> 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.

# XNN-assisted LLM explanations

The XNN-assisted LLM view generates a written interpretation of built-in Adult Income sample data. Review the sample and the generated claims together before reusing the explanation.

## Before you begin

Use this page to explore the supplied sample-data workflow. It does not query records from your deployment. Choose the sample size appropriate to what you want to inspect, then read the sample before reviewing the narrative.

## Generate an XNN-assisted LLM explanation

1. Select **Explanation Lab > XNN LLM Explanations**.

2. Review the built-in **Adult Income** sample on the **XNN Assisted LLM Explanations** page. This page uses supplied sample data; it does not use your selected deployment's records.

3. Choose **Sample Size**: **1**, **10** or **50** sample rows. The example uses **10 Sample Rows**.

4. Select **Generate Explanation**.

5. Review the **Sample Data** table and wait for the **LLM Explanation** to finish.

6. Read the generated sections and check their claims against the **Sample Data** table.

## Interpret and review the result

The example uses ten sample records and produces sections including **Summary of Key Trends**, **Data Anomalies and Implications**, and **Summary Table**. Generated wording may vary between runs.

## Regenerate, copy or export

Use **Regenerate** to request another narrative, **Copy** to copy the generated text, or **Export** to choose a download format. The export options include Markdown and PDF. Review the text before reusing it.

The generated text is an interpretation to review. Statements about population patterns, anomalies, fairness, or causality require supporting evidence. A ten-row sample is not enough to establish those conclusions for the full population.

Before reusing an explanation, retain the sample data context and sample size, verify quoted numbers and distinguish the generated interpretation from live model results. To explain your own deployed model, use [Local explanations](/ui-guides/explanations/local-explanations), [Model structure and decision rules](/ui-guides/explanations/decision-rules) and [Symbolic explanations](/ui-guides/explanations/symbolic-explanations).

![The XNN-assisted explanation uses ten sample rows.](/_fern-files/_img/8c246107bc0768dcb5ec87a43e596817bdce07ca9720fe64c33162bd8aa79b97.webp)

![Generated LLM text with Copy and Export controls. Review its claims before reuse.](/_fern-files/_img/4eeed9baa9a630adab738ff7eec2d4f6bb2dd2d338de79ef500190228adec96b.webp)

## Next steps

Continue to [Symbolic explanations](/ui-guides/explanations/symbolic-explanations).