# Autodistill: Train YOLOv8 with ZERO Annotations

**URL:** <https://discuss.roboflow.com/t/autodistill-train-yolov8-with-zero-annotations/2401>\
**Category:** 🤝  Community Help\
**Created:** [June 9, 2023, 10:46am UTC](https://discuss.roboflow.com/t/autodistill-train-yolov8-with-zero-annotations/2401 "2023-06-09T10:46:00Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Elia\_Wise](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/elia_wise/32/2120_2.png) [@Elia\_Wise](https://discuss.roboflow.com/u/Elia_Wise)\
**Post date:** [June 9, 2023, 10:46am UTC](https://discuss.roboflow.com/t/autodistill-train-yolov8-with-zero-annotations/2401/1 "2023-06-09T10:46:00Z")

</div>

I’m trying to create a distilled model of SAM [Google Colab](https://colab.research.google.com/drive/1PKc5x1XjxOPP-2msFp6uMSULvnm9dorw?usp=sharing)

I’m able to label the data and train, but the predict doesn’t find anything

Please help

The purpose is to find bounded area in a document

SAM can do it well but it’s heavy and slow, therefore I’m trying to create a Autodistill YOLOv8 that will find the bounded area in a document

I’m not sure that YOLOv8, because the concept of document is foreign to YOLOv8, is there a better way to create a SAM distilled model?

---

<div class="post-metadata">

**Author:** ![Mohamed](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/mohamed/32/62_2.png) [@Mohamed](https://discuss.roboflow.com/u/Mohamed)\
**Post date:** [June 9, 2023, 1:48pm UTC](https://discuss.roboflow.com/t/autodistill-train-yolov8-with-zero-annotations/2401/2 "2023-06-09T13:48:36Z")

</div>

@Elia_Wise did you try the Autodistill workflow? And were there any issues?

Asking because we do have a blog post that helps to evaluate prompts for use with Autodistill:

- descriptive prompts of what you’re trying to auto-label will help with improving success, maybe this project’s classes offers ideas that help for formulating your prompts?: [Table Extraction PDF Object Detection Dataset and Pre-Trained Model by Mohamed Traore](https://universe.roboflow.com/mohamed-traore-2ekkp/table-extraction-pdf) (example: `bordered table`, `borderless table`, `data table`, `data table cells`)

![image](https://canada1.discourse-cdn.com/flex029/uploads/roboflow1/original/2X/b/b781fdca5afe73b0711e3474690746300673d733.png)

- more documents projects here: [Documents Datasets and Pre-Trained Models](https://universe.roboflow.com/browse/documents)

> **[How to Evaluate Autodistill Prompts with CVevals](https://blog.roboflow.com/autodistill-prompt-evaluation/)**
>
> In this guide, learn how to evaluate prompts for use with Autodistill with CVevals.

This will help you to get a better idea of what the best prompt will be for labelling the document and/or parts of the document you are looking to identify and classify with the model.

> **[Distill Large Vision Models into Smaller, Efficient Models with Autodistill](https://blog.roboflow.com/autodistill/)**
>
> Autodistill is a new ecosystem of packages that enable you to distill knowledge from large vision models into smaller, edge-ready models.

> **[GitHub - autodistill/autodistill: Images to inference with no labeling (use...](https://github.com/autodistill/autodistill/#-available-models)**
>
> Images to inference with no labeling (use foundation models to train supervised models) - GitHub - autodistill/autodistill: Images to inference with no labeling (use foundation models to train supe...

 ![image](https://canada1.discourse-cdn.com/flex029/uploads/roboflow1/original/2X/c/c705fe8c60bc1bca3c503cbd53cfdea6c2334583.png)

---

<div class="post-metadata">

**Author:** ![Elia\_Wise](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/elia_wise/32/2120_2.png) [@Elia\_Wise](https://discuss.roboflow.com/u/Elia_Wise)\
**Post date:** [June 10, 2023, 4:38am UTC](https://discuss.roboflow.com/t/autodistill-train-yolov8-with-zero-annotations/2401/3 "2023-06-10T04:38:52Z")

</div>

I tried the [Table Extraction PDF Object Detection Dataset and Pre-Trained Model by Mohamed Traore](https://universe.roboflow.com/mohamed-traore-2ekkp/table-extraction-pdf)  
The precision is OK, but I was not able to get any data about the cell structure  
Is it the original model from the article?