README.md

metadata

dataset_info:
  - config_name: de
    features:
      - name: text
        dtype: string
      - name: url
        dtype: string
      - name: domain
        dtype: string
      - name: tlsh
        dtype: string
      - name: text_len
        dtype: uint32
      - name: lang_identification_prob
        dtype: float32
      - name: harmful_pp
        dtype: float32
      - name: warc_record_id
        dtype: string
      - name: warc_date
        dtype: string
    splits:
      - name: '2021_17'
        num_bytes: 16614818320
        num_examples: 7316629
      - name: '2021_21'
        num_bytes: 329787348
        num_examples: 164456
      - name: '2021_25'
        num_bytes: 13490283920
        num_examples: 5359845
      - name: '2021_39'
        num_bytes: 12298808823
        num_examples: 4856931
      - name: '2021_43'
        num_bytes: 19052080067
        num_examples: 6802704
      - name: '2022_05'
        num_bytes: 12019070936
        num_examples: 4033390
      - name: '2022_33'
        num_bytes: 8787302552
        num_examples: 3263896
      - name: '2024_18'
        num_bytes: 9749883212
        num_examples: 4881888
      - name: '2024_22'
        num_bytes: 8410414005
        num_examples: 4280514
    download_size: 100752449183
    dataset_size: 403009796732
configs:
  - config_name: de
    data_files:
      - split: '2021_17'
        path: data/2021-17/de/*.parquet
      - split: '2021_21'
        path: data/2021-21/de/*.parquet
      - split: '2021_25'
        path: data/2021-25/de/*.parquet
      - split: '2021_39'
        path: data/2021-39/de/*.parquet
      - split: '2021_43'
        path: data/2021-43/de/*.parquet
      - split: '2022_05'
        path: data/2022-05/de/*.parquet
      - split: '2022_33'
        path: data/2022-33/de/*.parquet
      - split: '2024_18'
        path: data/2024-18/de/*.parquet
      - split: '2024_22'
        path: data/2024-22/de/*.parquet

Some data name

Just testing... this dataset is temporary and will be removed soon.

Usage

from datasets import load_dataset

# init dataset in streaming mode
ds = load_dataset("malteos/clean-oscar-de", name="de", split="2021_17", streaming=True)

# print the first item of the dataset
print(next(iter(ds)))

License

Internal testing only

Statistics

	num_bytes	num_examples
de	100752449183	40960253

de

	num_bytes	num_examples
2021_17	16614818320	7316629
2021_21	329787348	164456
2021_25	13490283920	5359845
2021_39	12298808823	4856931
2021_43	19052080067	6802704
2022_05	12019070936	4033390
2022_33	8787302552	3263896
2024_18	9749883212	4881888
2024_22	8410414005	4280514