Showing preview only (364K chars total). Download the full file or copy to clipboard to get everything.
Repository: google-research/bert
Branch: master
Commit: eedf5716ce12
Files: 21
Total size: 352.6 KB
Directory structure:
gitextract_6m8hx9o7/
├── .gitignore
├── CONTRIBUTING.md
├── LICENSE
├── README.md
├── __init__.py
├── create_pretraining_data.py
├── extract_features.py
├── modeling.py
├── modeling_test.py
├── multilingual.md
├── optimization.py
├── optimization_test.py
├── predicting_movie_reviews_with_bert_on_tf_hub.ipynb
├── requirements.txt
├── run_classifier.py
├── run_classifier_with_tfhub.py
├── run_pretraining.py
├── run_squad.py
├── sample_text.txt
├── tokenization.py
└── tokenization_test.py
================================================
FILE CONTENTS
================================================
================================================
FILE: .gitignore
================================================
# Initially taken from Github's Python gitignore file
# Byte-compiled / optimized / DLL files
__pycache__/
*.py[cod]
*$py.class
# C extensions
*.so
# Distribution / packaging
.Python
build/
develop-eggs/
dist/
downloads/
eggs/
.eggs/
lib/
lib64/
parts/
sdist/
var/
wheels/
*.egg-info/
.installed.cfg
*.egg
MANIFEST
# PyInstaller
# Usually these files are written by a python script from a template
# before PyInstaller builds the exe, so as to inject date/other infos into it.
*.manifest
*.spec
# Installer logs
pip-log.txt
pip-delete-this-directory.txt
# Unit test / coverage reports
htmlcov/
.tox/
.nox/
.coverage
.coverage.*
.cache
nosetests.xml
coverage.xml
*.cover
.hypothesis/
.pytest_cache/
# Translations
*.mo
*.pot
# Django stuff:
*.log
local_settings.py
db.sqlite3
# Flask stuff:
instance/
.webassets-cache
# Scrapy stuff:
.scrapy
# Sphinx documentation
docs/_build/
# PyBuilder
target/
# Jupyter Notebook
.ipynb_checkpoints
# IPython
profile_default/
ipython_config.py
# pyenv
.python-version
# celery beat schedule file
celerybeat-schedule
# SageMath parsed files
*.sage.py
# Environments
.env
.venv
env/
venv/
ENV/
env.bak/
venv.bak/
# Spyder project settings
.spyderproject
.spyproject
# Rope project settings
.ropeproject
# mkdocs documentation
/site
# mypy
.mypy_cache/
.dmypy.json
dmypy.json
# Pyre type checker
.pyre/
================================================
FILE: CONTRIBUTING.md
================================================
# How to Contribute
BERT needs to maintain permanent compatibility with the pre-trained model files,
so we do not plan to make any major changes to this library (other than what was
promised in the README). However, we can accept small patches related to
re-factoring and documentation. To submit contributes, there are just a few
small guidelines you need to follow.
## Contributor License Agreement
Contributions to this project must be accompanied by a Contributor License
Agreement. You (or your employer) retain the copyright to your contribution;
this simply gives us permission to use and redistribute your contributions as
part of the project. Head over to <https://cla.developers.google.com/> to see
your current agreements on file or to sign a new one.
You generally only need to submit a CLA once, so if you've already submitted one
(even if it was for a different project), you probably don't need to do it
again.
## Code reviews
All submissions, including submissions by project members, require review. We
use GitHub pull requests for this purpose. Consult
[GitHub Help](https://help.github.com/articles/about-pull-requests/) for more
information on using pull requests.
## Community Guidelines
This project follows
[Google's Open Source Community Guidelines](https://opensource.google.com/conduct/).
================================================
FILE: LICENSE
================================================
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
================================================
FILE: README.md
================================================
# BERT
**\*\*\*\*\* New March 11th, 2020: Smaller BERT Models \*\*\*\*\***
This is a release of 24 smaller BERT models (English only, uncased, trained with WordPiece masking) referenced in [Well-Read Students Learn Better: On the Importance of Pre-training Compact Models](https://arxiv.org/abs/1908.08962).
We have shown that the standard BERT recipe (including model architecture and training objective) is effective on a wide range of model sizes, beyond BERT-Base and BERT-Large. The smaller BERT models are intended for environments with restricted computational resources. They can be fine-tuned in the same manner as the original BERT models. However, they are most effective in the context of knowledge distillation, where the fine-tuning labels are produced by a larger and more accurate teacher.
Our goal is to enable research in institutions with fewer computational resources and encourage the community to seek directions of innovation alternative to increasing model capacity.
You can download all 24 from [here][all], or individually from the table below:
| |H=128|H=256|H=512|H=768|
|---|:---:|:---:|:---:|:---:|
| **L=2** |[**2/128 (BERT-Tiny)**][2_128]|[2/256][2_256]|[2/512][2_512]|[2/768][2_768]|
| **L=4** |[4/128][4_128]|[**4/256 (BERT-Mini)**][4_256]|[**4/512 (BERT-Small)**][4_512]|[4/768][4_768]|
| **L=6** |[6/128][6_128]|[6/256][6_256]|[6/512][6_512]|[6/768][6_768]|
| **L=8** |[8/128][8_128]|[8/256][8_256]|[**8/512 (BERT-Medium)**][8_512]|[8/768][8_768]|
| **L=10** |[10/128][10_128]|[10/256][10_256]|[10/512][10_512]|[10/768][10_768]|
| **L=12** |[12/128][12_128]|[12/256][12_256]|[12/512][12_512]|[**12/768 (BERT-Base)**][12_768]|
Note that the BERT-Base model in this release is included for completeness only; it was re-trained under the same regime as the original model.
