The NuNER dataset is a large-scale named entity recognition corpus in CoNLL format, containing nearly 1 million annotated examples. It's designed for training robust NER models across various domains and entity types.
Structure
Format: CoNLL
Split: Train only (971,842 examples)
Features:
id: Numeric identifier for each example
words: Sequence of tokens
ner_tags: Named entity tags for each token
labels: Label annotations… See the full description on the dataset page: https://huggingface.co/datasets/milistu/NuNER-conll.
