Lightweight piece tokenization library
Project Links
Meta
Author: Explosion
Requires Python: >=3.9
Classifiers
Development Status
- 5 - Production/Stable
Environment
- Console
Intended Audience
- Developers
- Science/Research
License
- OSI Approved :: MIT License
Operating System
- MacOS :: MacOS X
Programming Language
- Cython
- Python :: 3
- Python :: 3.9
- Python :: 3.10
- Python :: 3.11
- Python :: 3.12
- Python :: 3.13
- Python :: 3.14
Topic
- Scientific/Engineering
🥢 Curated Tokenizers
This Python library provides word-/sentencepiece tokenizers. The following types of tokenizers are currenty supported:
| Tokenizer | Binding | Example model |
|---|---|---|
| BPE | sentencepiece | |
| Byte BPE | Native | RoBERTa/GPT-2 |
| Unigram | sentencepiece | XLM-RoBERTa |
| Wordpiece | Native | BERT |
⚠️ Warning: experimental package
This package is experimental and it is likely that the APIs will change in incompatible ways.
⏳ Install
Curated tokenizers is availble through PyPI:
pip install curated_tokenizers
🚀 Quickstart
The best way to get started with curated tokenizers is through the
curated-transformers
library. curated-transformers also provides functionality to load tokenization
models from Huggingface Hub.
2.0.1
Aug 24, 2026
2.0.0
Apr 15, 2024
2.0.0.dev0
Apr 12, 2024
0.9.2
Feb 07, 2024
0.9.1
Sep 14, 2023
0.9.0
Jul 11, 2023
0.9.0.dev0
Jul 06, 2023
0.0.10
Aug 24, 2026
0.0.9
Jan 18, 2024
0.0.8
Jul 06, 2023
0.0.7
May 23, 2023
0.0.6
May 23, 2023
Wheel compatibility matrix
Files in release
curated_tokenizers-2.0.1-cp310-cp310-macosx_10_9_x86_64.whl (721.5KiB)
curated_tokenizers-2.0.1-cp310-cp310-macosx_11_0_arm64.whl (700.3KiB)
curated_tokenizers-2.0.1-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.0MiB)
curated_tokenizers-2.0.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp310-cp310-musllinux_1_2_aarch64.whl (6.7MiB)
curated_tokenizers-2.0.1-cp310-cp310-musllinux_1_2_x86_64.whl (6.9MiB)
curated_tokenizers-2.0.1-cp310-cp310-win32.whl (712.2KiB)
curated_tokenizers-2.0.1-cp310-cp310-win_amd64.whl (759.3KiB)
curated_tokenizers-2.0.1-cp310-cp310-win_arm64.whl (729.2KiB)
curated_tokenizers-2.0.1-cp311-cp311-macosx_10_9_x86_64.whl (723.5KiB)
curated_tokenizers-2.0.1-cp311-cp311-macosx_11_0_arm64.whl (702.7KiB)
curated_tokenizers-2.0.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.2MiB)
curated_tokenizers-2.0.1-cp311-cp311-musllinux_1_2_aarch64.whl (6.8MiB)
curated_tokenizers-2.0.1-cp311-cp311-musllinux_1_2_x86_64.whl (6.9MiB)
curated_tokenizers-2.0.1-cp311-cp311-win32.whl (711.4KiB)
curated_tokenizers-2.0.1-cp311-cp311-win_amd64.whl (759.5KiB)
curated_tokenizers-2.0.1-cp311-cp311-win_arm64.whl (728.7KiB)
curated_tokenizers-2.0.1-cp312-cp312-macosx_10_13_x86_64.whl (725.2KiB)
curated_tokenizers-2.0.1-cp312-cp312-macosx_11_0_arm64.whl (704.1KiB)
