Datasets:

pbedrin
/

news-toy-long

Modalities:

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

Dataset Viewer

The dataset viewer is not available for this dataset.

Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Top-20 longest HTML pages (by token count) from NewsXLM dataset. Each page is from a unique website. HTMLs range from ~230K to 2.3M tokens.

Columns Explanation

html: Original HTML
html_en: html with text nodes translated into English

annotations: List of labeled page nodes in the following format:

[
  {
    "xpath": "string",        // May end with "/tail"
    "text": "string",
    "labels": ["string"]      // In this dataset, the list always contains exactly one label
  },
  // ...
]

Labeled Attributes

title, publication_date, text, author, tag. A very limited number of category labels may appear.

Downloads last month: 35

Total file size: