home/categories/machine-learning/orchestra-research-ai-research-skills-02-tokenization-huggingface-tokenizers-skill-md
machine-learningdata-ai
huggingface-tokenizers
Fast tokenizers optimized for research and production. Rust-based implementation tokenizes 1GB in <20 seconds. Supports BPE, WordPiece, and Unigram algorithms. Train custom vocabularies, track alignments, handle padding/truncation. Integrates seamlessly with transformers. Use when you need high-performance tokenization or custom tokenizer training.
maintainer
Orchestra-Research
更新於 11/20/2025
星標
6563
分支
515
quick start
Installation and usage
Fast tokenizers optimized for research and production. Rust-based implementation tokenizes 1GB in <20 seconds. Supports BPE, WordPiece, and Unigram algorithms. Train custom vocabularies, track alignments, handle padding/truncation. Integrates seamlessly with transformers. Use when you need high-performance tokenization or custom tokenizer training.
安裝
$ install --globalskills.sh
使用
安裝後，您可以透過在終端機執行以下指令來使用此技能：
skills use huggingface-tokenizers