Here are the corresponding GLUE scores on the test set:
|Model|Score|CoLA|SST-2|MRPC|STS-B|QQP|MNLI-m|MNLI-mm|QNLI(v2)|RTE|WNLI|AX|
|---|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
|BERT-Tiny|64.2|0.0|83.2|81.1/71.1|74.3/73.6|62.2/83.4|70.2|70.3|81.5|57.2|62.3|21.0|
|BERT-Mini|65.8|0.0|85.9|81.1/71.8|75.4/73.3|66.4/86.2|74.8|74.3|84.1|57.9|62.3|26.1|
|BERT-Small|71.2|27.8|89.7|83.4/76.2|78.8/77.0|68.1/87.0|77.6|77.0|86.4|61.8|62.3|28.6|
|BERT-Medium|73.5|38.0|89.6|86.6/81.6|80.4/78.4|69.6/87.9|80.0|79.1|87.7|62.2|62.3|30.5|
For each task, we selected the best fine-tuning hyperparameters from the lists below, and trained for 4 epochs:
- batch sizes: 8, 16, 32, 64, 128
- learning rates: 3e-4, 1e-4, 5e-5, 3e-5
If you use these models, please cite the following paper:
```
@article{turc2019,
title={Well-Read Students Learn Better: On the Importance of Pre-training Compact Models},
author={Turc, Iulia and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
journal={arXiv preprint arXiv:1908.08962v2 },
year={2019}
}
```
[2_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-128_A-2.zip
[2_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-256_A-4.zip
[2_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-512_A-8.zip
[2_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-768_A-12.zip
[4_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-128_A-2.zip
[4_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-256_A-4.zip
[4_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-512_A-8.zip
[4_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-768_A-12.zip
[6_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-128_A-2.zip
[6_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-256_A-4.zip
[6_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-512_A-8.zip
[6_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-768_A-12.zip
[8_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-128_A-2.zip
[8_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-256_A-4.zip
[8_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-512_A-8.zip
[8_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-768_A-12.zip
[10_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-128_A-2.zip
[10_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-256_A-4.zip
[10_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-512_A-8.zip
[10_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-768_A-12.zip
[12_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-128_A-2.zip
[12_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-256_A-4.zip
[12_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-512_A-8.zip
[12_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-768_A-12.zip
[all]: https://storage.googleapis.com/bert_models/2020_02_20/all_bert_models.zip
**\*\*\*\*\* New May 31st, 2019: Whole Word Masking Models \*\*\*\*\***
This is a release of several new models which were the result of an improvement
the pre-processing code.
In the original pre-processing code, we randomly select WordPiece tokens to
mask. For example:
`Input Text: the man jumped up , put his basket on phil ##am ##mon ' s head`
`Original Masked Input: [MASK] man [MASK] up , put his [MASK] on phil
[MASK] ##mon ' s head`
The new technique is called Whole Word Masking. In this case, we always mask
*all* of the the tokens corresponding to a word at once. The overall masking
rate remains the same.
`Whole Word Masked Input: the man [MASK] up , put his basket on [MASK] [MASK]
[MASK] ' s head`
The training is identical -- we still predict each masked WordPiece token
independently. The improvement comes from the fact that the original prediction
task was too 'easy' for words that had been split into multiple WordPieces.
This can be enabled during data generation by passing the flag
`--do_whole_word_mask=True` to `create_pretraining_data.py`.
Pre-trained models with Whole Word Masking are linked below. The data and
training were otherwise identical, and the models have identical structure and
vocab to the original models. We only include BERT-Large models. When using
these models, please make it clear in the paper that you are using the Whole
Word Masking variant of BERT-Large.
* **[`BERT-Large, Uncased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_uncased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
* **[`BERT-Large, Cased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_cased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
Model | SQUAD 1.1 F1/EM | Multi NLI Accuracy
---------------------------------------- | :-------------: | :----------------:
BERT-Large, Uncased (Original) | 91.0/84.3 | 86.05
BERT-Large, Uncased (Whole Word Masking) | 92.8/86.7 | 87.07
BERT-Large, Cased (Original) | 91.5/84.8 | 86.09
BERT-Large, Cased (Whole Word Masking) | 92.9/86.7 | 86.46
**\*\*\*\*\* New February 7th, 2019: TfHub Module \*\*\*\*\***
BERT has been uploaded to [TensorFlow Hub](https://tfhub.dev). See
`run_classifier_with_tfhub.py` for an example of how to use the TF Hub module,
or run an example in the browser on
[Colab](https://colab.sandbox.google.com/github/google-research/bert/blob/master/predicting_movie_reviews_with_bert_on_tf_hub.ipynb).
**\*\*\*\*\* New November 23rd, 2018: Un-normalized multilingual model + Thai +
Mongolian \*\*\*\*\***
We uploaded a new multilingual model which does *not* perform any normalization
on the input (no lower casing, accent stripping, or Unicode normalization), and
additionally inclues Thai and Mongolian.
**It is recommended to use this version for developing multilingual models,
especially on languages with non-Latin alphabets.**
This does not require any code changes, and can be downloaded here:
* **[`BERT-Base, Multilingual Cased`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
**\*\*\*\*\* New November 15th, 2018: SOTA SQuAD 2.0 System \*\*\*\*\***
We released code changes to reproduce our 83% F1 SQuAD 2.0 system, which is
currently 1st place on the leaderboard by 3%. See the SQuAD 2.0 section of the
README for details.
