lm1b

References:

plain_text

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:lm1b/plain_text')
  • Description:
A benchmark corpus to be used for measuring progress in statistical language modeling. This has almost one billion words in the training data.
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 306688
'train' 30301028
  • Features:
{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}