openpecha/tibetan-outline-boundary-snippets-unique

Tibetan Outline Boundary Snippets (Unique)

Deduplicated benchmark corpus of 31,591 context snippets (one snippet per single-file source). Use this split for honest evaluation.

Each row has a snippet field with a </b> marker at the boundary position.

Related repos

• Full training corpus: ganga4364/tibetan-outline-boundary-snippets-full
• CRF full model: ganga4364/tibetan-outline-boundary-crf-full
• CRF unbiased model: ganga4364/tibetan-outline-boundary-crf-unbiased

Load

from datasets import load_dataset

ds = load_dataset("ganga4364/tibetan-outline-boundary-snippets-unique", split="train")
snippets = ds["snippet"]

hf download ganga4364/tibetan-outline-boundary-snippets-unique breakpoints_context_snippets_unique.json --repo-type dataset

Format

["...closing text།</b> ༄༅། །opening text...", "..."]

Citation

If you use this data, please cite the OpenPecha outline-detection project and link to this dataset.