OpenVINO Tokenizers

OpenVINO Tokenizers adds text processing operations to OpenVINO.

Features

• Perform tokenization and detokenization without third-party dependencies
• Convert a HuggingFace tokenizer into OpenVINO model tokenizer and detokenizer
• Combine OpenVINO models into a single model
• Add greedy decoding pipeline to text generation model

Installation

(Recommended) Create and activate virtual env:

python3 -m venv venv
source venv/bin/activate
 # or
conda create --name openvino_tokenizers
conda activate openvino_tokenizers

Minimal Installation

Use minimal installation when you have a converted OpenVINO tokenizer:

pip install openvino-tokenizers
 # or
conda install -c conda-forge openvino openvino-tokenizers

Convert Tokenizers Installation

If you want to convert HuggingFace tokenizers into OpenVINO tokenizers:

pip install openvino-tokenizers[transformers]
 # or
conda install -c conda-forge openvino openvino-tokenizers && pip install transformers[sentencepiece] tiktoken

Install Pre-release Version

Use openvino-tokenizers[transformers] to install tokenizers conversion dependencies.

pip install --pre -U openvino openvino-tokenizers --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly

Build and Install from Source

Using OpenVINO PyPI package

openvino-tokenizers build depends on openvino package which will be automatically installed from PyPI during the build process. To install unreleased versions, you would need to install openvino package from the nightly distribution channel using --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly

git clone https://github.com/openvinotoolkit/openvino_tokenizers.git
cd openvino_tokenizers
pip install . --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly

This command is the equivalent of minimal installation. Install tokenizers conversion dependencies if needed:

pip install transformers[sentencepiece] tiktoken

⚠️ Latest commit of OpenVINO Tokenizers might rely on features that are not present in the release OpenVINO version. Use a nightly build of OpenVINO or build OpenVINO Tokenizers from a release branch if you have issues with the build process.

Using OpenVINO archive

Install OpenVINO archive distribution. Use --no-deps to avoid OpenVINO installation from PyPI into your current environment. --extra-index-url is needed to resolve build dependencies only.

source path/to/installed/openvino/setupvars.sh
git clone https://github.com/openvinotoolkit/openvino_tokenizers.git
cd openvino_tokenizers
pip install --no-deps . --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly

This command is the equivalent of minimal installation. Install tokenizers conversion dependencies if needed:

pip install transformers[sentencepiece] tiktoken

⚠️ Latest commit of OpenVINO Tokenizers might rely on features that are not present in the release OpenVINO version. Use a nightly build of OpenVINO or build OpenVINO Tokenizers from a release branch if you have issues with the build process.

Build and install for development

Using OpenVINO PyPI package

git clone https://github.com/openvinotoolkit/openvino_tokenizers.git
cd openvino_tokenizers
pip install -e .[all] --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
# verify installation by running tests
cd tests/
pytest .

Using OpenVINO archive

Install OpenVINO archive distribution. Use --no-deps to avoid OpenVINO installation from PyPI into your current environment. --extra-index-url is needed to resolve build dependencies only.

source path/to/installed/openvino/setupvars.sh
git clone https://github.com/openvinotoolkit/openvino_tokenizers.git
cd openvino_tokenizers
pip install -e .[all] --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
# verify installation by running tests
cd tests/
pytest .

C++ Installation

You can use converted tokenizers in C++ pipelines with prebuild binaries.

1. Download OpenVINO archive distribution for your OS from here and extract the archive.
2. Download OpenVINO Tokenizers prebuild libraries from here. To ensure compatibility first three numbers of OpenVINO Tokenizers version should match OpenVINO version and OS.
3. Extract OpenVINO Tokenizers archive into OpenVINO installation directory. OpenVINO Tokenizers archive maintains the structure to be aligned with OpenVINO archive:
   • Windows: <openvino_dir>\runtime\bin\intel64\Release\
   • MacOS_x86: <openvino_dir>/runtime/lib/intel64/Release
   • MacOS_arm64: <openvino_dir>/runtime/lib/arm64/Release/
   • Linux_x86: <openvino_dir>/runtime/lib/intel64/
   • Linux_arm64: <openvino_dir>/runtime/lib/aarch64/

After that you can add binary extension in the code with:

• core.add_extension("openvino_tokenizers.dll") for Windows
• core.add_extension("libopenvino_tokenizers.dylib") for MacOS
• core.add_extension("libopenvino_tokenizers.so") for Linux

and read/compile converted (de)tokenizers models. If you use version 2023.3.0.0, the binary extension file is called (lib)user_ov_extension.(dll/dylib/so).