**\*\*\*\*\* New November 5th, 2018: Third-party PyTorch and Chainer versions of
BERT available \*\*\*\*\***
NLP researchers from HuggingFace made a
[PyTorch version of BERT available](https://github.com/huggingface/pytorch-pretrained-BERT)
which is compatible with our pre-trained checkpoints and is able to reproduce
our results. Sosuke Kobayashi also made a
[Chainer version of BERT available](https://github.com/soskek/bert-chainer)
(Thanks!) We were not involved in the creation or maintenance of the PyTorch
implementation so please direct any questions towards the authors of that
repository.
**\*\*\*\*\* New November 3rd, 2018: Multilingual and Chinese models available
\*\*\*\*\***
We have made two new BERT models available:
* **[`BERT-Base, Multilingual`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)
(Not recommended, use `Multilingual Cased` instead)**: 102 languages,
12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
parameters
We use character-based tokenization for Chinese, and WordPiece tokenization for
all other languages. Both models should work out-of-the-box without any code
changes. We did update the implementation of `BasicTokenizer` in
`tokenization.py` to support Chinese character tokenization, so please update if
you forked it. However, we did not change the tokenization API.
For more, see the
[Multilingual README](https://github.com/google-research/bert/blob/master/multilingual.md).
**\*\*\*\*\* End new information \*\*\*\*\***
## Introduction
**BERT**, or **B**idirectional **E**ncoder **R**epresentations from
**T**ransformers, is a new method of pre-training language representations which
obtains state-of-the-art results on a wide array of Natural Language Processing
(NLP) tasks.
Our academic paper which describes BERT in detail and provides full results on a
number of tasks can be found here:
[https://arxiv.org/abs/1810.04805](https://arxiv.org/abs/1810.04805).
To give a few numbers, here are the results on the
[SQuAD v1.1](https://rajpurkar.github.io/SQuAD-explorer/) question answering
task:
SQuAD v1.1 Leaderboard (Oct 8th 2018) | Test EM | Test F1
------------------------------------- | :------: | :------:
1st Place Ensemble - BERT | **87.4** | **93.2**
2nd Place Ensemble - nlnet | 86.0 | 91.7
1st Place Single Model - BERT | **85.1** | **91.8**
2nd Place Single Model - nlnet | 83.5 | 90.1
And several natural language inference tasks:
System | MultiNLI | Question NLI | SWAG
----------------------- | :------: | :----------: | :------:
BERT | **86.7** | **91.1** | **86.3**
OpenAI GPT (Prev. SOTA) | 82.2 | 88.1 | 75.0
Plus many other tasks.
Moreover, these results were all obtained with almost no task-specific neural
network architecture design.
If you already know what BERT is and you just want to get started, you can
[download the pre-trained models](#pre-trained-models) and
[run a state-of-the-art fine-tuning](#fine-tuning-with-bert) in only a few
minutes.
## What is BERT?
BERT is a method of pre-training language representations, meaning that we train
a general-purpose "language understanding" model on a large text corpus (like
Wikipedia), and then use that model for downstream NLP tasks that we care about
(like question answering). BERT outperforms previous methods because it is the
first *unsupervised*, *deeply bidirectional* system for pre-training NLP.
*Unsupervised* means that BERT was trained using only a plain text corpus, which
is important because an enormous amount of plain text data is publicly available
on the web in many languages.
Pre-trained representations can also either be *context-free* or *contextual*,
and contextual representations can further be *unidirectional* or
*bidirectional*. Context-free models such as
[word2vec](https://www.tensorflow.org/tutorials/representation/word2vec) or
[GloVe](https://nlp.stanford.edu/projects/glove/) generate a single "word
embedding" representation for each word in the vocabulary, so `bank` would have
the same representation in `bank deposit` and `river bank`. Contextual models
instead generate a representation of each word that is based on the other words
in the sentence.
BERT was built upon recent work in pre-training contextual representations —
including [Semi-supervised Sequence Learning](https://arxiv.org/abs/1511.01432),
[Generative Pre-Training](https://blog.openai.com/language-unsupervised/),
[ELMo](https://allennlp.org/elmo), and
[ULMFit](http://nlp.fast.ai/classification/2018/05/15/introducting-ulmfit.html)
— but crucially these models are all *unidirectional* or *shallowly
bidirectional*. This means that each word is only contextualized using the words
to its left (or right). For example, in the sentence `I made a bank deposit` the
unidirectional representation of `bank` is only based on `I made a` but not
`deposit`. Some previous work does combine the representations from separate
left-context and right-context models, but only in a "shallow" manner. BERT
represents "bank" using both its left and right context — `I made a ... deposit`
— starting from the very bottom of a deep neural network, so it is *deeply
bidirectional*.
BERT uses a simple approach for this: We mask out 15% of the words in the input,
run the entire sequence through a deep bidirectional
[Transformer](https://arxiv.org/abs/1706.03762) encoder, and then predict only
the masked words. For example:
```
Input: the man went to the [MASK1] . he bought a [MASK2] of milk.
Labels: [MASK1] = store; [MASK2] = gallon
```
In order to learn relationships between sentences, we also train on a simple
task which can be generated from any monolingual corpus: Given two sentences `A`
and `B`, is `B` the actual next sentence that comes after `A`, or just a random
sentence from the corpus?
