README.md

---
annotations_creators:
- no-annotation
language_creators:
- crowdsourced
license: cc-by-sa-4.0
task_categories:
- text-generation
- fill-mask
task_ids:
- language-modeling
- masked-language-modeling
source_datasets:
- original
language:
- en
pretty_name: Lecture Gratuits

configs:
  - config_name: default
    data_files:
      - split: final
        path: data/*
---

# Dataset Card for LectureGratuits

![](Recursalberg.png "Clara makes her return. See recursal/Recursalberg for her description!")

*Waifu to catch your attention.*

## Dataset Details

### Dataset Description

*LectureGratuits* is a cleaned dataset of [*Ebooks Gratuits*](https://www.ebooksgratuits.com/) books. We downloaded all the publicly available ebooks books at the time and processed them.  
Filtering to a total amount of tokens of **~265.46M** (llama-2-7b-chat-tokenizer) / **~253.51M** (RWKV Tokenizer) from primarily English language.  

- **Curated by:** Darok
- **Funded by:** Recursal.ai
- **Shared by:** KaraKaraWitch
- **Language(s) (NLP):** English
- **License:** Public domain

### Dataset Sources

- **Source Data:** [ebooksgratuits.com](https://www.ebooksgratuits.com)

### Processing

KaraKaraWitch doesn't have specifics on how it's processed. We have postiluated the following workflow / processing:

0. Get the higher ID
1. Enumerate and download all the epub files: `https://www.ebooksgratuits.com/newsendbook.php?id=<ID>&format=epub`
2. Put them in a folder called `books`
3. extract content to each json file in `output` folder. (See filtering steps in `extract-text.py`)
4. Combine into a single file.

### Data Keys

```
text (str): The book's text. Converted to markdown.
meta (dict): A dictionary of metadata with the following keys:
  - title
  - author
  - publisher
```

### Dataset Curators

This dataset was mainly Darok's work. I (KaraKaraWitch) only assisted them with questions and the writing of the dataset card.

### Licensing Information

The books itself is in public domain. For the post processed data under Recursal work, it's licensed as CC-BY-SA.

Recursal Waifus (The banner image) are licensed under CC-BY-SA. 
They do not represent the related websites in any official capacity unless otherwise or announced by the website. 
You may use them as a banner image. However, you must always link back to the dataset.

### Citation Information

```
@ONLINE{lecturegratuits,
  title         = {LectureGratuits},
  author        = {Darok, KaraKaraWitch, recursal.ai},
  year          = {2024},
  howpublished  = {\url{https://huggingface.co/datasets/recursal/Recursalberg}},
}
```