# Training Line Fields/Table Training

## Table Training: When and How to Use It

DocBits uses AI-based table extraction by default, which works across suppliers. However, when you need more control over how line items are extracted for a specific supplier, you can use **Table Training** to define the structure manually.

#### **Important: Supplier-Specific Training**

Table training is **specific to each supplier**. If you train a table layout for one supplier, the rules apply only to that supplier’s documents.\
Other suppliers will continue to use AI-based table extraction unless trained separately.

#### When Should You Use Table Training?

Use Table Training **only when the document layout is stable**, and you want to ensure consistent, high-quality extraction. It is best suited for suppliers who:

* Use a **fixed invoice format** over time.
* Require **precise mapping** of line items to specific fields.
* Are **high-volume suppliers**, where consistent automation is valuable.

#### When Not to Use Table Training

Avoid using Table Training if:

* The supplier **frequently changes their invoice layout** (e.g., every couple of months).
* You do not regularly receive documents from that supplier.
* The default AI extraction works **well enough** without manual training.

In such cases, retraining the table repeatedly would be inefficient and time-consuming.

## How to use It:

**Training Mode** is used to teach DocBits how to correctly extract line items from a specific supplier’s document layout.

When you're in Training Mode, you define:

* **Where the table is located** on the document.
* **Which areas correspond to each column** (e.g., Quantity, Description, Unit Price).

This mode creates a **reusable extraction template** for that supplier, so future documents with the same layout will be processed automatically — no need to retrain.

**Key Characteristics:**

* **Mapped columns are read-only** — you can’t manually change the text.
* You **cannot add or delete rows**.
* The focus is on **mapping positions**, not correcting values.
* Once you’ve finished mapping the table and its columns, click **“Save Rules”** so DocBits learns and applies this training to similar documents from the same supplier.

**Correction Mode** allows you to **fix extraction errors** in the table data for the current document. It’s used when DocBits has already attempted to extract the table, but the results need small corrections.

Use Correction Mode when:

* Some **values are missing or misaligned**.
* An **empty or duplicate line** was extracted and needs to be removed.

**What You Can Do in Correction Mode:**

* **Manually adjust values** in mapped columns (e.g., fix misplacements).
* **Delete rows** that shouldn’t be part of the table.
* **Add rows** that were missed entirely during extraction.


---

# Agent Instructions: Querying This Documentation

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

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.docbits.com/administration-and-setup/setup/document-training/training-line-fields-table-training.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
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.