```
Sentence A: the man went to the store .
Sentence B: he bought a gallon of milk .
Label: IsNextSentence
```
```
Sentence A: the man went to the store .
Sentence B: penguins are flightless .
Label: NotNextSentence
```
We then train a large model (12-layer to 24-layer Transformer) on a large corpus
(Wikipedia + [BookCorpus](http://yknzhu.wixsite.com/mbweb)) for a long time (1M
update steps), and that's BERT.
Using BERT has two stages: *Pre-training* and *fine-tuning*.
**Pre-training** is fairly expensive (four days on 4 to 16 Cloud TPUs), but is a
one-time procedure for each language (current models are English-only, but
multilingual models will be released in the near future). We are releasing a
number of pre-trained models from the paper which were pre-trained at Google.
Most NLP researchers will never need to pre-train their own model from scratch.
**Fine-tuning** is inexpensive. All of the results in the paper can be
replicated in at most 1 hour on a single Cloud TPU, or a few hours on a GPU,
starting from the exact same pre-trained model. SQuAD, for example, can be
trained in around 30 minutes on a single Cloud TPU to achieve a Dev F1 score of
91.0%, which is the single system state-of-the-art.
The other important aspect of BERT is that it can be adapted to many types of
NLP tasks very easily. In the paper, we demonstrate state-of-the-art results on
sentence-level (e.g., SST-2), sentence-pair-level (e.g., MultiNLI), word-level
(e.g., NER), and span-level (e.g., SQuAD) tasks with almost no task-specific
modifications.
## What has been released in this repository?
We are releasing the following:
* TensorFlow code for the BERT model architecture (which is mostly a standard
[Transformer](https://arxiv.org/abs/1706.03762) architecture).
* Pre-trained checkpoints for both the lowercase and cased version of
`BERT-Base` and `BERT-Large` from the paper.
* TensorFlow code for push-button replication of the most important
fine-tuning experiments from the paper, including SQuAD, MultiNLI, and MRPC.
All of the code in this repository works out-of-the-box with CPU, GPU, and Cloud
TPU.
## Pre-trained models
We are releasing the `BERT-Base` and `BERT-Large` models from the paper.
`Uncased` means that the text has been lowercased before WordPiece tokenization,
e.g., `John Smith` becomes `john smith`. The `Uncased` model also strips out any
accent markers. `Cased` means that the true case and accent markers are
preserved. Typically, the `Uncased` model is better unless you know that case
information is important for your task (e.g., Named Entity Recognition or
Part-of-Speech tagging).
These models are all released under the same license as the source code (Apache
2.0).
For information about the Multilingual and Chinese model, see the
[Multilingual README](https://github.com/google-research/bert/blob/master/multilingual.md).
**When using a cased model, make sure to pass `--do_lower=False` to the training
scripts. (Or pass `do_lower_case=False` directly to `FullTokenizer` if you're
using your own script.)**
The links to the models are here (right-click, 'Save link as...' on the name):
* **[`BERT-Large, Uncased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_uncased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
* **[`BERT-Large, Cased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_cased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
* **[`BERT-Base, Uncased`](https://storage.googleapis.com/bert_models/2018_10_18/uncased_L-12_H-768_A-12.zip)**:
12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Large, Uncased`](https://storage.googleapis.com/bert_models/2018_10_18/uncased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
* **[`BERT-Base, Cased`](https://storage.googleapis.com/bert_models/2018_10_18/cased_L-12_H-768_A-12.zip)**:
12-layer, 768-hidden, 12-heads , 110M parameters
* **[`BERT-Large, Cased`](https://storage.googleapis.com/bert_models/2018_10_18/cased_L-24_H-1024_A-16.zip)**:
24-layer, 1024-hidden, 16-heads, 340M parameters
* **[`BERT-Base, Multilingual Cased (New, recommended)`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Base, Multilingual Uncased (Orig, not recommended)`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)
(Not recommended, use `Multilingual Cased` instead)**: 102 languages,
12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
parameters
Each .zip file contains three items:
* A TensorFlow checkpoint (`bert_model.ckpt`) containing the pre-trained
weights (which is actually 3 files).
* A vocab file (`vocab.txt`) to map WordPiece to word id.
* A config file (`bert_config.json`) which specifies the hyperparameters of
the model.
## Fine-tuning with BERT
**Important**: All results on the paper were fine-tuned on a single Cloud TPU,
which has 64GB of RAM. It is currently not possible to re-produce most of the
`BERT-Large` results on the paper using a GPU with 12GB - 16GB of RAM, because
the maximum batch size that can fit in memory is too small. We are working on
adding code to this repository which allows for much larger effective batch size
on the GPU. See the section on [out-of-memory issues](#out-of-memory-issues) for
more details.
This code was tested with TensorFlow 1.11.0. It was tested with Python2 and
Python3 (but more thoroughly with Python2, since this is what's used internally
in Google).
The fine-tuning examples which use `BERT-Base` should be able to run on a GPU
that has at least 12GB of RAM using the hyperparameters given.
### Fine-tuning with Cloud TPUs
Most of the examples below assumes that you will be running training/evaluation
on your local machine, using a GPU like a Titan X or GTX 1080.
However, if you have access to a Cloud TPU that you want to train on, just add
the following flags to `run_classifier.py` or `run_squad.py`:
```
--use_tpu=True \
--tpu_name=$TPU_NAME
```
Please see the
[Google Cloud TPU tutorial](https://cloud.google.com/tpu/docs/tutorials/mnist)
for how to use Cloud TPUs. Alternatively, you can use the Google Colab notebook
"[BERT FineTuning with Cloud TPUs](https://colab.research.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb)".
