Not established from the available sources.
NVIDIA; IIT MADRAS (AI4BHARAT) · MODEL RELEASE TRACKER
Nemotron-pretraining-Indic-v1
A large open pretraining corpus announced by NVIDIA and IIT Madras / AI4Bharat: Nemotron-pretraining-Indic-v1 is described as approximately 2.8 trillion tokens of synthetic textual data for Indian languages, to be released alongside an open, GPU‑accelerated data generation pipeline.CURRENT SNAPSHOT2/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
Not established from the available sources.
ModalitySynthetic textual pretraining data (native-script and romanized text), including translated and curated English material such as mathematics, code, and question–answer content rendered into Indian languages.
Not established from the available sources.
AvailabilityAnnounced as an open-source release for the community, to be published alongside the reusable GPU-accelerated data generation pipeline.
VERIFIABLE FACTS
Every value stays attached to a source and date.
RELEASE · Release announcementDEVELOPER CLAIM
Introduced as Nemotron-pretraining-Indic-v1, a pretraining corpus of approximately 2.8 trillion tokens; announced as releasing soon alongside the open, GPU-accelerated data generation pipeline used to create it.
CAPABILITIES · Languages coveredDEVELOPER CLAIM
Contains native-script data for 22 Indian languages (the 22 languages listed in India's Constitution) and romanized text for twelve of those languages.
MODALITIES · ModalityDEVELOPER CLAIM
Synthetic textual pretraining data (native-script and romanized text), including translated and curated English material such as mathematics, code, and question–answer content rendered into Indian languages.
AVAILABILITY · AvailabilityDEVELOPER CLAIM
Announced as an open-source release for the community, to be published alongside the reusable GPU-accelerated data generation pipeline.
SAFETY · Quality and groundingDEVELOPER CLAIM
Every record traces back to a curated source and outputs are checked against their inputs so the synthetic text stays grounded in real information; the pipeline includes prompts, LLM workflows, and quality checks for translation, assessment, and repair.
WHAT CHANGED