OpenArtha
Corpus distribution

Take the whole text.

Immutable, checksummed exports generated from one canonical corpus build.

gita-1.0.0-rc.1
Current release candidate

gita-1.0.0-rc.1

700 passages · 18 chapters
JSONLgita.jsonlOne complete passage record per line · 921 KBCSVgita.csvFlat table for spreadsheets and analysis · 452 KBXMLgita.xmlCanonical TEI P5 source representation · 687 KB
Manifest.jsonSHA-256 checksumsRights manifest
Commentary research preview

gita-commentaries-0.2.0

20,300 records · 29 public witnesses
JSONLcommentaries-en.jsonlVerse-aligned records, split by language · 7.7 MBXMLcommentaries-en.xmlSimplified TEI, split by language · 5.4 MBJSONLcommentaries-hi.jsonlVerse-aligned records, split by language · 14.6 MBXMLcommentaries-hi.xmlSimplified TEI, split by language · 13.5 MBJSONLcommentaries-sa.jsonlVerse-aligned records, split by language · 19.7 MBXMLcommentaries-sa.xmlSimplified TEI, split by language · 17.0 MBJSONrights.jsonDataset licensing and author-field coverage · 8 KBJSONalignment-report.jsonCanonical alignment counts and exceptions · 1 KBTXTLICENSE-GPL-3.0.txtGPL-3.0 license text · 34 KB
Manifest.jsonSHA-256 checksumsAuthor registry API
Release gate

Structured does not mean final.

Every verse requires two independent human review passes before gita-1.0.0.

The current files deliberately identify themselves as a research preview. Stable gita-1.0.0 will be cut only when all 700 passages satisfy the review gate.

See the validation process →
Read-only interface

Prefer an API?

GET /api/v2/works/gita/passages/2/47GET /api/v2/works/gita/search?q=yogaGET /api/v1/worksInspect the works response →