# Buddhist AI datasets. One entry per dataset. Rendered into directory/README.md by scripts/build.py.
# Same fields as projects.yml; kind is one of the dataset kinds in scripts/check.py.
# access: open (open license, reuse allowed), open-nc (open license, non-commercial use only),
#         free (free to download or read, but no license stated or reuse restricted),
#         gated (you must ask or sign in to get it), closed (not available).
# collaborative: "yes" (the maintainers ask for corrections or new data; give the link in contribute),
#                "no", or not stated (we found no invitation). Say "yes" only when the source says so.
#                Always quote "yes" and "no", or YAML reads them as true/false.
# "updated" is the last change we saw on the dataset page (YYYY-MM).

# Benchmarks
- name: Thai Buddhist Studies Exam (Nak Tham)
  url: https://huggingface.co/datasets/biodatlab/thai_buddhist_studies_exam
  org: Mahidol University (biodatlab)
  kind: Benchmarks
  what: About 4,100 multiple-choice questions from the Thai national Dhamma exams (2020, 2022, 2023), three levels. Imported into BuddhistBench.
  languages: [Thai]
  open: Apache-2.0
  access: open
  collaborative: not stated
  updated: 2025-01
  tasks: [buddhist-knowledge-qa]
- name: BudHistBenchmark
  url: https://mbingenheimer.net/budhistbenchmark/
  org: Marcus Bingenheimer
  kind: Benchmarks
  what: 260 multiple-choice questions on Buddhist history (mostly Chinese and Japanese Buddhism). Top models now score about 95%.
  languages: [English]
  open: license not stated
  access: free
  collaborative: not stated
  updated: ""
  tasks: [buddhist-knowledge-qa]
- name: DharmaBench
  url: https://github.com/Intellexus-DSI/DharmaBench
  org: Intellexus (Reichman University, University of Hamburg)
  kind: Benchmarks
  what: 13 classification and span-detection tasks on Indian and Tibetan canonical texts, about 4,900 test items.
  languages: [Sanskrit, Tibetan]
  open: data CC-BY-4.0, code Apache-2.0
  access: open
  collaborative: not stated
  updated: 2026-01
  tasks: [segmentation, citation-mapping]
- name: MITRA-zh-eval
  url: https://aclanthology.org/2025.nlp4dh-1.12
  org: Dharmamitra team
  kind: Benchmarks
  what: 2,662 Classical Chinese to English sentence pairs from the Chinese canon for evaluating translation.
  languages: [Classical Chinese, English]
  open: paper CC BY 4.0; data on GitHub
  access: free
  collaborative: not stated
  updated: 2025
  tasks: [translation]
- name: BK-EVAL (佛典知識評測資料集)
  url: https://github.com/DILA-edu/BK-EVAL
  org: Dharma Drum Institute of Liberal Arts
  kind: Benchmarks
  what: Evaluation set on knowledge of the Chinese Buddhist canon.
  languages: [Classical Chinese, Chinese]
  open: on GitHub (license not stated)
  access: free
  collaborative: not stated
  updated: 2026-11
  tasks: [buddhist-knowledge-qa]
- name: Bo-bench
  url: https://huggingface.co/datasets/MonlamAI/Bo-bench-v1.0.0
  org: Monlam AI
  kind: Benchmarks
  what: Multiple-choice questions on Tibetan grammar, literature and history, written by subject experts, for evaluating Tibetan language models.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-04
  tasks: [buddhist-knowledge-qa]
- name: Daham Danuma
  url: https://github.com/sdkaluwitharana/daham-danuma
  org: sdkaluwitharana (community)
  kind: Benchmarks
  what: Open bank of multiple-choice questions for Dhamma school students in Sri Lanka.
  languages: [Sinhala]
  open: CC-BY-SA-4.0
  access: open
  collaborative: "yes"
  contribute: https://github.com/sdkaluwitharana/daham-danuma/blob/main/CONTRIBUTING.md
  updated: 2026-10
  tasks: [buddhist-knowledge-qa]
- name: TLUE
  url: https://arxiv.org/abs/2503.12051
  org: UESTC, Tibet University and others
  kind: Benchmarks
  what: 22,963 multiple-choice questions on general Tibetan knowledge, translated from Chinese benchmarks (no Buddhism category).
