can't I see the Delete Page button?Because it appears only after you click Edit Page. It doesn't appear if there's only one Page.

Cause: You can delete a Page only in edit mode. You can't delete the last remaining Page.
Solution: Click Edit Page. If there's only one Page, create a new Page, then delete it.

Why does Preview show a message instead of a chart?Because the chart can't be drawn with the conditions you selected.

Cause: This happens when there isn't enough data or the conditions don't fit the chart type. For example, a Radar Chart needs 3 or more X-axis items.
Solution: Change one of X-Axis, Group By, the analysis scope, or Chart Type. See Chart types for the requirements of each type.

Why does the chart show only some items?Because the number of items exceeds what the chart can show.

Cause: If there are too many items, the chart can't fit them all.
Solution: Change one of X-Axis, Group By, the analysis scope, or Chart Type to reduce the number of items.