Tibetan large language model DeepZang unveiled in China

LHASA -- DeepZang, a large language model designed for the Tibetan language, was unveiled Sunday in Lhasa, capital of Southwest China's Xizang autonomous region. This language model is the first of its kind in the country.

Developed by a local company, CHOKNOR, the platform supports intelligent interaction in Tibetan, Chinese and English, and it integrates multiple functions including artificial intelligence (AI) dialogues, real-time translation and speech-to-text transcription, according to the company founder Tenzin Norbu.

So far, the model has accumulated nearly 70 million entries of standardized parallel corpora and over 30,500 hours of voice data covering Tibetan's three major dialects.

DeepZang is China's first Tibetan large language model to complete national registration for generative AI algorithms and models, filling a technological gap in the field and helping promote the high-quality development of Tibetan-language AI.

The app has seen impressive downloads, with the number of users rising by several thousand within just one hour of its release, according to the company.

"The launch of DeepZang means ethnic minority languages now have their own voice in the age of AI," said Tenzin Norbu.