On Cloud TPUs, the pretrained model and the output directory will need to be on
Google Cloud Storage. For example, if you have a bucket named `some_bucket`, you
might use the following flags instead:
```
--output_dir=gs://some_bucket/my_output_dir/
```
The unzipped pre-trained model files can also be found in the Google Cloud
Storage folder `gs://bert_models/2018_10_18`. For example:
```
export BERT_BASE_DIR=gs://bert_models/2018_10_18/uncased_L-12_H-768_A-12
```
### Sentence (and sentence-pair) classification tasks
Before running this example you must download the
[GLUE data](https://gluebenchmark.com/tasks) by running
[this script](https://gist.github.com/W4ngatang/60c2bdb54d156a41194446737ce03e2e)
and unpack it to some directory `$GLUE_DIR`. Next, download the `BERT-Base`
checkpoint and unzip it to some directory `$BERT_BASE_DIR`.
This example code fine-tunes `BERT-Base` on the Microsoft Research Paraphrase
Corpus (MRPC) corpus, which only contains 3,600 examples and can fine-tune in a
few minutes on most GPUs.
```shell
export BERT_BASE_DIR=/path/to/bert/uncased_L-12_H-768_A-12
export GLUE_DIR=/path/to/glue
python run_classifier.py \
--task_name=MRPC \
--do_train=true \
--do_eval=true \
--data_dir=$GLUE_DIR/MRPC \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
--max_seq_length=128 \
--train_batch_size=32 \
--learning_rate=2e-5 \
--num_train_epochs=3.0 \
--output_dir=/tmp/mrpc_output/
```
You should see output like this:
```
***** Eval results *****
eval_accuracy = 0.845588
eval_loss = 0.505248
global_step = 343
loss = 0.505248
```
This means that the Dev set accuracy was 84.55%. Small sets like MRPC have a
high variance in the Dev set accuracy, even when starting from the same
pre-training checkpoint. If you re-run multiple times (making sure to point to
different `output_dir`), you should see results between 84% and 88%.
A few other pre-trained models are implemented off-the-shelf in
`run_classifier.py`, so it should be straightforward to follow those examples to
use BERT for any single-sentence or sentence-pair classification task.
Note: You might see a message `Running train on CPU`. This really just means
that it's running on something other than a Cloud TPU, which includes a GPU.
#### Prediction from classifier
Once you have trained your classifier you can use it in inference mode by using
the --do_predict=true command. You need to have a file named test.tsv in the
input folder. Output will be created in file called test_results.tsv in the
output folder. Each line will contain output for each sample, columns are the
class probabilities.
```shell
export BERT_BASE_DIR=/path/to/bert/uncased_L-12_H-768_A-12
export GLUE_DIR=/path/to/glue
export TRAINED_CLASSIFIER=/path/to/fine/tuned/classifier
python run_classifier.py \
--task_name=MRPC \
--do_predict=true \
--data_dir=$GLUE_DIR/MRPC \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$TRAINED_CLASSIFIER \
--max_seq_length=128 \
--output_dir=/tmp/mrpc_output/
```
### SQuAD 1.1
The Stanford Question Answering Dataset (SQuAD) is a popular question answering
benchmark dataset. BERT (at the time of the release) obtains state-of-the-art
results on SQuAD with almost no task-specific network architecture modifications
or data augmentation. However, it does require semi-complex data pre-processing
and post-processing to deal with (a) the variable-length nature of SQuAD context
paragraphs, and (b) the character-level answer annotations which are used for
SQuAD training. This processing is implemented and documented in `run_squad.py`.
To run on SQuAD, you will first need to download the dataset. The
[SQuAD website](https://rajpurkar.github.io/SQuAD-explorer/) does not seem to
link to the v1.1 datasets any longer, but the necessary files can be found here:
* [train-v1.1.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json)
* [dev-v1.1.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v1.1.json)
* [evaluate-v1.1.py](https://github.com/allenai/bi-att-flow/blob/master/squad/evaluate-v1.1.py)
Download these to some directory `$SQUAD_DIR`.
The state-of-the-art SQuAD results from the paper currently cannot be reproduced
on a 12GB-16GB GPU due to memory constraints (in fact, even batch size 1 does
not seem to fit on a 12GB GPU using `BERT-Large`). However, a reasonably strong
`BERT-Base` model can be trained on the GPU with these hyperparameters:
```shell
python run_squad.py \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
--do_train=True \
--train_file=$SQUAD_DIR/train-v1.1.json \
--do_predict=True \
--predict_file=$SQUAD_DIR/dev-v1.1.json \
--train_batch_size=12 \
--learning_rate=3e-5 \
--num_train_epochs=2.0 \
--max_seq_length=384 \
--doc_stride=128 \
--output_dir=/tmp/squad_base/
```
The dev set predictions will be saved into a file called `predictions.json` in
the `output_dir`:
```shell
python $SQUAD_DIR/evaluate-v1.1.py $SQUAD_DIR/dev-v1.1.json ./squad/predictions.json
```
Which should produce an output like this:
```shell
{"f1": 88.41249612335034, "exact_match": 81.2488174077578}
```
You should see a result similar to the 88.5% reported in the paper for
`BERT-Base`.
