sentencepiece-rs

Library Listed

Rust runtime reimplementation of SentencePiece model loading, normalization, encoding, and decoding.

Checked · Sources

Downloads
30,659
Dependents
5
Stars
2
Forks
1
Contributors
2
Commits
8
Latest release
0.2.2 ·

About

From the project README ·

SentencePiece model loading, normalization, encoding, and decoding in pure Rust. Load a .model file, tokenize text, and decode tokens back into text.

Install

[dependencies]
sentencepiece-rs = "0.2"

Quick Start

use sentencepiece_rs::SentencePieceProcessor;

fn main() -> sentencepiece_rs::Result<()> {
    let sp = SentencePieceProcessor::open("tokenizer.model")?;

    let pieces = sp.encode("hello rust world")?;
    let ids = sp.encode_to_ids("hello rust world")?;
    let text = sp.decode_ids(&ids)?;

    println!("{pieces:?}");
    println!("{ids:?}");
    println!("{text}");

    Ok(())
}

Decode Pieces

use sentencepiece_rs::SentencePieceProcessor;

let sp = SentencePieceProcessor::open("tokenizer.model")?;
let text = sp.decode(&["▁hello", "▁world"])?;
assert_eq!(text, "hello world");
# Ok::<(), sentencepiece_rs::Error>(())

Extra Options

use sentencepiece_rs::SentencePieceProcessor;

let mut sp = SentencePieceProcessor::open("tokenizer.model")?;
sp.set_encode_extra_options("bos:eos")?;

let ids = sp.encode_to_ids("ship it")?;
# Ok::<(), sentencepiece_rs::Error>(())

Supported options: bos, eos, reverse, unk_piece.

Read the full README

Compatibility Notes

The crate reads standard SentencePiece model protobufs and uses the embedded normalizer trie, so normal .model files should load without drama. The runtime behavior follows the C++ implementation, but this is not a line-by-line rewrite.

The big missing chunk is training. Bring your own existing model for now.

Documentation

docs.rs/sentencepiece-rs

References

Test Locally

cargo fmt --all
cargo test --all-features
cargo clippy --all-targets --all-features -- -D warnings
cargo doc --all-features --no-deps

License and Terms of Use

This crate is licensed under the Apache License, Version 2.0. See ./LICENSE.

The original SentencePiece source is also licensed under the Apache License, Version 2.0. This crate is a Rust reimplementation that uses the upstream source as the behavioral reference; it is not a C++ binding and does not link against the C++ library.

Apache 2.0 is permissive. In normal-human terms, you can use, modify, distribute, sublicense, and ship derivative work, including commercially, as long as you follow the license terms.

If you redistribute this crate, the upstream source, or a modified version, keep the important bits intact:

  • include a copy of the Apache 2.0 license
  • preserve copyright, patent, trademark, and attribution notices that apply
  • mark modified files when you change Apache-licensed source
  • include upstream NOTICE content if a distributed upstream package includes one
  • do not imply Google or SentencePiece trademark endorsement

The software is provided as-is, without warranty, and the Apache 2.0 patent grant terminates if you sue over patent infringement involving the licensed work. That is the deal. Pretty reasonable, honestly.

This README is a practical summary, not legal advice. The actual license text in ./LICENSE and the upstream SentencePiece license are the source of truth.

Runs on
GPU support unknown
What shows it
  • no dependency or feature matches a backend row
Native code
None found
  • No native code found
Python bindings
No Python binding crate found
License
Apache-2.0

Compare with similar projects

Order: strongest tier first, then most downloads, then name, with unmeasured values last. A library counts downloads over 30 days. An app counts binary downloads per release. An app with none counts its crate downloads over 30 days. ML Rust runs this catalogue, so its closest project always gets a row.

sentencepiece-rs beside the projects that share a topic with it
ProjectDownloadsDependentsStarsLatest releaseRuns on (GPU)Native codeLicense
sentencepiece-rs This page Listed

Rust runtime reimplementation of SentencePiece model loading, normalization, encoding, and decoding.

30,659 downloads over 30 days 5 2 0.2.2 GPU support unknown No native code found Apache-2.0
tiktoken-rs Featured

Library for encoding and decoding with the tiktoken library in Rust

3,939,182 downloads over 30 days 449 410 0.12.1 GPU support unknown No native code found MIT
whatlang-rs Featured

Fast and lightweight language identification library for Rust.

882,019 downloads over 30 days 74 1,087 0.18.0 GPU support unknown No native code found MIT
text-splitter Featured

Split text into semantic chunks, up to a desired chunk size. Supports calculating length by characters and tokens, and is callable from Rust and Python.

271,174 downloads over 30 days 56 645 0.33.1 GPU support unknown Binds native code: tree-sitter (feature code), tree-sitter-language (feature code) MIT

This crate aims to emulate https://github.com/google/sentencepiece Dart::DoubleArray struct and it's Normalizer. This crate is highly specialized and not intended for general use.

