Flat, one-row-per-span dataset intended for span/token-classification
(BIO-tagging style) hallucination detection over agent tool-calling traces,
derived from the same judging pipeline as the reasoning-distillation set in
this collection.
ds_bert_spans_full.jsonl — 11,942 rows. Already self-contained — no join
needed. Each row is one hallucinated span: span (verbatim text), type
(taxonomy label), avg_iou / exact / n_judges… See the full description on the dataset page:
https://huggingface.co/datasets/ssurface/hallucination-bert-spans.
