Awesome

Bling Fire

Introduction

Hi, we are a team at Microsoft called Bling (Beyond Language Understanding), we help Bing be smarter. Here we wanted to share with all of you our FInite State machine and REgular expression manipulation library (FIRE). We use Fire for many linguistic operations inside Bing such as Tokenization, Multi-word expression matching, Unknown word-guessing, Stemming / Lemmatization just to mention a few.

Bling Fire Tokenizer Overview

Bling Fire Tokenizer provides state of the art performance for Natural Language text tokenization. Bling Fire supports the following tokenization algorithms:

1. Pattern-based tokenization
2. WordPiece tokenization
3. SentencePiece Unigram LM
4. SentencePiece BPE
5. Induced/learned syllabification patterns (identifies possible hyphenation points within a token)

Bling Fire provides uniform interface for working with all four algorithms so there is no difference for the client whether to use tokenizer for XLNET, BERT or your own custom model.

Model files describe the algorithms they are built for and are loaded on demand from external file. There are also two default models for NLTK-style tokenization and sentence breaking, which does not need to be loaded. The default tokenization model follows logic of NLTK, except hyphenated words are split and a few "errors" are fixed.

Normalization can be added to each model, but is optional.

Diffrences between algorithms are summarized here.

Bling Fire Tokenizer high level API designed in a way that it requires minimal or no configuration, or initialization, or additional files and is friendly for use from languages like Python, Ruby, Rust, C#, JavaScript (via WASM), etc.

We have precompiled some popular models and listed with the source code reference below:

Oh yes, it is also the fastest! We did a comparison of Bling Fire with tokenizers from Hugging Face, Bling Fire runs 4-5 times faster than Hugging Face Tokenizers, see also Bing Blog Post. We did comparison of Bling Fire Unigram LM and BPE implementaion to the same one in SentencePiece library and our implementation is ~2x faster, see XLNET benchmark and BPE benchmark. Not to mention our default models are 10x faster than the same functionality from SpaCy, see benchmark wiki and this Bing Blog Post.

So if low latency inference is what you need then you have to try Bling Fire!

Python API Description

If you simply want to use it in Python, you can install the latest release using pip:

pip install -U blingfire

Examples

1. Python example, using default pattern-based tokenizer:

from blingfire import *

text = 'After reading this post, you will know: What "natural language" is and how it is different from other types of data. What makes working with natural language so challenging. [1]'

print(text_to_sentences(text))
print(text_to_words(text))

Expected output:

After reading this post, you will know: What "natural language" is and how it is different from other types of data.
What makes working with natural language so challenging. [1]
After reading this post , you will know : What " natural language " is and how it is different from other types of data . What makes working with natural language so challenging . [ 1 ]

2. Python example, load a custom model for a pattern-based tokenizer:

from blingfire import *

# load a custom model from file
h = load_model("./wbd_chuni.bin")

text = 'This is the Bling-Fire tokenizer. 2007年9月日历表_2007年9月农历阳历一览表-万年历'

# custom model output
print(text_to_words_with_model(h, text))

# default model output
print(text_to_words(text))

free_model(h)

Expected output:

This is the Bling - Fire tokenizer . 2007 年 9 月 日 历 表 _2007 年 9 月 农 历 阳 历 一 览 表 - 万 年 历
This is the Bling - Fire tokenizer . 2007年9月日历表_2007年9月农历阳历一览表 - 万年历

3. Python example, calling BERT BASE tokenizer

On one thread, it works 14x faster than orignal BERT tokenizer written in Python. Given this code is written in C++ it can be called from multiple threads without blocking on global interpreter lock thus achieving higher speed-ups for batch mode.

import os
import blingfire

s = "Эpple pie. How do I renew my virtual smart card?: /Microsoft IT/ 'virtual' smart card certificates for DirectAccess are valid for one year. In order to get to microsoft.com we need to type pi@1.2.1.2."

