laya-indic Collection 322M zero-shot typed-decision model for Indian languages + Hinglish, and its open training data (real Sangraha text + synthetic, Qwen3-labelled). • 3 items • Updated about 5 hours ago • 1