  languages: [Tibetan]
  open: CC-BY-NC-SA 4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2025
- name: canoncite
  url: https://huggingface.co/datasets/pralia-labs/canoncite
  org: pralia-labs
  kind: Benchmarks
  what: 622 citation and abstention items over ten public-domain corpora, including the Dhammapada.
  languages: [English]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: ""
  tasks: [citation-mapping]
- name: BDRC Tibetan OCR benchmark
  url: https://huggingface.co/datasets/BDRC/tibetan-ocr-benchmark
  org: BDRC
  kind: Benchmarks
  what: Evaluation set for Tibetan OCR across many kinds of prints and manuscripts. Has a public leaderboard.
  languages: [Tibetan]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-11
  tasks: [ocr]
- name: Tibetan OCR diagnostic benchmark
  url: https://huggingface.co/datasets/BDRC/tibetan-ocr-diagnostic-benchmark
  org: BDRC
  kind: Benchmarks
  what: 604 synthetic Tibetan page images that change one difficulty at a time (for example rotation or page shape) to show where OCR breaks.
  languages: [Tibetan]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-11
  tasks: [ocr]
- name: Tibetan metadata benchmark
  url: https://huggingface.co/datasets/BDRC/Bo_metadata_benchmark
  org: BDRC
  kind: Benchmarks
  what: Tests how well models extract title, author, translator, date and other catalog data from the start and end of a Tibetan text.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-08
  tasks: [entity-extraction]
- name: AI text outline benchmark
  url: https://huggingface.co/datasets/openpecha/ai-text-outline-benchmark
  org: OpenPecha
  kind: Benchmarks
  what: 124 Tibetan Buddhist texts with human-confirmed section breaks, for evaluating outline extraction.
  languages: [Tibetan]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-06
  tasks: [toc-extraction]
- name: Page-to-character outline benchmark
  url: https://huggingface.co/datasets/BDRC/PageToChar-Outline-benchmark
  org: BDRC
  kind: Benchmarks
  what: Finds where a text starts and ends inside OCR pages when page-level outlines are turned into character-level ones.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-09
  tasks: [toc-extraction]
- name: Tibetan speech-to-text benchmark
  url: https://huggingface.co/datasets/openpecha/tibetan-voice-benchmark
  org: Monlam AI and OpenPecha
  kind: Benchmarks
  what: Reviewed Tibetan speech transcripts from audio books, teachings and other sources, for evaluating speech recognition.
  languages: [Tibetan]
  open: MIT
  access: open
  collaborative: not stated
  updated: 2026-02
- name: BuddhistAI TOC extraction benchmark
  url: https://huggingface.co/datasets/BuddhistAI/toc-extraction-tibetan
  org: BuddhistAI
  kind: Benchmarks
  what: Table-of-contents extraction from Tibetan texts, with a scorer and leaderboard.
  languages: [Tibetan]
  open: MIT
  access: open
  collaborative: "yes"
  contribute: https://github.com/BuddhistAI/buddhist-benchmarks/blob/main/CONTRIBUTING.md
  updated: 2026-11
  tasks: [toc-extraction]
- name: BuddhistBench
  url: https://github.com/BuddhistAI/buddhist-bench
  org: BuddhistAI
  kind: Benchmarks
  what: Question and answer benchmark on Buddhist knowledge across traditions and languages, built from expert-written and exam sets.
  languages: [Thai, English, more]
  open: MIT (imported sets keep their own license)
  access: open
  collaborative: "yes"
  contribute: https://github.com/BuddhistAI/buddhist-bench/blob/main/CONTRIBUTING.md
  updated: 2026-11
  tasks: [buddhist-knowledge-qa]
- name: PaliBench
  url: https://github.com/MateMetzger/palibench
  org: Máté Metzger with Mahachulalongkornrajavidyalaya University
  kind: Benchmarks
  what: Pali to English translation benchmark of 1,700 passages with several reference translations, with scores for 10 LLMs. The repo has code, prompts and scores, but not the Pali passages or the reference translations.
  languages: [Pali, English]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-05
  tasks: [translation]
- name: WYWEB
  url: https://github.com/baudzhou/WYWEB
  org: baudzhou (ACL Findings 2023 authors)
  kind: Benchmarks
  what: Nine Classical Chinese tasks, including punctuation, named entities and translation into modern Chinese (about 26,000 pairs). Part of the punctuation data comes from Buddhist scriptures.
