SentencePiece
SentencePiece is a fast, lightweight, and unsupervised text tokenizer and detokenizer designed for neural network-based text generation systems (such as Large Language Models) where the vocabulary size is fixed prior to training.
It implements subword units—including Byte-Pair-Encoding (BPE) [Sennrich et al.] and the unigram language model [Kudo.]—with the ability to train directly from raw sentences. By treating input text as a raw sequence of Unicode characters, SentencePiece enables a purely end-to-end, language-independent pipeline that completely eliminates the need for language-specific pre- or post-processing.
This is not an official Google product.
Quick Start (Python)
SentencePiece provides an easy-to-use Python module. Install it via pip:
pip install sentencepiece
Basic Example
Here is how to train a model, encode text into tokens/IDs, and decode them back to the original string:
import sentencepiece as spm
# 1. Train a model directly from a raw text file.
# (No pre-tokenization or language-specific preprocessing required!)
spm.SentencePieceTrainer.train(
input='data/botchan.txt',
model_prefix='m',
vocab_size=1000
)
# 2. Load the trained model.
sp = spm.SentencePieceProcessor(model_file='m.model')
# 3. Encode raw text into subword pieces (strings) or vocabulary IDs (integers).
text = "I saw a girl with a telescope."
pieces = sp.encode(text, out_type=str)
ids = sp.encode(text, out_type=int)
print(f"Pieces: {pieces}")
# Output: ['▁I', '▁saw', '▁a', '▁girl', '▁with', '▁a', '▁', 'te', 'le', 's', 'c', 'o', 'pe', '.']
print(f"IDs: {ids}")
# Output: [9, 459, 11, 939, 44, 11, 4, 142, 82, 8, 28, 21, 132, 6]
# 4. Decode IDs or pieces back into the original text.
# The reconstruction is completely lossless and reversible!
print(sp.decode(ids))
# Output: "I saw a girl with a telescope."
print(sp.decode(pieces))
# Output: "I saw a girl with a telescope."
Why SentencePiece?
1. Reversible & Lossless Tokenization (Whitespace as a Basic Symbol)
Traditional tokenizers drop whitespace information (e.g., treating Tokenize("World.") identically to Tokenize("World .")), making detokenization ambiguous and language-dependent.
SentencePiece treats the input text as a raw sequence of Unicode characters. It escapes whitespaces with a meta-symbol ▁ (U+2581) and includes it in the tokenization. This design ensures that detokenization is a simple, lossless string join operation, entirely independent of the language:
# Lossless detokenization
original_text = "".join(pieces).replace("▁", " ")
2. Purely Data-Driven & Language-Independent
SentencePiece trains tokenization and detokenization models directly from raw sentences. It does not require language-specific pre-tokenizers (such as Moses, MeCab, or KyTea). This makes it highly effective for languages without explicit word boundaries, such as Chinese, Japanese, and Thai.
3. Subword Regularization & BPE-Dropout
To improve the robustness and accuracy of translation and language models, SentencePiece supports on-the-fly subword sampling during training. By sampling different segmentations for the same input text (Subword Regularization for Unigram, BPE-Dropout for BPE), it virtually augments your training data and makes the model more resilient to spelling variations and noise.
# Sample different segmentations on-the-fly
for _ in range(3):
print(sp.encode('New York', out_type=str, enable_sampling=True, alpha=0.1, nbest_size=-1))
# May output:
# ['▁', 'N', 'e', 'w', '▁York']
# ['▁New', '▁York']
# ['▁New', '▁Y', 'o', 'r', 'k']
4. Fast, Lightweight, and Self-Contained
- Performance: Written in highly optimized C++. Segmentation speed is around 50,000 sentences per second, with a memory footprint of only ~6MB.
- Self-Contained: The generated
.model file contains the entire normalization rules, vocabulary mapping, and segmentation model. You are guaranteed to get the exact same tokenization results in any environment (C++, Python, Go, etc.) as long as you use the same model file.
Benchmark Setup
- Environment: 24-core CPU, Python 3.13.
- Dataset: Balanced raw multilingual text from FLORES-200 (parallel sentences in English, Chinese, Japanese, and Thai; 11.29 MB, 60,720 lines). CJK and Thai texts are raw and do not contain artificial space delimiters.
- Batch Request Size: The entire dataset (60,720 sentences) is fed as a single batch request (a single Python
list[str]) in one call.
- Metric: Encoding throughput in MB/s (higher is better).
1. Unigram Model: T5-base (32k vocab)
| Tokenizer |
1 Thread |
2 Threads |
4 Threads |
8 Threads |
16 Threads |
24 Threads |
| SentencePiece |
27.41 |
43.83 |
71.62 |
102.08 |
123.33 |
127.60 |
| Hugging Face Fast |
3.78 |
7.15 |
12.45 |
20.33 |
27.00 |
31.49 |
2. BPE Model: Gemma 3 (256k vocab)
| Tokenizer |
1 Thread |
2 Threads |
4 Threads |
8 Threads |
16 Threads |
24 Threads |
| SentencePiece |
7.44 |
12.82 |
23.03 |
36.66 |
48.65 |
52.43 |
| Hugging Face Fast |
3.66 |
6.37 |
10.45 |
15.54 |
21.05 |
20.48 |
While the core tokenization (C++ or Rust) runs in parallel, the final step of converting the native results (C++ vector of vectors or Rust vector of vectors) into Python objects (list[list[int]] or list[Encoding]) is sequential and must be done on Python's main thread (GIL-locked). At high thread counts, this single-threaded serialization step becomes the dominant bottleneck, capping the scaling performance.
For the detailed analysis and single-thread reference comparison, see Performance Benchmark Details.
To run these benchmarks yourself, see the reproduction instructions and scripts.
Documentation & Resources
For detailed guides, API references, and advanced usage, please refer to the following resources:
License
SentencePiece is licensed under the Apache 2.0 License.
SentencePiece 是一个快速、轻量的无监督文本分词与还原工具，专为 LLM 等神经网络文本生成系统设计，这类系统的词表大小在训练前就已固定。它以子词单元的形式实现了 BPE 和 unigram 语言模型，并且可以直接从原始句子中训练。