# Training

Training your models is an integral part of Typless. Since you are the author of the document types and the structure is not known in advance, you have to [build a dataset](https://typless.gitbook.io/typlessapi/typless/training/building-a-dataset) for your documents that you are automating. Providing context is done by feeding the data to Typless, which then starts recognizing and building connections between the documents.

### Number of documents for precise extraction

<table><thead><tr><th width="313">Quality</th><th width="229">Metadata</th><th>Line items</th></tr></thead><tbody><tr><td>Under 300DPI, skewed scans, ...</td><td>1 - 4</td><td>1 - 7</td></tr><tr><td>300 DPI, original PDFs, ...</td><td>1 - 3</td><td>1 - 5</td></tr></tbody></table>

{% hint style="info" %}
**📘&#x20;**<mark style="color:blue;">**Train using just values - No additional data needed**</mark>

Forget about special annotation or template building process. Typless takes care of everything - Automatically!
{% endhint %}

![Typless training cycle - No human intervention](https://files.readme.io/89a779c-capture_3.PNG)

### Triggering training

{% hint style="success" %}
**👍&#x20;**<mark style="color:green;">**Training is executed automatically every day at 10 PM CET**</mark>

For **all of your suppliers** with new documents in the [dataset](https://typless.gitbook.io/typlessapi/typless/training/building-a-dataset) of all your document types.\
**Free of charge**
{% endhint %}

If you want to see results immediately, you can trigger the training process on the [Dashboard page](https://app.typless.com). Look for your document type in the list, and click on the![cogs icon](https://typless-public.s3-eu-west-1.amazonaws.com/cogs.png).