# one time load the model (we are using the one that comes with the package)
h = blingfire.load_model(os.path.join(os.path.dirname(blingfire.__file__), "bert_base_tok.bin"))
print("Model Handle: %s" % h)

# use the model from one or more threads
print(s)
ids = blingfire.text_to_ids(h, s, 128, 100)  # sequence length: 128, oov id: 100
print(ids)                                   # returns a numpy array of length 128 (padded or trimmed)

# free the model at the end
blingfire.free_model(h)
print("Model Freed")

Expected output:

Model Handle: 2854016629088
Эpple pie. How do I renew my virtual smart card?: /Microsoft IT/ 'virtual' smart card certificates for DirectAccess are valid for one year. In order to get to microsoft.com we need to type pi@1.2.1.2.
[ 1208  9397  2571 11345  1012  2129  2079  1045 20687  2026  7484  6047
  4003  1029  1024  1013  7513  2009  1013  1005  7484  1005  6047  4003
 17987  2005  3622  6305  9623  2015  2024  9398  2005  2028  2095  1012
  1999  2344  2000  2131  2000  7513  1012  4012  2057  2342  2000  2828
 14255  1030  1015  1012  1016  1012  1015  1012  1016  1012     0     0
     0     0     0     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0     0     0     0     0     0     0
     0     0     0     0     0     0     0     0]
Model Freed

4. Python example to call a XLM-R tokenizer and prepare a pytorch batch

import os
import torch
from torch.nn.utils.rnn import pad_sequence
from blingfire import load_model, text_to_ids, free_model

# Load the XLM-RoBERTa tokenizer model provided by BlingFire
model_path = os.path.join("./data", 'xlm_roberta_base.bin')
tokenizer_model = load_model(model_path)

if __name__ == "__main__":
    # Sample input text
    input_texts = [
        "+1 (678) 274-9543 US https://lookup.robokiller.com/p/678-274-9543 (678) 274-9543 - roboKiller lookup",
        "+1 (678) 274-9543 US https://lookup.robokiller.com/p/678-274-9543 (678) 274-9543 - super robo killer lookup"
    ]

    # Tokenize input texts and create tensors
    token_ids = [
        torch.cat([
            torch.tensor([0], dtype=torch.long),  # 0 - <s> token
            torch.tensor(text_to_ids(tokenizer_model, t, 256, 3, True).astype('int64')), # 3 - <unk>
            torch.tensor([2], dtype=torch.long)   # 2 - </s> token
        ])
        for t in input_texts
    ]

    # Use torch.nn.utils.rnn.pad_sequence to pad the sequences
    token_ids = pad_sequence(token_ids, batch_first=True, padding_value=1) # 1 - <pad>

    print("\nAfter Padding:")
    print(token_ids)

    # Free the model after use
    free_model(tokenizer_model)

5. Python example, doing tokenization and hyphenation of a text

Since hyphenation API's take one word at a time with the limit of 300 Unicode characters, we need to break the text into words first and then run hyphenation for each token.

import os
import blingfire

# load a provided with the package model
h = blingfire.load_model(os.path.join(os.path.dirname(blingfire.__file__), "syllab.bin"))

# get a text
text = "Like Curiosity, the Perseverance rover was built by engineers and scientists at NASA's Jet Propulsion Laboratory in Pasadena, California. Roughly 85% of Perseverance's mass is based on Curiosity \"heritage hardware,\" saving NASA time and money and reducing risk considerably, agency officials have said.  Как и Curiosity, марсоход Perseverance был построен инженерами и учеными из Лаборатории реактивного движения НАСА в Пасадене, Калифорния. По словам официальных лиц агентства, примерно 85% массы Perseverance основано на «традиционном оборудовании» Curiosity, что экономит время и деньги NASA и значительно снижает риски."