C++ Build

To build OpenVINO Tokenizers binaries locally, use this command:

source path/to/installed/openvino/setupvars.sh
git clone https://github.com/openvinotoolkit/openvino_tokenizers.git
cd openvino_tokenizers
mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

After that, you can transfer all binaries from build/src to <openvino_dir> as described in the C++ installation instruction above.

Reducing the ICU Data Size

By default, all available ICU locales are supported, which significantly increases the package size. To reduce the size of the ICU libraries included in your final package, follow these steps:

1. Use the ICU Data Configuration File:

   • This file specifies which features and locales to include in a custom data bundle. You can find more information here.

2. Set the ICU Data Filter File as an Environment Variable:

   • On Unix-like systems (Linux, macOS): Set the ICU_DATA_FILTER_FILE environment variable to the path of your configuration file (filters.json):

     export ICU_DATA_FILTER_FILE="filters.json"

   • On Windows: Set the ICU_DATA_FILTER_FILE environment variable using the Command Prompt or PowerShell:

     Command Prompt:

     set ICU_DATA_FILTER_FILE=filters.json

     PowerShell:

     $env:ICU_DATA_FILTER_FILE="filters.json"

3. Create a Configuration File:

   • An example configuration file (filters.json) might look like this:

   {
     "localeFilter": {
       "filterType": "language",
       "includelist": [
         "en"
       ]
     }
   }

4. Configure OpenVINO Tokenizers:

   • When building OpenVINO tokenizers, set the following CMake option during the project configuration:

   -DBUILD_FAST_TOKENIZERS=ON

   • Example for a pip installation path:

   ICU_DATA_FILTER_FILE=</path/to/filters.json> pip install git+https://github.com/openvinotoolkit/openvino_tokenizers.git --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly --config-settings=override=cmake.options.BUILD_FAST_TOKENIZERS=ON

By following these instructions, you can effectively reduce the size of the ICU libraries in your final package.

Build OpenVINO Tokenizers without FastTokenizer Library

If a tokenizer doesn't use CaseFold, UnicodeNormalization or Wordpiece operations, you can drastically reduce package binary size by building OpenVINO Tokenizers without FastTokenizer dependency with this flag:

-DENABLE_FAST_TOKENIZERS=OFF

This option can also help with building for platform that is supported by FastTokenizer, for example Android x86_64.

Example for a pip installation path:

pip install git+https://github.com/openvinotoolkit/openvino_tokenizers.git --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly --config-settings=override=cmake.options.ENABLE_FAST_TOKENIZERS=OFF

Usage

⚠️ OpenVINO Tokenizers can be inferred on a CPU device only.

Convert HuggingFace tokenizer

OpenVINO Tokenizers ships with CLI tool that can convert tokenizers from Huggingface Hub or Huggingface tokenizers saved on disk:

convert_tokenizer codellama/CodeLlama-7b-hf --with-detokenizer -o output_dir

There is also convert_tokenizer function that can convert tokenizer python object.

import numpy as np
from transformers import AutoTokenizer
from openvino import compile_model, save_model
from openvino_tokenizers import convert_tokenizer

hf_tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
ov_tokenizer = convert_tokenizer(hf_tokenizer)

compiled_tokenzier = compile_model(ov_tokenizer)
text_input = ["Test string"]

hf_output = hf_tokenizer(text_input, return_tensors="np")
ov_output = compiled_tokenzier(text_input)

for output_name in hf_output:
    print(f"OpenVINO {output_name} = {ov_output[output_name]}")
    print(f"HuggingFace {output_name} = {hf_output[output_name]}")
# OpenVINO input_ids = [[ 101 3231 5164  102]]
# HuggingFace input_ids = [[ 101 3231 5164  102]]
# OpenVINO token_type_ids = [[0 0 0 0]]
# HuggingFace token_type_ids = [[0 0 0 0]]
# OpenVINO attention_mask = [[1 1 1 1]]
# HuggingFace attention_mask = [[1 1 1 1]]

# save tokenizer for later use
save_model(ov_tokenizer, "openvino_tokenizer.xml")

loaded_tokenizer = compile_model("openvino_tokenizer.xml")
loaded_ov_output = loaded_tokenizer(text_input)
for output_name in hf_output:
    assert np.all(loaded_ov_output[output_name] == ov_output[output_name])

Connect Tokenizer to a Model

To infer and convert the original model, install torch or torch-cpu to the virtual environment.