If you have access to a Cloud TPU, you can train with `BERT-Large`. Here is a
set of hyperparameters (slightly different than the paper) which consistently
obtain around 90.5%-91.0% F1 single-system trained only on SQuAD:
```shell
python run_squad.py \
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
--do_train=True \
--train_file=$SQUAD_DIR/train-v1.1.json \
--do_predict=True \
--predict_file=$SQUAD_DIR/dev-v1.1.json \
--train_batch_size=24 \
--learning_rate=3e-5 \
--num_train_epochs=2.0 \
--max_seq_length=384 \
--doc_stride=128 \
--output_dir=gs://some_bucket/squad_large/ \
--use_tpu=True \
--tpu_name=$TPU_NAME
```
For example, one random run with these parameters produces the following Dev
scores:
```shell
{"f1": 90.87081895814865, "exact_match": 84.38978240302744}
```
If you fine-tune for one epoch on
[TriviaQA](http://nlp.cs.washington.edu/triviaqa/) before this the results will
be even better, but you will need to convert TriviaQA into the SQuAD json
format.
### SQuAD 2.0
This model is also implemented and documented in `run_squad.py`.
To run on SQuAD 2.0, you will first need to download the dataset. The necessary
files can be found here:
* [train-v2.0.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v2.0.json)
* [dev-v2.0.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v2.0.json)
* [evaluate-v2.0.py](https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/)
Download these to some directory `$SQUAD_DIR`.
On Cloud TPU you can run with BERT-Large as follows:
```shell
python run_squad.py \
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
--do_train=True \
--train_file=$SQUAD_DIR/train-v2.0.json \
--do_predict=True \
--predict_file=$SQUAD_DIR/dev-v2.0.json \
--train_batch_size=24 \
--learning_rate=3e-5 \
--num_train_epochs=2.0 \
--max_seq_length=384 \
--doc_stride=128 \
--output_dir=gs://some_bucket/squad_large/ \
--use_tpu=True \
--tpu_name=$TPU_NAME \
--version_2_with_negative=True
```
We assume you have copied everything from the output directory to a local
directory called ./squad/. The initial dev set predictions will be at
./squad/predictions.json and the differences between the score of no answer ("")
and the best non-null answer for each question will be in the file
./squad/null_odds.json
Run this script to tune a threshold for predicting null versus non-null answers:
python $SQUAD_DIR/evaluate-v2.0.py $SQUAD_DIR/dev-v2.0.json
./squad/predictions.json --na-prob-file ./squad/null_odds.json
Assume the script outputs "best_f1_thresh" THRESH. (Typical values are between
-1.0 and -5.0). You can now re-run the model to generate predictions with the
derived threshold or alternatively you can extract the appropriate answers from
./squad/nbest_predictions.json.
```shell
python run_squad.py \
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
--do_train=False \
--train_file=$SQUAD_DIR/train-v2.0.json \
--do_predict=True \
--predict_file=$SQUAD_DIR/dev-v2.0.json \
--train_batch_size=24 \
--learning_rate=3e-5 \
--num_train_epochs=2.0 \
--max_seq_length=384 \
--doc_stride=128 \
--output_dir=gs://some_bucket/squad_large/ \
--use_tpu=True \
--tpu_name=$TPU_NAME \
--version_2_with_negative=True \
--null_score_diff_threshold=$THRESH
```
### Out-of-memory issues
All experiments in the paper were fine-tuned on a Cloud TPU, which has 64GB of
device RAM. Therefore, when using a GPU with 12GB - 16GB of RAM, you are likely
to encounter out-of-memory issues if you use the same hyperparameters described
in the paper.
The factors that affect memory usage are:
* **`max_seq_length`**: The released models were trained with sequence lengths
up to 512, but you can fine-tune with a shorter max sequence length to save
substantial memory. This is controlled by the `max_seq_length` flag in our
example code.
* **`train_batch_size`**: The memory usage is also directly proportional to
the batch size.
* **Model type, `BERT-Base` vs. `BERT-Large`**: The `BERT-Large` model
requires significantly more memory than `BERT-Base`.
* **Optimizer**: The default optimizer for BERT is Adam, which requires a lot
of extra memory to store the `m` and `v` vectors. Switching to a more memory
efficient optimizer can reduce memory usage, but can also affect the
results. We have not experimented with other optimizers for fine-tuning.
Using the default training scripts (`run_classifier.py` and `run_squad.py`), we
benchmarked the maximum batch size on single Titan X GPU (12GB RAM) with
TensorFlow 1.11.0:
System | Seq Length | Max Batch Size
------------ | ---------- | --------------
`BERT-Base` | 64 | 64
... | 128 | 32
... | 256 | 16
... | 320 | 14
... | 384 | 12
... | 512 | 6
`BERT-Large` | 64 | 12
... | 128 | 6
... | 256 | 2
... | 320 | 1
... | 384 | 0
... | 512 | 0
Unfortunately, these max batch sizes for `BERT-Large` are so small that they
will actually harm the model accuracy, regardless of the learning rate used. We
are working on adding code to this repository which will allow much larger
effective batch sizes to be used on the GPU. The code will be based on one (or
both) of the following techniques:
* **Gradient accumulation**: The samples in a minibatch are typically
independent with respect to gradient computation (excluding batch
normalization, which is not used here). This means that the gradients of
multiple smaller minibatches can be accumulated before performing the weight
update, and this will be exactly equivalent to a single larger update.
* [**Gradient checkpointing**](https://github.com/openai/gradient-checkpointing):
The major use of GPU/TPU memory during DNN training is caching the
intermediate activations in the forward pass that are necessary for
efficient computation in the backward pass. "Gradient checkpointing" trades
memory for compute time by re-computing the activations in an intelligent
way.