  languages: [Classical Chinese, Chinese]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2023
  tasks: [translation]

# Texts and corpora
- name: CBETA XML P5
  url: https://github.com/cbeta-org/xml-p5
  org: CBETA
  kind: Texts and corpora
  what: The full Chinese Buddhist canon of CBETA in TEI XML.
  languages: [Classical Chinese]
  open: most texts CC BY-NC-SA 4.0 (some collections not CC)
  access: open-nc
  collaborative: not stated
  updated: 2026-10
- name: CBETA Chinese canon (Hugging Face copy)
  url: https://huggingface.co/datasets/linhgiang/cbeta-chinese-canon
  org: linhgiang (community)
  kind: Texts and corpora
  what: The CBETA canon as a ready-to-use dataset.
  languages: [Classical Chinese]
  open: CC-BY-NC-SA-4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2026-09
- name: SuttaCentral data (sc-data)
  url: https://github.com/suttacentral/sc-data
  org: SuttaCentral
  kind: Texts and corpora
  what: Texts, translations and parallels behind SuttaCentral.
  languages: [Pali, Chinese, Sanskrit, Tibetan, English, many translations]
  open: mostly CC0 (see repo)
  access: open
  collaborative: "yes"
  contribute: https://github.com/suttacentral/sc-data/blob/main/sc_bilara_data/CONTRIBUTING.md
  updated: 2026-11
  tasks: [text-translation-alignment, citation-mapping]
- name: Early Buddhist Texts, Pali and English
  url: https://huggingface.co/datasets/linhgiang/early-buddhist-texts
  org: linhgiang (community)
  kind: Texts and corpora
  what: 5,764 suttas of the Pali Sutta Piṭaka aligned segment by segment with Bhikkhu Sujato's English translations.
  languages: [Pali, English]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-09
  tasks: [text-translation-alignment]
- name: BoCorpus
  url: https://huggingface.co/datasets/openpecha/BoCorpus
  org: OpenPecha
  kind: Texts and corpora
  what: Classical Tibetan texts from several digital collections, for training language models.
  languages: [Tibetan]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-06
- name: Tibetan corpus (Derge Kangyur and Tengyur)
  url: https://github.com/tokushige-koyasan/tibetan-corpus
  org: tokushige-koyasan (community)
  kind: Texts and corpora
  what: Plain text of 3,452 Tibetan Buddhist texts from the Derge Kangyur and Tengyur.
  languages: [Tibetan]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-08
- name: Derge Kangyur and Tengyur e-texts
  url: https://github.com/Esukhia/derge-kangyur
  org: Esukhia
  kind: Texts and corpora
  what: Digital Derge Kangyur (and a Tengyur repo next to it).
  languages: [Tibetan]
  open: on GitHub (license not stated)
  access: free
  collaborative: not stated
  updated: 2022-06
- name: Thai Tipitaka and commentaries (MBU edition)
  url: https://huggingface.co/datasets/uisp/tripitaka-mbu
  org: uisp (community)
  kind: Texts and corpora
  what: The Thai Tipitaka with commentaries, Mahamakut edition, 91 volumes, one file per page.
  languages: [Thai]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2024-12
- name: Pali Tipitaka in Thai script
  url: https://huggingface.co/datasets/uisp/pali-tripitaka-thai-script-siamrath-version
  org: uisp (community)
  kind: Texts and corpora
  what: The Pali Tipitaka (Siamrath edition) in Thai script, with a matching commentary set.
  languages: [Pali, Thai]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2024-12
- name: Tipitaka Pali in 15 scripts
  url: https://huggingface.co/datasets/freococo/tipitaka_pali_in_15_scripts
  org: freococo (community)
  kind: Texts and corpora
  what: The Pali Tipitaka written in 15 scripts.
  languages: [Pali]
  open: CC-BY-NC-4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2026-02
- name: Myanmar Tipitaka translation
  url: https://huggingface.co/datasets/freococo/tipitaka_myanmar_translation_books
  org: freococo (community)
  kind: Texts and corpora
  what: Burmese translation of the Tipitaka with the main commentaries and the Visuddhimagga, as JSONL.
  languages: [Burmese]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-02
- name: SiPaKosa
  url: https://github.com/RanxduG/SiPaKosa-Dataset
  org: RanxduG (community)
  kind: Texts and corpora
  what: Corpus of canonical and classical Buddhist texts in Sinhala and Pali.
