---
configs:
- config_name: default
  data_files:
  - split: train
    path: data/clean-*.parquet
---

# websoup

This dataset is a collection of websites that were scraped based off of their simplicity (size, quality, obfuscation level) and inlined into simple & large html files.

If you want to build a stochastic parrot that spews random web pages (including js and css in a single output) then this dataset is decent (not perfect by any means).

Load with:

```python
from datasets import load_dataset
ds = load_dataset("lowerkey/websoup", split="train")
```