**However, this is not implemented in the current release.**
## Using BERT to extract fixed feature vectors (like ELMo)
In certain cases, rather than fine-tuning the entire pre-trained model
end-to-end, it can be beneficial to obtained *pre-trained contextual
embeddings*, which are fixed contextual representations of each input token
generated from the hidden layers of the pre-trained model. This should also
mitigate most of the out-of-memory issues.
As an example, we include the script `extract_features.py` which can be used
like this:
```shell
# Sentence A and Sentence B are separated by the ||| delimiter for sentence
# pair tasks like question answering and entailment.
# For single sentence inputs, put one sentence per line and DON'T use the
# delimiter.
echo 'Who was Jim Henson ? ||| Jim Henson was a puppeteer' > /tmp/input.txt
python extract_features.py \
--input_file=/tmp/input.txt \
--output_file=/tmp/output.jsonl \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
--layers=-1,-2,-3,-4 \
--max_seq_length=128 \
--batch_size=8
```
This will create a JSON file (one line per line of input) containing the BERT
activations from each Transformer layer specified by `layers` (-1 is the final
hidden layer of the Transformer, etc.)
Note that this script will produce very large output files (by default, around
15kb for every input token).
If you need to maintain alignment between the original and tokenized words (for
projecting training labels), see the [Tokenization](#tokenization) section
below.
**Note:** You may see a message like `Could not find trained model in model_dir:
/tmp/tmpuB5g5c, running initialization to predict.` This message is expected, it
just means that we are using the `init_from_checkpoint()` API rather than the
saved model API. If you don't specify a checkpoint or specify an invalid
checkpoint, this script will complain.
## Tokenization
For sentence-level tasks (or sentence-pair) tasks, tokenization is very simple.
Just follow the example code in `run_classifier.py` and `extract_features.py`.
The basic procedure for sentence-level tasks is:
1. Instantiate an instance of `tokenizer = tokenization.FullTokenizer`
2. Tokenize the raw text with `tokens = tokenizer.tokenize(raw_text)`.
3. Truncate to the maximum sequence length. (You can use up to 512, but you
probably want to use shorter if possible for memory and speed reasons.)
4. Add the `[CLS]` and `[SEP]` tokens in the right place.
Word-level and span-level tasks (e.g., SQuAD and NER) are more complex, since
you need to maintain alignment between your input text and output text so that
you can project your training labels. SQuAD is a particularly complex example
because the input labels are *character*-based, and SQuAD paragraphs are often
longer than our maximum sequence length. See the code in `run_squad.py` to show
how we handle this.
Before we describe the general recipe for handling word-level tasks, it's
important to understand what exactly our tokenizer is doing. It has three main
steps:
1. **Text normalization**: Convert all whitespace characters to spaces, and
(for the `Uncased` model) lowercase the input and strip out accent markers.
E.g., `John Johanson's, → john johanson's,`.
2. **Punctuation splitting**: Split *all* punctuation characters on both sides
(i.e., add whitespace around all punctuation characters). Punctuation
characters are defined as (a) Anything with a `P*` Unicode class, (b) any
non-letter/number/space ASCII character (e.g., characters like `$` which are
technically not punctuation). E.g., `john johanson's, → john johanson ' s ,`
3. **WordPiece tokenization**: Apply whitespace tokenization to the output of
the above procedure, and apply
[WordPiece](https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/data_generators/text_encoder.py)
tokenization to each token separately. (Our implementation is directly based
on the one from `tensor2tensor`, which is linked). E.g., `john johanson ' s
, → john johan ##son ' s ,`
The advantage of this scheme is that it is "compatible" with most existing
English tokenizers. For example, imagine that you have a part-of-speech tagging
task which looks like this:
```
Input: John Johanson 's house
Labels: NNP NNP POS NN
```
The tokenized output will look like this:
```
Tokens: john johan ##son ' s house
```
Crucially, this would be the same output as if the raw text were `John
Johanson's house` (with no space before the `'s`).
If you have a pre-tokenized representation with word-level annotations, you can
simply tokenize each input word independently, and deterministically maintain an
original-to-tokenized alignment:
```python
### Input
orig_tokens = ["John", "Johanson", "'s", "house"]
labels = ["NNP", "NNP", "POS", "NN"]
### Output
bert_tokens = []
# Token map will be an int -> int mapping between the `orig_tokens` index and
# the `bert_tokens` index.
orig_to_tok_map = []
tokenizer = tokenization.FullTokenizer(
vocab_file=vocab_file, do_lower_case=True)
bert_tokens.append("[CLS]")
for orig_token in orig_tokens:
orig_to_tok_map.append(len(bert_tokens))
bert_tokens.extend(tokenizer.tokenize(orig_token))
bert_tokens.append("[SEP]")
# bert_tokens == ["[CLS]", "john", "johan", "##son", "'", "s", "house", "[SEP]"]
# orig_to_tok_map == [1, 2, 4, 6]
```
Now `orig_to_tok_map` can be used to project `labels` to the tokenized
representation.
There are common English tokenization schemes which will cause a slight mismatch
between how BERT was pre-trained. For example, if your input tokenization splits
off contractions like `do n't`, this will cause a mismatch. If it is possible to
do so, you should pre-process your data to convert these back to raw-looking
text, but if it's not possible, this mismatch is likely not a big deal.
