metadata

language:
  - sk
license: odc-by
tags:
  - Slovak
  - Datasets
  - FineWeb2
dataset_info:
  features:
    - name: text
      dtype: string
    - name: id
      dtype: string
    - name: dump
      dtype: string
    - name: url
      dtype: string
    - name: date
      dtype: string
    - name: file_path
      dtype: string
    - name: language
      dtype: string
    - name: language_score
      dtype: float64
    - name: language_script
      dtype: string
    - name: minhash_cluster_size
      dtype: int64
    - name: top_langs
      dtype: string
    - name: fasttext
      dtype: string
    - name: langdetect
      dtype: string
    - name: lingua
      dtype: string
    - name: bad_words
      sequence: string
  splits:
    - name: test
      num_bytes: 111283118
      num_examples: 28720
    - name: train
      num_bytes: 101664155529
      num_examples: 26470482
  download_size: 63706633020
  dataset_size: 101775438647
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
      - split: train
        path: data/train-*

FineWeb2 Slovak

This is the Slovak Portion of The FineWeb2 Dataset.

Known within subsets as slk_Latn, this language boasts an extensive corpus of over 14.1 billion words across more than 26.5 million documents.

Purpose of This Repository

This repository provides easy access to the Slovak portion of the extensive FineWeb2 dataset.

The existing dataset was extended with additional information, especially language identification using FastText, langdetect and lingua using XTranslator package. This language identification aims to filter out the texts that are in languages other than Slovak.

In addition, we also identified adult words within the text, which we added to the column bad_words, which can be used to filter out texts with the problematic content.

Additional information:

fasttext - language identified using FastText
langdetect - language identified using langdetect
lingua - languages along with the proportion of the text in the particular language, using lingua-language-detector
bad_words - list of adult words (Slovak and English) identified in each text

Credit to the Original Work

The dataset is released under the Open Data Commons Attribution License (ODC-By) v1.0, with additional usage subject to CommonCrawl's Terms of Use.

Citation

If you use this dataset, please cite it as follows:

@software{penedo2024fineweb-2,
  author = {Penedo, Guilherme and Kydlíček, Hynek and Sabolčec, Vinko and Messmer, Bettina and Foroutan, Negar and Jaggi, Martin and von Werra, Leandro and Wolf, Thomas},
  title = {FineWeb2: A sparkling update with 1000s of languages},
  month = dec,
  year = 2024,
  doi = {10.57967/hf/3744},
  url = {https://huggingface.co/datasets/HuggingFaceFW/fineweb-2}
}