curated_tokenizers-2.0.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.2MiB)
curated_tokenizers-2.0.1-cp312-cp312-musllinux_1_2_aarch64.whl (6.8MiB)
curated_tokenizers-2.0.1-cp312-cp312-musllinux_1_2_x86_64.whl (7.0MiB)
curated_tokenizers-2.0.1-cp312-cp312-win32.whl (712.1KiB)
curated_tokenizers-2.0.1-cp312-cp312-win_amd64.whl (760.4KiB)
curated_tokenizers-2.0.1-cp312-cp312-win_arm64.whl (727.7KiB)
curated_tokenizers-2.0.1-cp313-cp313-macosx_10_13_x86_64.whl (723.1KiB)
curated_tokenizers-2.0.1-cp313-cp313-macosx_11_0_arm64.whl (702.1KiB)
curated_tokenizers-2.0.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.2MiB)
curated_tokenizers-2.0.1-cp313-cp313-musllinux_1_2_aarch64.whl (6.8MiB)
curated_tokenizers-2.0.1-cp313-cp313-musllinux_1_2_x86_64.whl (7.0MiB)
curated_tokenizers-2.0.1-cp313-cp313-win32.whl (711.7KiB)
curated_tokenizers-2.0.1-cp313-cp313-win_amd64.whl (759.7KiB)
curated_tokenizers-2.0.1-cp313-cp313-win_arm64.whl (727.6KiB)
curated_tokenizers-2.0.1-cp314-cp314-macosx_10_15_x86_64.whl (723.0KiB)
curated_tokenizers-2.0.1-cp314-cp314-macosx_11_0_arm64.whl (702.1KiB)
curated_tokenizers-2.0.1-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.2MiB)
curated_tokenizers-2.0.1-cp314-cp314-musllinux_1_2_aarch64.whl (6.8MiB)
curated_tokenizers-2.0.1-cp314-cp314-musllinux_1_2_x86_64.whl (7.0MiB)
curated_tokenizers-2.0.1-cp314-cp314-win32.whl (738.1KiB)
curated_tokenizers-2.0.1-cp314-cp314-win_amd64.whl (787.4KiB)
curated_tokenizers-2.0.1-cp314-cp314-win_arm64.whl (755.7KiB)
curated_tokenizers-2.0.1-cp314-cp314t-macosx_10_15_x86_64.whl (730.0KiB)
curated_tokenizers-2.0.1-cp314-cp314t-macosx_11_0_arm64.whl (711.3KiB)
curated_tokenizers-2.0.1-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.2MiB)
curated_tokenizers-2.0.1-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.3MiB)
curated_tokenizers-2.0.1-cp314-cp314t-musllinux_1_2_aarch64.whl (6.9MiB)
curated_tokenizers-2.0.1-cp314-cp314t-musllinux_1_2_x86_64.whl (7.0MiB)
curated_tokenizers-2.0.1-cp314-cp314t-win32.whl (748.4KiB)
curated_tokenizers-2.0.1-cp314-cp314t-win_amd64.whl (796.9KiB)
curated_tokenizers-2.0.1-cp314-cp314t-win_arm64.whl (763.8KiB)
curated_tokenizers-2.0.1-cp39-cp39-macosx_10_9_x86_64.whl (723.0KiB)
curated_tokenizers-2.0.1-cp39-cp39-macosx_11_0_arm64.whl (702.0KiB)
curated_tokenizers-2.0.1-cp39-cp39-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl (6.0MiB)
curated_tokenizers-2.0.1-cp39-cp39-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.1MiB)
curated_tokenizers-2.0.1-cp39-cp39-musllinux_1_2_aarch64.whl (6.7MiB)
curated_tokenizers-2.0.1-cp39-cp39-musllinux_1_2_x86_64.whl (6.9MiB)
curated_tokenizers-2.0.1-cp39-cp39-win32.whl (713.1KiB)
curated_tokenizers-2.0.1-cp39-cp39-win_amd64.whl (760.2KiB)
curated_tokenizers-2.0.1-cp39-cp39-win_arm64.whl (730.0KiB)
curated_tokenizers-2.0.1.tar.gz (2.2MiB)
Extras:
None
Dependencies:
regex
(>=2022)