{
  "entities_by_split": {
    "test": 184,
    "train": 1352,
    "validation": 137
  },
  "entity_labels": {
    "org.ent.pressagency.afp": 290,
    "org.ent.pressagency.ansa": 2,
    "org.ent.pressagency.ap": 88,
    "org.ent.pressagency.apa": 1,
    "org.ent.pressagency.ats-sda": 277,
    "org.ent.pressagency.belga": 3,
    "org.ent.pressagency.ctk": 1,
    "org.ent.pressagency.ddp-dapd": 3,
    "org.ent.pressagency.dnb": 34,
    "org.ent.pressagency.domei": 2,
    "org.ent.pressagency.dpa": 37,
    "org.ent.pressagency.europapress": 8,
    "org.ent.pressagency.extel": 23,
    "org.ent.pressagency.havas": 386,
    "org.ent.pressagency.kipa": 13,
    "org.ent.pressagency.reuters": 278,
    "org.ent.pressagency.spk-smp": 16,
    "org.ent.pressagency.stefani": 11,
    "org.ent.pressagency.tass": 8,
    "org.ent.pressagency.up-upi": 119,
    "org.ent.pressagency.wolff": 68,
    "org.ent.pressagency.xinhua": 5
  },
  "files": [
    "data/train.jsonl",
    "data/validation.jsonl",
    "data/test.jsonl",
    "label_map.json",
    "dataset_summary.json",
    "audit/curation_summary.json",
    "audit/curation_changes.jsonl",
    "audit/curation_changes_tags.tsv"
  ],
  "format": "jsonl",
  "label_count": 37,
  "languages_by_split": {
    "test": {
      "de": 32,
      "fr": 113
    },
    "train": {
      "de": 325,
      "fr": 875
    },
    "validation": {
      "de": 32,
      "fr": 107
    }
  },
  "legacy_trace_fields": [
    "source_format",
    "source_file",
    "news_agency_as_source"
  ],
  "public_entity_fields": [
    "entity_id",
    "label",
    "entity_family",
    "token_start",
    "token_stop",
    "start",
    "stop",
    "surface",
    "normalized_surface",
    "nel",
    "wikidata_url",
    "has_ocr_correction",
    "max_ocr_levenshtein"
  ],
  "public_row_fields": [
    "schema_version",
    "id",
    "document_id",
    "split",
    "language",
    "newspaper",
    "date",
    "year",
    "text",
    "tokens",
    "token_start_offsets",
    "token_end_offsets",
    "token_labels",
    "token_label_ids",
    "entities",
    "quality_flags",
    "legacy"
  ],
  "repo_id": "impresso-project/impresso-mediaagencies-ner-dataset",
  "splits": {
    "test": 145,
    "train": 1200,
    "validation": 139
  },
  "tokens_by_split": {
    "test": 90741,
    "train": 825329,
    "validation": 100723
  }
}