TokenDagger
lastmod 2025-07-02created 2025-07-021 edits
A tokenizer faster than OpenAI's Tiktoken
About
- A high-performance tokenizer 100% compatible with OpenAI's TikToken, delivering more than 2x throughput on large-scale text processing and 4x faster code tokenization
- Maximizes token pattern matching speed with a fast PCRE2-based regular expression parsing engine
- A simplified BPE algorithm minimizes performance degradation when handling large numbers of special tokens
- In real benchmarks, code tokenization is more than 4x faster, and it can be used as a drop-in replacement in existing TikToken code
- Supports Python 3.8+, easy to install from PyPI with
pip install tokendagger, and depends on PCRE2