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.
- Score
- Unranked
- Evidence
- No signals yet
- Last commit
- Not known
- Listed
Install
armory install huggingface-tokenizers --cli claudewrites the skill into.claude/skills/huggingface-tokenizers/SKILL.mdListed as compatible
# fetches the source and writes it to:
.claude/skills/huggingface-tokenizers/SKILL.mdNeeds the armory CLI · not on npm yet, build it from cli/ in the repository
What it is
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.
When to use it
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.
How to install / invoke
# Copy the skill into your .claude/skills/ directory
curl -sL https://raw.githubusercontent.com/davila7/claude-code-templates/main/cli-tool/components/skills/ai-research/tokenization-huggingface-tokenizers/SKILL.md -o .claude/skills/tokenization-huggingface-tokenizers/SKILL.md
Notes
Extracted from davila7/claude-code-templates, tokenization-huggingface-tokenizers category.