  languages: [Sinhala, Pali]
  open: MIT
  access: open
  collaborative: not stated
  updated: 2026-05
- name: Daizhige
  url: https://huggingface.co/datasets/buddhist-nlp/daizhige
  org: Dharmamitra team
  kind: Texts and corpora
  what: Large Classical Chinese corpus (Daizhige collection).
  languages: [Classical Chinese]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2023-07
- name: CBETA-txt
  url: https://github.com/DILA-edu/CBETA-txt
  org: Dharma Drum Institute of Liberal Arts
  kind: Texts and corpora
  what: Plain text of Taishō volumes 1-55 and 85, made from the CBETA XML, with and without punctuation.
  languages: [Classical Chinese]
  open: license not stated (CBETA texts are CC BY-NC-SA 4.0)
  access: free
  collaborative: not stated
  updated: ""
- name: Tipitaka in Myanmar script (DatarrX)
  url: https://huggingface.co/datasets/DatarrX/tipitaka-dataset
  org: DatarrX (community)
  kind: Texts and corpora
  what: Pali Mula, Atthakatha and Tika texts in Myanmar script, about 463,000 paragraphs as JSON.
  languages: [Pali, Burmese]
  open: MIT
  access: open
  collaborative: not stated
  updated: ""
- name: INEL Kalmyk Corpus
  url: https://www.fdr.uni-hamburg.de/record/17676
  org: Universität Hamburg (INEL project)
  kind: Texts and corpora
  what: 55 spoken Kalmyk texts (19,742 tokens, about 4.4 hours of audio), glossed, with Russian and English translations.
  languages: [Kalmyk, Russian, English]
  open: CC-BY-NC-SA-4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2025-07

# Parallel texts and translation
- name: 84000 translation memory
  url: https://github.com/84000/data-translation-memory
  org: 84000
  kind: Parallel texts and translation
  what: Tibetan source segments aligned with 84000's English translations.
  languages: [Tibetan, English]
  open: on GitHub (license not stated)
  access: free
  collaborative: not stated
  updated: 2025-03
  tasks: [text-translation-alignment, translation]
- name: 84000 TEI translations
  url: https://github.com/84000/data-tei
  org: 84000
  kind: Parallel texts and translation
  what: 84000's published translations in TEI XML.
  languages: [English, Tibetan]
  open: on GitHub (license not stated)
  access: free
  collaborative: not stated
  updated: 2025-02
- name: MITRA parallel data
  url: https://github.com/dharmamitra/mitra-parallel
  org: Dharmamitra team
  kind: Parallel texts and translation
  what: Parallel sentence data from the Dharmamitra project.
  languages: [Sanskrit, Tibetan, Classical Chinese, English]
  open: on GitHub (license not stated)
  access: free
  collaborative: not stated
  updated: 2026-10
  tasks: [text-translation-alignment]
- name: Mitrasaṃgraha
  url: https://huggingface.co/datasets/buddhist-nlp/mitrasamgraha-released-data-only
  org: Dharmamitra team
  kind: Parallel texts and translation
  what: Released part of the Dharmamitra parallel data collection. The paper version on GitHub (dharmamitra/mitrasamgraha-dataset) has 391,548 Sanskrit-English pairs with a test set.
  languages: [Sanskrit, Tibetan, Classical Chinese, English]
  open: license not stated here; GitHub release CC BY-NC-ND 4.0 (a 372,791-pair subset CC BY-SA 4.0)
  access: free
  collaborative: not stated
  updated: 2026-02
  tasks: [translation]
- name: Tibetan-English sentence alignment
  url: https://huggingface.co/datasets/buddhist-nlp/sentence-alignment-tib-eng
  org: Dharmamitra team
  kind: Parallel texts and translation
  what: Aligned Tibetan and English sentences.
  languages: [Tibetan, English]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2024-09
  tasks: [text-translation-alignment]
- name: Tibetan to English, Spanish, French and German pairs
  url: https://huggingface.co/datasets/billingsmoore/tibetan-to-english-translation-dataset
  org: billingsmoore (community)
  kind: Parallel texts and translation
  what: Tibetan sentences paired with translations, with sister sets for Spanish, French and German and a phonetics set.
