subword-tokenization
Akshara-aware tokenizer for six Brahmic scripts, with a trained SentencePiece model