Transformers
tokenizer
sentencepiece
unigram
indic
multilingual
romanized
code-mixed
spearindic-tokenizer / tokenizer.json

Commit History

4 days ago

  1. SpearIndic v2: 22 languages, Unigram 200k/128k, full script coverage, ablation models

    Verified signature
    8570306

23 days ago

  1. SpearIndic tokenizer v1: Unigram-64k (default) + BPE 16k/32k/64k

    Verified signature
    5238ab7