# break text into words with default model and hyphenate each word
output = " ".join([blingfire.word_hyphenation_with_model(h, w) for w in blingfire.text_to_words(text).split(' ')])
print(output)

# free the model after we are all done
blingfire.free_model(h)

The output should be something like this:

Li-ke Cu-rios-i-ty , the Per-se-ve-rance ro-ver was built by en-gi-neers and sci-en-tists at NASA 's Jet Pro-pul-sion La-bo-ra-to-ry in Pa-sa-dena , Cali-for-nia . Roughly 85 % of Per-se-ve-rance 's mass is ba-se-d on Cu-rios-i-ty " he-r-i-tage hard-ware , " sa-ving NASA time and money and re-du-c-ing risk con-si-de-r-ably , agen-cy of-fi-cials ha-ve said . Ка-к и Cu-rios-i-ty , мар-со-ход Per-se-ve-rance бы-л построен ин-же-не-рами и у-че-ны-ми из Ла-бора-то-рии ре-актив-ного дви-же-ния НАСА в Па-са-дене , Ка-ли-фор-ния . По сло-вам офи-ци-аль-ных ли-ц агент-ства , при-мерно 85 % мас-сы Per-se-ve-rance осно-вано на « тра-ди-ци-он-ном обо-ру-до-ва-нии » Cu-rios-i-ty , что эко-но-мит вре-мя и деньги NASA и зна-чи-те-льно сни-жа-ет риски .

Note you can specify any other Unicode character as a hyphen that API inserts into the output string.

6. C# example, calling XLM Roberta tokenizer and getting ids and offsets

Note, everything that is supported in Python is supported by C# API as well. C# also has ability to use parallel computations since all models and functions are stateless you can share the same model across the threads without locks. Let's load XLM Roberta model and tokenize a string, for each token let's get ID and offsets in the original text.

using System;
using BlingFire;

namespace BlingUtilsTest
{
    class Program
    {
        static void Main(string[] args)
        {
            // load XLM Roberta tokenization model
            var h = BlingFireUtils.LoadModel("./xlm_roberta_base.bin");


            // input string
            string input = "Autophobia, also called monophobia, isolophobia, or eremophobia, is the specific phobia of isolation. I saw a girl with a telescope. Я увидел девушку с телескопом.";
            // get its UTF8 representation
            byte[] inBytes = System.Text.Encoding.UTF8.GetBytes(input);


            // allocate space for ids and offsets
            int[] Ids =  new int[128];
            int[] Starts =  new int[128];
            int[] Ends =  new int[128];

            // tokenize with loaded XLM Roberta tokenization and output ids and start and end offsets
            outputCount = BlingFireUtils.TextToIdsWithOffsets(h, inBytes, inBytes.Length, Ids, Starts, Ends, Ids.Length, 0);
            Console.WriteLine(String.Format("return length: {0}", outputCount));
            if (outputCount >= 0)
            {
                Console.Write("tokens from offsets: [");
                for(int i = 0; i < outputCount; ++i)
                {
                    int startOffset = Starts[i];
                    int surfaceLen = Ends[i] - Starts[i] + 1;

                    string token = System.Text.Encoding.UTF8.GetString(new ArraySegment<byte>(inBytes, startOffset, surfaceLen));
                    Console.Write(String.Format("'{0}'/{1} ", token, Ids[i]));
                }
                Console.WriteLine("]");
            }

            // free loaded models
            BlingFireUtils.FreeModel(h);
        }
    }
}

This code will print the following output:

return length: 49
tokens from offsets: ['Auto'/4396 'pho'/22014 'bia'/9166 ','/4 ' also'/2843 ' called'/35839 ' mono'/22460 'pho'/22014 'bia'/9166 ','/4 ' is'/83 'olo'/7537 'pho'/22014 'bia'/9166 ','/4 ' or'/707 ' '/6 'eremo'/102835 'pho'/22014 'bia'/9166 ','/4 ' is'/83 ' the'/70 ' specific'/29458 ' pho'/53073 'bia'/9166 ' of'/111 ' '/6 'isolation'/219488 '.'/5 ' I'/87 ' saw'/24124 ' a'/10 ' girl'/23040 ' with'/678 ' a'/10 ' tele'/5501 'scope'/70820 '.'/5 ' Я'/1509 ' увидел'/79132 ' дев'/29513 'у'/105 'шку'/46009 ' с'/135 ' теле'/18293 'скоп'/41333 'ом'/419 '.'/5 ]

See this project for more C# examples: https://github.com/microsoft/BlingFire/tree/master/nuget/test .