  languages: [Tibetan, English, Spanish, French, German]
  open: CC-BY-NC-4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2024-10
  tasks: [translation]
- name: Pali Tripitaka parallel corpus
  url: https://github.com/x39826/Pali_Tripitaka
  org: x39826 (community)
  kind: Parallel texts and translation
  what: Pali scriptures as used in 15 countries, with a parallel corpus.
  languages: [Pali, many]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2019-04
- name: BDK English Tripiṭaka
  url: https://bdk-seiten.com/translation.php?lang=en
  org: Bukkyō Dendō Kyōkai (BDK)
  kind: Parallel texts and translation
  what: English translations of texts from the Taishō canon (139 works in the first stage), as free PDFs.
  languages: [Classical Chinese, English]
  open: free PDFs; all rights reserved
  access: free
  collaborative: not stated
  updated: ""
  tasks: [translation]
- name: Sāmayik
  url: https://github.com/ayushbits/saamayik
  org: IIT Bombay, University of Utah and others
  kind: Parallel texts and translation
  what: About 53,000 English-Sanskrit sentence pairs of modern prose, with training and evaluation scripts. Not Buddhist-specific.
  languages: [Sanskrit, English]
  open: license not stated
  access: free
  collaborative: not stated
  updated: ""
  tasks: [translation]
- name: KanbunKundoku
  url: https://github.com/shiryusann/KanbunKundoku
  org: Zilong Li and Jie Cao
  kind: Parallel texts and translation
  what: Classical Chinese sentences with kundoku reading marks and Japanese readings, for studying translation of Classical Chinese into Japanese.
  languages: [Classical Chinese, Japanese]
  open: MIT
  access: open
  collaborative: not stated
  updated: 2025-11
  tasks: [translation]
- name: Buryat-Russian parallel corpus
  url: https://huggingface.co/datasets/SaranaAbidueva/buryat-russian_parallel_corpus
  org: Sarana Abidueva
  kind: Parallel texts and translation
  what: Buryat-Russian sentence pairs (10,000 to 100,000), used to train an NLLB-200 Buryat model.
  languages: [Buryat, Russian]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: ""
  tasks: [translation]
- name: Tuvan-Russian parallel 50k
  url: https://huggingface.co/datasets/Agisight/tyvan-russian-parallel-50k
  org: Agisight (tyvan.ru)
  kind: Parallel texts and translation
  what: 50,000 Russian-Tuvan sentence pairs taken from the tyvan.ru corpus.
  languages: [Tuvan, Russian]
  open: CC-BY-SA-4.0
  access: open
  collaborative: not stated
  updated: ""
  tasks: [translation]
- name: Asian Language Treebank (ALT)
  url: https://huggingface.co/datasets/alt
  org: NICT and partners
  kind: Parallel texts and translation
  what: About 20,000 English Wikinews sentences translated into 12 Asian languages, including Thai, Lao, Khmer and Burmese. Not Buddhist-specific.
  languages: [Thai, Lao, Khmer, Burmese, English, more]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: ""
  tasks: [translation]

# OCR and page layout
- name: BDRC OCR transcriptions (several sets)
  url: https://huggingface.co/BDRC
  org: BDRC
  kind: OCR and page layout
  what: Page images with checked transcriptions from many sources (ACIP, Berkeley, Monlam AI handwritten, Palri Parkhang and more), as separate datasets.
  languages: [Tibetan]
  open: mostly CC0
  access: open
  collaborative: not stated
  updated: 2026-09
  tasks: [ocr]
- name: TiBLAD - Tibetan Book Layout Analysis Dataset
  url: https://huggingface.co/datasets/BDRC/TiBLAD
  org: BDRC
  kind: OCR and page layout
  what: Tibetan book pages with layout regions marked.
  languages: [Tibetan]
  open: see dataset card
  access: gated
  collaborative: not stated
  updated: 2026-08
  tasks: [ocr]
- name: OpenPecha OCR datasets
  url: https://huggingface.co/openpecha
  org: OpenPecha
  kind: OCR and page layout
  what: Line images and transcriptions from Kangyur and other prints and manuscripts (Lhasa, Lithang, Derge, Stok, Drutsa and more), plus line and layout annotation.