6,527,079 downloads over 30 days 7 23 0.1.4 GPU support unknown No native code found Apache-2.0
splintr Listed

Tokenizer for Rust and Python: pure Rust by default, with 18 bundled vocabularies and four backends covering BPE, SentencePiece and WordPiece. Its output matches Hugging Face tokenizers id for id on 14 model families, including GPT-2, Qwen, Whisper, T5 and BERT.

Python bindings

8,204 downloads over 30 days 7 63 0.22.2 GPU support unknown Binds native code: pcre2-sys (feature pcre2) MIT
Why ListedNo archive, New or Featured rule places it, so it is Listed.

What it lacks for Featured

  • It has 6 active dependents and 6 new dependents in 365 days. Featured needs 11 active dependents or 9 new dependents in 365 days. The repository route reads 2 stars, 1 fork and 2 contributors. It needs 4,539 stars, 434 forks and 79 contributors together.

What else the check found

  • The last activity is on 20 May 2026, 173 days ago. The rules expect an update within 365 days. Dormancy rule

Exact rules: Listed, Featured library. The check on applied thresholds version 9.

Every rule outcome, as the check records it
  • expected-interval: version 0.2.2 base 300, capped at 4 releases = 200, raised to the 365-day floor: E = 365 days
  • dormancy: last activity 2026-05-20, 173 days ago, E = 365: active
  • featured: not featured: active_users 6 < 11 and new_adopters 6 < 9 and repository route stars 2 (needs 4539), forks 1 (needs 434), contributors 2 (needs 79)
  • listed: listed: no archive, new or featured rule applies

Measurements and sources

Each signal row shows the raw value, its source and the fetch date. The Sources list gathers every distinct source on this page. Movement needs more history.

Sources6 sources
Adoption6 active dependents, 6 new dependents in 365 days1 not measured
Adoption measurements
SignalValueSourceFetchedCounts
Active dependents6crates.io database dumpYes
New dependents (365 days)6crates.io database dumpYes
Peak active dependents5.5crates.io database dumpYes
Active dependents as a share of the peak1crates.io database dumpYes
Binary downloads per releaseNot measuredYes
MaintenanceLatest human commit: 5 months ago3 not measured
Maintenance measurements
SignalValueSourceFetchedCounts
Latest human commit2026-05-20GitHub repositoryYes
Latest release2026-05-20crates.io database dumpYes
First release2026-05-10crates.io database dumpYes
Latest version0.2.2crates.io database dumpYes
Days the oldest outside issue waits for a reply0GitHub repositoryYes
Maintainer replies over 365 days1GitHub repositoryYes
Repository archivedfalseGitHub repositoryYes
Repository gonefalseGitHub repositoryYes
Release commit or manifest in the repotruecrates.io packageYes
Published docstruecrates.io database dumpYes
Latest release fails cargo checkNot measuredYes
Build check dateNot measuredYes
Build check toolchainNot measuredYes
SecurityUnpatched vulnerabilities: none
Security measurements
SignalValueSourceFetchedCounts
Unpatched vulnerabilities past the grace windownoneRustSec advisory databaseYes
RustSec unmaintained advisorynoneRustSec advisory databaseYes
Deprecation noticenonecrates.io database dumpYes
Deprecated requirementnonecrates.io database dumpYes
Distribution4 releases, 30,659 downloads over 30 days5 not measured
Distribution measurements
SignalValueSourceFetchedCounts
Releases4crates.io database dumpYes
Downloads over 30 days30659crates.io database dumpYes
Binary downloads over the last 3 releasesNot measuredYes
Homebrew installs on request over 365 daysNot measuredYes
Repology repository familiesNot measuredYes
Docker pull growth over 90 daysNot measuredYes
Docker pulls at the growth baselineNot measuredYes
Every version a prereleasefalsecrates.io database dumpYes
Every version 0.0.xfalsecrates.io database dumpYes
Every version yankedfalsecrates.io database dumpYes
Crate deleted from crates.iofalsecrates.io database dumpYes
Repository2 stars, 1 forks, 2 contributors, 8 commits. Stars alone never qualify a project.
Repository measurements
SignalValueSourceFetchedCounts
GitHub stars2GitHub repositoryYes — one part of the repository route
GitHub forks1GitHub repositoryYes — one part of the repository route
Contributors with a GitHub account2GitHub contributorsYes — one part of the repository route
Commits on the default branch8GitHub repositoryNo
Repository's main crate left out of the entrynonecrates.io database dumpYes
Facts that never count
Facts that never count measurements
SignalValueSourceFetchedCounts
LicenseApache-2.0crates.io database dumpNo
Runs on, from listed crates and their direct dependenciesGPU support unknowncrates.io database dumpNo
What shows each backendno dependency or feature matches a backend rowcrates.io database dumpNo
Native code, from listed crates and their dependenciesNo native code foundcrates.io database dumpNo
Python bindings, from direct dependencies of listed cratesNo Python binding crate foundcrates.io database dumpNo
Crates that depend on a listed crate directly5crates.io database dumpNo

Corrections and review history

Report a correction

No correction applies to this entry.

Judgments

A judgment enters only through an approved issue and a merged pull request.

No judgment applies to this entry.