Anjishnu Mukherjee

Pretraining Data: English NOW corpus (english-corpora.org/now). Paper: arxiv.org/abs/2601.15236. Code: github.com/iamshnoo/metadata_localization

LLaMa 2 instruction tuned on cleaned Alpaca dataset using QLoRA. For languages other than English, Alpaca data was first translated using NLLB-1.3B