  languages: [Tibetan]
  open: mostly ODC-BY or CC0
  access: open
  collaborative: not stated
  updated: 2025-10
  tasks: [ocr]
- name: Sinhala and Pali OCR data
  url: https://huggingface.co/datasets/pathnirvana/sinhala-pali-ocr
  org: Path Nirvana
  kind: OCR and page layout
  what: Old Sinhala and Pali prints with transcriptions.
  languages: [Sinhala, Pali]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-11
  tasks: [ocr]
- name: Tripitaka Koreana in Han (TKH, MTH)
  url: https://github.com/HCIILAB/TKH_MTH_Datasets_Release
  org: HCIILAB
  kind: OCR and page layout
  what: Page images of the Tripitaka Koreana and other Chinese canons with character annotation.
  languages: [Classical Chinese]
  open: research use (see repo)
  access: free
  collaborative: not stated
  updated: 2020-09
  tasks: [ocr]

# Speech
- name: Tibetan voice datasets
  url: https://huggingface.co/datasets/openpecha/tibetan_voice
  org: OpenPecha
  kind: Speech
  what: Tibetan speech with transcripts, including teachings, audio books and more. Monlam AI also publishes Bo-voice.
  languages: [Tibetan]
  open: see dataset cards
  access: free
  collaborative: not stated
  updated: 2026-05
- name: Garchen Rinpoche teachings speech data
  url: https://huggingface.co/datasets/openpecha/garchen_rinpoche_data
  org: OpenPecha
  kind: Speech
  what: 5.6 hours of transcribed teachings by Garchen Rinpoche, for speech recognition of one teacher.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2025-10
- name: Nwāchā Munā (Newari speech)
  url: https://huggingface.co/collections/ilprl-docse/nwacha-muna
  org: Kathmandu University (ILPRL)
  kind: Speech
  what: 5.39 hours of transcribed Newari (Nepal Bhasa) speech in Devanagari, with speech recognition baseline models.
  languages: [Newari]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-04
- name: Khmer speech corpus
  url: https://huggingface.co/datasets/seanghay/km-speech-corpus
  org: seanghay (community)
  kind: Speech
  what: About 10.4 hours of Khmer speech in 14,943 clips.
  languages: [Khmer]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: ""

# Language annotation
- name: Tibetan quotation detection
  url: https://huggingface.co/datasets/BDRC/Bo-Quotation-Dataset
  org: BDRC
  kind: Language annotation
  what: 254 classical Tibetan books with 23,095 human-reviewed quotation spans, for finding quotations in commentaries.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-11
  tasks: [citation-mapping]
- name: Tibetan yigchung (small-print notes) detection
  url: https://huggingface.co/datasets/BDRC/Bo-Yigchung-Dataset
  org: BDRC
  kind: Language annotation
  what: 762 classical Tibetan books with marked spans of yigchung, the small-print notes written under or beside the main line.
  languages: [Tibetan]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-11
  tasks: [annotation-detection]
- name: Tibetan spelling correction
  url: https://huggingface.co/datasets/BDRC/tibetan-spelling-correction-dataset
  org: BDRC
  kind: Language annotation
  what: Sentence pairs of a first transcription and the reviewer's corrected version, from BDRC's double-annotation work.
  languages: [Tibetan]
  open: CC0
  access: open
  collaborative: not stated
  updated: 2026-08
  tasks: [spell-check]
- name: Tibetan word segmentation
  url: https://huggingface.co/datasets/billingsmoore/tibetan-word-segmentation-ds
  org: billingsmoore (community)
  kind: Language annotation
  what: Expert word boundaries for 100 sentences of modern Lhasa Tibetan news speech.
  languages: [Tibetan]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-07
  tasks: [segmentation]
- name: Tibetan NER
  url: https://huggingface.co/datasets/TibetanAI/TibetanAI_NERv1.0
  org: TibetanAI
  kind: Language annotation
  what: Named entity data for Tibetan.
  languages: [Tibetan]
  open: Apache-2.0
  access: open
  collaborative: not stated
  updated: 2023-08
  tasks: [entity-extraction]
- name: Tibetan normalisation test data
  url: https://huggingface.co/datasets/pagantibet/Tibetan-normalisation-testdata
  org: pagantibet
  kind: Language annotation
  what: Test data for normalising historical Tibetan spelling, with an abbreviation dictionary next to it.