7. JavaScript example, fetching and loading model file, using the model to compute ids

The goal of integration with JavaScript is ability to run the code in a browser with ML frameworks like TensorFlow.js and FastText web assembly.

Note: this work is still in progress, we are likely to make some changes/improvements there.

import { GetVersion, TextToWords, TextToSentences, LoadModel, FreeModel, TextToIds } from './blingfire_wrapper.js';

$(document).ready(function() {

  var text = "I saw a girl with a telescope. Я видел девушку с телескопом.";

  var modelHandle1 = null;

  $("#btn4").click(function () {
    if(modelHandle1 == null) {
      (async function () {
        modelHandle1 = await LoadModel("./bert_base_tok.bin");
        console.log("Model handle: " + modelHandle1);
      })();
    }
  });

  $("#btn5").click(function () {
    if(modelHandle1 != null) {
      FreeModel(modelHandle1);
      modelHandle1 = null;
      console.log("Model Freed!");
    }
  });

  $("#btn6").click(function () {
    if(modelHandle1 != null) {
      console.log(TextToIds(modelHandle1, text, 128));
    } else {
      console.log("Load the model first!");
    }
  });

});

Full example code can be found here. Details of the API are described in the wasm folder.

8. Example of making a difference with using Bling Fire default tokenizer in a classification task

This notebook demonstrates how Bling Fire tokenizer helps in Stack Overflow posts classification problem.

9. Example of reaching 99% accuracy for language detection

This document describes how to improve FastText language detection model with Bling Fire and achive 99% accuracy in language detection task for 365 languages.

How to create your own models

If you want to create your own tokenization or any other finite-state model, you need to compile the C++ tools first. Then use these tools to compile linugusitc resources from human readble format into binary finite-state machines.

1. Setup your environment, once. You need to do this step once, it compiles retail version of the tools and adds the build directory to the PATH.
2. Adding BERT-like tokenization model is describing how to add new tokenization model similar to BERT.
3. How to add a new Unigram LM model.
4. How to add a new BPE model.

Note: please read the documents above in the order before creating your own model. If you have any questions please start an Issue in Github.

Support for other programming languages

1. Rust wrapper
2. Ruby wrapper

Supported Platforms

Bling Fire is supported for Windows, Linux and Mac (Thanks to Andrew Kane!)

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.

When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Working Branch

To contribute directly to code base, you should create a personal fork and create feature branches there when you need them. This keeps the main repository clean and your personal workflow out of sight.

Pull Request

Before we can accept a pull request from you, you'll need to sign a Contributor License Agreement (CLA). It is an automated process and you only need to do it once.

However, you don't have to do this up-front. You can simply clone, fork, and submit your pull-request as usual. When your pull-request is created, it is classified by a CLA bot. If the change is trivial (i.e. you just fixed a typo) then the PR is labelled with cla-not-required. Otherwise, it's classified as cla-required. In that case, the system will also tell you how you can sign the CLA. Once you have signed a CLA, the current and all future pull-requests will be labelled as cla-signed.

To enable us to quickly review and accept your pull requests, always create one pull request per issue and link the issue in the pull request if possible. Never merge multiple requests in one unless they have the same root cause. Besides, keep code changes as small as possible and avoid pure formatting changes to code that has not been modified otherwise.

Feedback

• Ask a question on Stack Overflow.
• File a bug in GitHub Issues.

Reporting Security Issues

Security issues and bugs should be reported privately, via email, to the Microsoft Security Response Center (MSRC) at secure@microsoft.com. You should receive a response within 24 hours. If for some reason you do not, please follow up via email to ensure we received your original message. Further information, including the MSRC PGP key, can be found in the Security TechCenter.

License

Copyright (c) Microsoft Corporation. All rights reserved.

Licensed under the MIT License.