from transformers import AutoTokenizer, AutoModelForSequenceClassification
from openvino import compile_model, convert_model
from openvino_tokenizers import convert_tokenizer, connect_models

checkpoint = "mrm8488/bert-tiny-finetuned-sms-spam-detection"
hf_tokenizer = AutoTokenizer.from_pretrained(checkpoint)
hf_model = AutoModelForSequenceClassification.from_pretrained(checkpoint)

text_input = ["Free money!!!"]
hf_input = hf_tokenizer(text_input, return_tensors="pt")
hf_output = hf_model(**hf_input)

ov_tokenizer = convert_tokenizer(hf_tokenizer)
ov_model = convert_model(hf_model, example_input=hf_input.data)
combined_model = connect_models(ov_tokenizer, ov_model)
compiled_combined_model = compile_model(combined_model)

openvino_output = compiled_combined_model(text_input)

print(f"OpenVINO logits: {openvino_output['logits']}")
# OpenVINO logits: [[ 1.2007061 -1.4698029]]
print(f"HuggingFace logits {hf_output.logits}")
# HuggingFace logits tensor([[ 1.2007, -1.4698]], grad_fn=<AddmmBackward0>)

Use Extension With Converted (De)Tokenizer or Model With (De)Tokenizer

Import openvino_tokenizers will add all tokenizer-related operations to OpenVINO, after which you can work with saved tokenizers and detokenizers.

import numpy as np
import openvino_tokenizers
from openvino import Core

core = Core()

# detokenizer from codellama sentencepiece model
compiled_detokenizer = core.compile_model("detokenizer.xml")

token_ids = np.random.randint(100, 1000, size=(3, 5))
openvino_output = compiled_detokenizer(token_ids)

print(openvino_output["string_output"])
# ['sc�ouition�', 'intvenord hasient', 'g shouldwer M more']

Text generation pipeline

import numpy as np
from openvino import compile_model, convert_model
from openvino_tokenizers import add_greedy_decoding, convert_tokenizer
from transformers import AutoModelForCausalLM, AutoTokenizer


model_checkpoint = "JackFram/llama-68m"
hf_tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
hf_model = AutoModelForCausalLM.from_pretrained(model_checkpoint, use_cache=False)

# convert hf tokenizer
text_input = ["Quick brown fox jumped "]
ov_tokenizer, ov_detokenizer = convert_tokenizer(hf_tokenizer, with_detokenizer=True)
compiled_tokenizer = compile_model(ov_tokenizer)

# transform input text into tokens
ov_input = compiled_tokenizer(text_input)
hf_input = hf_tokenizer(text_input, return_tensors="pt")

# convert Pytorch model to OpenVINO IR and add greedy decoding pipeline to it
ov_model = convert_model(hf_model, example_input=hf_input.data)
ov_model_with_greedy_decoding = add_greedy_decoding(ov_model)
compiled_model = compile_model(ov_model_with_greedy_decoding)

# generate new tokens
new_tokens_size = 10
prompt_size = ov_input["input_ids"].shape[-1]
input_dict = {
    output.any_name: np.hstack([tensor, np.zeros(shape=(1, new_tokens_size), dtype=np.int_)])
    for output, tensor in ov_input.items()
}
for idx in range(prompt_size, prompt_size + new_tokens_size):
    output = compiled_model(input_dict)["token_ids"]
    input_dict["input_ids"][:, idx] = output[:, idx - 1]
    input_dict["attention_mask"][:, idx] = 1
ov_token_ids = input_dict["input_ids"]

hf_token_ids = hf_model.generate(
    **hf_input,
    min_new_tokens=new_tokens_size,
    max_new_tokens=new_tokens_size,
    temperature=0,  # greedy decoding
)

# decode model output
compiled_detokenizer = compile_model(ov_detokenizer)
ov_output = compiled_detokenizer(ov_token_ids)["string_output"]
hf_output = hf_tokenizer.batch_decode(hf_token_ids, skip_special_tokens=True)
print(f"OpenVINO output string: `{ov_output}`")
# OpenVINO output string: `['Quick brown fox was walking through the forest. He was looking for something']`
print(f"HuggingFace output string: `{hf_output}`")
# HuggingFace output string: `['Quick brown fox was walking through the forest. He was looking for something']`

