Nemotron-pretraining-Indic-v1: 2.8 trillion tokens and open pipeline
NVIDIA and AI4Bharat announce Nemotron-pretraining-Indic-v1 with approximately 2.8 trillion tokens across 22 Indian languages and an open, GPU-accelerated data-generation pipeline for translation, quality assessment, and repair.
What changed
Released Nemotron-pretraining-Indic-v1 with approximately 2.8 trillion tokens and an open GPU-accelerated pipeline for translation, quality assessment, and repair; includes native-script for 22 languages and romanized text for twelve.
Why it matters
Developers can access a traceable synthetic corpus and a reusable GPU pipeline to generate, audit, and extend Indic training data for model development in Indian languages.
Who should care
DevelopersResearchersEnterprises
Worth trying
Maybe — Open pipeline available; corpus is described as "releasing soon".