Author

Raviraj Joshi

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Review Oct 2026

Scaling Synthetic Pre-training Data and Evaluation for Indic LLMs

Building capable language models for India requires scalable multilingual data and robust evaluation. We introduce Nemotron-Pretraining-Indic-v1, a synthetic pretraining corpus of approximately 2.83 trillion tokens covering all 22 scheduled Indian languages in native script and twelve in romanized form. It is the large...

Abhinav Khattar, Aishwarya Chhabra, A. Velingker et al. · 0 citations

Preprint Aug 2026

Beyond Initialization Loss: A Systematic Study of Token Embedding Initialization Strategies for LLM Vocabulary Extension

It is shown that initialization loss and initialization bits-per-byte (Init BPB) are unreliable predictors of downstream convergence, whereas lightweight CPT, as few as 50 steps, provides a cost-effective and reliable signal for selecting the best initialization strategy.

Raviraj Joshi, Utkarsh Vaidya, S. Chauhan et al. · 0 citations