About
About us
MITRA’s Mission is to harness the rapidly evolving AI technologies to promote the scholarly study and personal practice of the dharma and to accelerate academic and individual research through open-source collaboration on datasets, models and applications.
Automated Translation
Sanskrit, classical Tibetan, and classical Chinese to modern languages including Chinese, Tibetan, Korean, and Japanese
Semantic Search
Refining semantic search across a growing body of Sanskrit, Classical Tibetan, and Classical Chinese texts
Philologist's Toolkit
Develop tools to facilitate the study of Sanskrit and Classical Tibetan
In the future we hope to provide Generative AI tools that enable elevating and transformative experiences through:
- Immersive 3D experiences in maṇḍala
- Generation of 2D art based on canonical guildelines
The MITRA project
MITRA is a research project in the Berkeley AI Research lab in EECS at the University of California, Berkeley. It is lead by Kurt Keutzer and Sebastian Nehrdich and focuses on bridging the linguistic divide between ancient wisdom source languages and contemporary languages through the application of advanced Deep Learning and AI technologies.
Initiated in 2023, the project quickly evolved from its conceptual phase to a dynamic development process, accelerated by its collaborative efforts with organizations such as monlam.ai and with contributions from a diverse array of sources, including translators and AI researchers. Leveraging a robust corpus of over four million sentence pairs from various sourcesand utilizing Google's MADLAD-400 model as a foundation, MITRA has fine-tuned a specialized translation model that not only promises enhanced fluency in translations but also aims to significantly expand access to ancient wisdom texts.
Through continuous improvements in data quality, sentence alignment, and model fine-tuning, the project seeks to overcome the challenges inherent in low-resource language translation. The MITRA project stands as a testament to the transformative potential of AI in transcending language barriers, embodying a commitment to cultural preservation, academic research, and the democratization of access to Tibetan literature and wisdom.
Collaborations
monlam.ai
Machine translation data collection for Tibetan and English
Kumarajiva project
Development of Tibetan<>Chinese capabilities and user interface for specific translation purposes.
IIT Kharagpur
Development of Sanskrit translation model development. Sanskrit<>English dataset compilation
AI4Bharat
Development of Sanskrit data collection