dataset_info.json

{
  "citation": "@inproceedings{Tiedemann2012ParallelData,\n  author = {Tiedemann, J},\n  title = {Parallel Data, Tools and Interfaces in OPUS},\n  booktitle = {LREC}\n  year = {2012}}",
  "description": "OPUS is a collection of translated texts from the web.\n\nCreate your own config to choose which data / language pair to load.\n\n```\nconfig = tfds.translate.opus.OpusConfig(\n    version=tfds.core.Version('0.1.0'),\n    language_pair=(\"de\", \"en\"),\n    subsets=[\"GNOME\", \"EMEA\"]\n)\nbuilder = tfds.builder(\"opus\", config=config)\n```\n\nmedical documents",
  "downloadSize": "35952852",
  "location": {
    "urls": [
      "http://opus.nlpl.eu/"
    ]
  },
  "name": "opus",
  "splits": [
    {
      "name": "train",
      "numBytes": "198021004",
      "shardLengths": [
        "554376",
        "554376"
      ]
    }
  ],
  "supervisedKeys": {
    "input": "de",
    "output": "en"
  },
  "version": "0.1.0"
}