TensorFlow Text Integration

OpenVINO Tokenizers include converters for certain TensorFlow Text operations. Currently, only the MUSE model is supported. Here is an example of model conversion and inference:

import numpy as np
import tensorflow_hub as hub
import tensorflow_text  # register tf text ops
from openvino import convert_model, compile_model
import openvino_tokenizers  # register ov tokenizer ops and translators


sentences = ["dog",  "I cuccioli sono carini.", "私は犬と一緒にビーチを散歩するのが好きです"]
tf_embed = hub.load(
    "https://www.kaggle.com/models/google/universal-sentence-encoder/frameworks/"
    "TensorFlow2/variations/multilingual/versions/2"
)
# convert model that uses Sentencepiece tokenizer op from TF Text
ov_model = convert_model(tf_embed)
ov_embed = compile_model(ov_model, "CPU")

ov_result = ov_embed(sentences)[ov_embed.output()]
tf_result = tf_embed(sentences)

assert np.all(np.isclose(ov_result, tf_result, atol=1e-4))

RWKV Tokenizer

from urllib.request import urlopen

from openvino import compile_model
from openvino_tokenizers import build_rwkv_tokenizer


rwkv_vocab_url = (
    "https://raw.githubusercontent.com/BlinkDL/ChatRWKV/main/tokenizer/rwkv_vocab_v20230424.txt"
)

with urlopen(rwkv_vocab_url) as vocab_file:
    vocab = map(bytes.decode, vocab_file)
    tokenizer, detokenizer = build_rwkv_tokenizer(vocab)

tokenizer, detokenizer = compile_model(tokenizer), compile_model(detokenizer)

print(tokenized := tokenizer(["Test string"])["input_ids"])  # [[24235 47429]]
print(detokenizer(tokenized)["string_output"])  # ['Test string']

C++ Usage example

This example shows how to run inference with C++ on a text-classification model from Hugging Face. It expects the path to a model directory as parameter, and prints the logits returned by the model inference.

Export an example model by running the following command after pip install optimum[openvino]:

optimum-cli export openvino microsoft/deberta-base-mnli deberta-base-mnli-ov

#include <openvino/openvino.hpp>
#include <iostream>
#include <filesystem>

int main(int argc, char* argv[]) {
   std::string dirname = argv[1];
   std::filesystem::path dir_path(dirname);
   std::filesystem::path model_xml = dir_path / "openvino_model.xml";
   std::filesystem::path tokenizer_xml = dir_path / "openvino_tokenizer.xml";

   ov::Core core;
   // use "openvino_tokenizers.dll" on Windows, "libopenvino_tokenizers.dylib" on macOS
   core.add_extension("libopenvino_tokenizers.so");

   ov::InferRequest tokenizer_request = core.compile_model(tokenizer_xml, "CPU").create_infer_request();

   std::string prompt="Hello world!";
   tokenizer_request.set_input_tensor(ov::Tensor{ov::element::string, {1}, &prompt});
   tokenizer_request.infer();
   ov::Tensor input_ids = tokenizer_request.get_tensor("input_ids");
   ov::Tensor attention_mask = tokenizer_request.get_tensor("attention_mask");

   ov::InferRequest infer_request = core.compile_model(model_xml, "CPU").create_infer_request();
   infer_request.set_tensor("input_ids", input_ids);
   infer_request.set_tensor("attention_mask", attention_mask);
   infer_request.infer();

   auto output = infer_request.get_tensor("logits");
   const float *output_buffer = output.data<const float>();

   size_t num_elements = output.get_size();

   for (size_t i = 0; i < num_elements; i++) {
       std::cout << output_buffer[i] << " ";
   }

   std::cout << std::endl;
   return 0;
}

Supported Tokenizer Types

Huggingface Tokenizer Type	Tokenizer Model Type	Tokenizer	Detokenizer
Fast	WordPiece	✅	❌
	BPE	✅	✅
	Unigram	❌	❌
Legacy	SentencePiece .model	✅	✅
Custom	tiktoken	✅	✅
RWKV	Trie	✅	✅

Test Results

This report is autogenerated and includes tokenizers and detokenizers tests. The Output Matched, % column shows the percent of test strings for which the results of OpenVINO and Huggingface Tokenizers are the same. To update the report run pytest --update_readme tokenizers_test.py in tests directory.

