Agastya AGI v01 — Bramha Medha Native INDIA
Publisher: Rabbit Industries Pvt Ltd Evidence release: R1 — 2026-09-30 Canonical public evidence: https://zippri.com/agastya-native-brain-evidence/ GitHub evidence mirror: https://github.com/RabbitIndustries/agastya-native-brain-evidence
Status
This is a public provenance and model-card repository. It does not redistribute proprietary model weights, raw corpus, full tokenizer vocabulary, or private architecture source.
Published evidence records:
- fresh random initialization at step 0
- teacher weight ancestry: ZERO
- third-party weight ancestry: false
- third-party tokenizer artifact import: false
- unique parameter count: 1,264,715,776
- tokenizer vocabulary size: 98,320
- canonical context: 4,096 tokens
- Clean-P0 R2 acceptance: PASS
Canonical lineage
Agastya corpus -> Agastya tokenizer -> Agastya architecture -> RANDOM INITIALIZATION -> Agastya pretraining -> Agastya post-training -> Agastya Native Brain
Independent verification
Independent third-party verification remains PENDING. The repository includes a public verification protocol that a qualified evaluator can follow.
Citation
See CITATION.cff.
Persistent identifier
Published Zenodo DOI: 10.5281/zenodo.23054425
This DOI identifies the published Zenodo R1 evidence record.
Zenodo record: https://zenodo.org/records/23054425