  languages: [Tibetan]
  open: CC-BY-NC-SA-4.0 (non-commercial)
  access: open-nc
  collaborative: not stated
  updated: 2026-04
  tasks: [text-cleanup, spell-check]
- name: Sanskrit sandhi, lemma and grammar data
  url: https://huggingface.co/chronbmm
  org: chronbmm
  kind: Language annotation
  what: Sanskrit datasets for splitting sandhi, lemmas, grammar tags and OCR post-correction.
  languages: [Sanskrit]
  open: see dataset cards
  access: free
  collaborative: not stated
  updated: 2024-05
  tasks: [segmentation, text-cleanup]

# Metadata, outlines and knowledge
- name: Tibetan metadata extracted
  url: https://huggingface.co/datasets/openpecha/tibetan-metadata-extracted
  org: OpenPecha
  kind: Metadata, outlines and knowledge
  what: Catalog data extracted from Tibetan texts, with training sets for language models.
  languages: [Tibetan]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-06
  tasks: [entity-extraction]
- name: Tibetan outline boundary snippets
  url: https://huggingface.co/datasets/openpecha/tibetan-outline-boundary-snippets-unique
  org: OpenPecha
  kind: Metadata, outlines and knowledge
  what: Tibetan text snippets around outline boundaries, for finding where works start and end.
  languages: [Tibetan]
  open: Apache-2.0
  access: open
  collaborative: not stated
  updated: 2026-06
  tasks: [toc-extraction]
- name: Buddhist Philosophy Graph
  url: https://huggingface.co/datasets/joyboseroy/buddhist-philosophy-graph
  org: joyboseroy (community)
  kind: Metadata, outlines and knowledge
  what: Knowledge graph of nine Buddhist schools, from Theravada to Tibetan Vajrayana. Every link cites a named text with a quote.
  languages: [English, Pali, Sanskrit, Tibetan, Chinese, Japanese]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-06
- name: Dhamma.Gift definitions
  url: https://huggingface.co/datasets/dhammagift/dg-dataset
  org: Dhamma.Gift
  kind: Metadata, outlines and knowledge
  what: Key terms of the suttas with the definitions the suttas themselves give.
  languages: [Pali, English, Russian]
  open: CC-BY-4.0
  access: open
  collaborative: not stated
  updated: 2026-11
  tasks: [term-extraction]
- name: INNER ATLAS SuttaCentral quotations
  url: https://huggingface.co/datasets/Zaneoiji/inneratlas-suttacentral
  org: Zaneoiji (community)
  kind: Metadata, outlines and knowledge
  what: 703,068 quotation records from SuttaCentral texts in many languages, each with its own rights and attribution.
  languages: [Pali, English, many]
  open: CC0 (920 records CC-BY-SA-3.0)
  access: open
  collaborative: not stated
  updated: 2026-09
  tasks: [citation-mapping]

# Q&A and instruction data
- name: MITRA instruction data
  url: https://huggingface.co/datasets/buddhist-nlp/mitra-instructions
  org: Dharmamitra team
  kind: Q&A and instruction data
  what: Instruction data from the MITRA project.
  languages: [Tibetan, Sanskrit, Classical Chinese, English]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2024-10
- name: Buddhist LLM fine-tuning data
  url: https://github.com/guyiicn/buddhist-llm-finetune
  org: guyiicn (community)
  kind: Q&A and instruction data
  what: 4,386 question and answer pairs on 18 CBETA scriptures, with fine-tuning scripts.
  languages: [Chinese]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2026-02
  tasks: [buddhist-knowledge-qa]
- name: Thai Dhamma books
  url: https://huggingface.co/datasets/uisp/dhamma-scholar-book
  org: uisp (community)
  kind: Q&A and instruction data
  what: Thai Dhamma study books, with a second set for the general public.
  languages: [Thai]
  open: license not stated
  access: free
  collaborative: not stated
  updated: 2024-12
  tasks: [buddhist-knowledge-qa]
- name: Vietnamese Buddhism Q&A
  url: https://huggingface.co/datasets/vanloc1808/buddism-qa-dataset
  org: vanloc1808 (community)
  kind: Q&A and instruction data
  what: Questions and answers on Buddhism in English and Vietnamese, merged from two source sets (about 16,000 rows in the dataset viewer).
  languages: [Vietnamese, English]
  open: MIT
  access: open
  collaborative: not stated
  updated: ""
  tasks: [buddhist-knowledge-qa]
