inltk

Why we included this project

iNLTK bundles the NLP tasks most Indic-language apps need into one Python package: tokenization, word and sentence embeddings, similarity scoring, text generation, and classification. It ships pretrained models for fourteen languages, from Hindi and Tamil to Sanskrit and Telugu, plus code-mixed Hinglish, Tanglish, and Manglish, so a team building for Indian users does not have to assemble a separate model stack for each language. The code-mixed support is the practical part, since it matches how people actually type in India, mixing English into Hindi, Tamil, or Malayalam. Built on PyTorch with per-language language models, it suits both quick experiments and production pipelines that need consistent behavior across several Indic languages.