Output Match by Tokenizer Type

Tokenizer Type	Output Matched, %	Number of Tests
BPE	97.18	4544
SentencePiece	89.19	6633
Tiktoken	96.56	524
WordPiece	98.39	747

Output Match by Model

Tokenizer Type	Model	Output Matched, %	Number of Tests
BPE	EleutherAI/gpt-neox-20b	95.92	245
BPE	NousResearch/Meta-Llama-3-8B-Instruct	100.00	247
BPE	Salesforce/codegen-16B-multi	96.17	261
BPE	Xenova/gpt-4o	100.00	261
BPE	ai-forever/rugpt3large_based_on_gpt2	94.64	261
BPE	bigscience/bloom	97.55	245
BPE	databricks/dolly-v2-3b	95.92	245
BPE	deepseek-ai/deepseek-coder-6.7b-instruct	99.24	263
BPE	facebook/galactica-120b	95.92	245
BPE	facebook/opt-66b	96.73	245
BPE	gpt2	95.40	261
BPE	koalajun/Gemma-2-9b-it-Ko-Crypto-Translate	100.00	247
BPE	laion/CLIP-ViT-bigG-14-laion2B-39B-b160k	100.00	261
BPE	microsoft/deberta-base	96.73	245
BPE	roberta-base	95.40	261
BPE	stabilityai/stablecode-completion-alpha-3b-4k	95.92	245
BPE	stabilityai/stablelm-2-1_6b	100.00	245
BPE	tiiuae/falcon-7b	93.87	261
SentencePiece	NousResearch/Llama-2-13b-hf	97.55	245
SentencePiece	NousResearch/Llama-2-13b-hf_legacy_sp_backend	97.55	245
SentencePiece	NousResearch/Llama-2-13b-hf_sp_backend	94.29	245
SentencePiece	TinyLlama/TinyLlama-1.1B-Chat-v1.0	100.00	247
SentencePiece	TinyLlama/TinyLlama-1.1B-Chat-v1.0_legacy_sp_backend	98.38	247
SentencePiece	TinyLlama/TinyLlama-1.1B-Chat-v1.0_sp_backend	100.00	247
SentencePiece	baichuan-inc/Baichuan2-7B-Chat_legacy_sp_backend	100.00	245
SentencePiece	camembert-base_legacy_sp_backend	75.51	245
SentencePiece	camembert-base_sp_backend	52.24	245
SentencePiece	facebook/musicgen-small_legacy_sp_backend	78.37	245
SentencePiece	facebook/musicgen-small_sp_backend	83.67	245
SentencePiece	microsoft/Phi-3-mini-128k-instruct	100.00	247
SentencePiece	microsoft/Phi-3-mini-128k-instruct_legacy_sp_backend	97.57	247
SentencePiece	microsoft/Phi-3-mini-128k-instruct_sp_backend	99.19	247
SentencePiece	microsoft/deberta-v3-base_legacy_sp_backend	100.00	245
SentencePiece	microsoft/deberta-v3-base_sp_backend	96.73	245
SentencePiece	mlx-community/quantized-gemma-7b-it	97.57	247
SentencePiece	mlx-community/quantized-gemma-7b-it_legacy_sp_backend	97.57	247
SentencePiece	mlx-community/quantized-gemma-7b-it_sp_backend	96.76	247
SentencePiece	rinna/bilingual-gpt-neox-4b_legacy_sp_backend	86.12	245
SentencePiece	rinna/bilingual-gpt-neox-4b_sp_backend	80.41	245
SentencePiece	t5-base_legacy_sp_backend	80.00	245
SentencePiece	t5-base_sp_backend	85.31	245
SentencePiece	xlm-roberta-base_legacy_sp_backend	95.10	245
SentencePiece	xlm-roberta-base_sp_backend	95.10	245
SentencePiece	xlnet-base-cased_legacy_sp_backend	57.96	245
SentencePiece	xlnet-base-cased_sp_backend	64.49	245
Tiktoken	Qwen/Qwen-14B-Chat	100.00	261
Tiktoken	THUDM/glm-4-9b-chat	93.16	263
WordPiece	ProsusAI/finbert	100.00	109
WordPiece	bert-base-multilingual-cased	100.00	109
WordPiece	cointegrated/rubert-tiny2	100.00	109
WordPiece	distilbert-base-uncased-finetuned-sst-2-english	100.00	109
WordPiece	google/mobilebert-uncased	100.00	93
WordPiece	rasa/LaBSE	88.99	109
WordPiece	sentence-transformers/all-MiniLM-L6-v2	100.00	109

Recreating Tokenizers From Tests

In some tokenizers, you need to select certain settings so that their output is closer to the Huggingface tokenizers:

• THUDM/chatglm2-6b detokenizer always skips special tokens. Use skip_special_tokens=True during conversion
• THUDM/chatglm3-6b detokenizer don't skips special tokens. Use skip_special_tokens=False during conversion
• All tested tiktoken based detokenizers leave extra spaces. Use clean_up_tokenization_spaces=False during conversion