## Pre-training with BERT
We are releasing code to do "masked LM" and "next sentence prediction" on an
arbitrary text corpus. Note that this is *not* the exact code that was used for
the paper (the original code was written in C++, and had some additional
complexity), but this code does generate pre-training data as described in the
paper.
Here's how to run the data generation. The input is a plain text file, with one
sentence per line. (It is important that these be actual sentences for the "next
sentence prediction" task). Documents are delimited by empty lines. The output
is a set of `tf.train.Example`s serialized into `TFRecord` file format.
You can perform sentence segmentation with an off-the-shelf NLP toolkit such as
[spaCy](https://spacy.io/). The `create_pretraining_data.py` script will
concatenate segments until they reach the maximum sequence length to minimize
computational waste from padding (see the script for more details). However, you
may want to intentionally add a slight amount of noise to your input data (e.g.,
randomly truncate 2% of input segments) to make it more robust to non-sentential
input during fine-tuning.
This script stores all of the examples for the entire input file in memory, so
for large data files you should shard the input file and call the script
multiple times. (You can pass in a file glob to `run_pretraining.py`, e.g.,
`tf_examples.tf_record*`.)
The `max_predictions_per_seq` is the maximum number of masked LM predictions per
sequence. You should set this to around `max_seq_length` * `masked_lm_prob` (the
script doesn't do that automatically because the exact value needs to be passed
to both scripts).
```shell
python create_pretraining_data.py \
--input_file=./sample_text.txt \
--output_file=/tmp/tf_examples.tfrecord \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--do_lower_case=True \
--max_seq_length=128 \
--max_predictions_per_seq=20 \
--masked_lm_prob=0.15 \
--random_seed=12345 \
--dupe_factor=5
```
Here's how to run the pre-training. Do not include `init_checkpoint` if you are
pre-training from scratch. The model configuration (including vocab size) is
specified in `bert_config_file`. This demo code only pre-trains for a small
number of steps (20), but in practice you will probably want to set
`num_train_steps` to 10000 steps or more. The `max_seq_length` and
`max_predictions_per_seq` parameters passed to `run_pretraining.py` must be the
same as `create_pretraining_data.py`.
```shell
python run_pretraining.py \
--input_file=/tmp/tf_examples.tfrecord \
--output_dir=/tmp/pretraining_output \
--do_train=True \
--do_eval=True \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
--train_batch_size=32 \
--max_seq_length=128 \
--max_predictions_per_seq=20 \
--num_train_steps=20 \
--num_warmup_steps=10 \
--learning_rate=2e-5
```
This will produce an output like this:
```
***** Eval results *****
global_step = 20
loss = 0.0979674
masked_lm_accuracy = 0.985479
masked_lm_loss = 0.0979328
next_sentence_accuracy = 1.0
next_sentence_loss = 3.45724e-05
```
Note that since our `sample_text.txt` file is very small, this example training
will overfit that data in only a few steps and produce unrealistically high
accuracy numbers.
### Pre-training tips and caveats
* **If using your own vocabulary, make sure to change `vocab_size` in
`bert_config.json`. If you use a larger vocabulary without changing this,
you will likely get NaNs when training on GPU or TPU due to unchecked
out-of-bounds access.**
* If your task has a large domain-specific corpus available (e.g., "movie
reviews" or "scientific papers"), it will likely be beneficial to run
additional steps of pre-training on your corpus, starting from the BERT
checkpoint.
* The learning rate we used in the paper was 1e-4. However, if you are doing
additional steps of pre-training starting from an existing BERT checkpoint,
you should use a smaller learning rate (e.g., 2e-5).
* Current BERT models are English-only, but we do plan to release a
multilingual model which has been pre-trained on a lot of languages in the
near future (hopefully by the end of November 2018).
* Longer sequences are disproportionately expensive because attention is
quadratic to the sequence length. In other words, a batch of 64 sequences of
length 512 is much more expensive than a batch of 256 sequences of
length 128. The fully-connected/convolutional cost is the same, but the
attention cost is far greater for the 512-length sequences. Therefore, one
good recipe is to pre-train for, say, 90,000 steps with a sequence length of
128 and then for 10,000 additional steps with a sequence length of 512. The
very long sequences are mostly needed to learn positional embeddings, which
can be learned fairly quickly. Note that this does require generating the
data twice with different values of `max_seq_length`.
* If you are pre-training from scratch, be prepared that pre-training is
computationally expensive, especially on GPUs. If you are pre-training from
scratch, our recommended recipe is to pre-train a `BERT-Base` on a single
[preemptible Cloud TPU v2](https://cloud.google.com/tpu/docs/pricing), which
takes about 2 weeks at a cost of about $500 USD (based on the pricing in
October 2018). You will have to scale down the batch size when only training
on a single Cloud TPU, compared to what was used in the paper. It is
recommended to use the largest batch size that fits into TPU memory.
### Pre-training data
We will **not** be able to release the pre-processed datasets used in the paper.
For Wikipedia, the recommended pre-processing is to download
[the latest dump](https://dumps.wikimedia.org/enwiki/latest/enwiki-latest-pages-articles.xml.bz2),
extract the text with
[`WikiExtractor.py`](https://github.com/attardi/wikiextractor), and then apply
any necessary cleanup to convert it into plain text.
Unfortunately the researchers who collected the
[BookCorpus](http://yknzhu.wixsite.com/mbweb) no longer have it available for
public download. The
[Project Guttenberg Dataset](https://web.eecs.umich.edu/~lahiri/gutenberg_dataset.html)
is a somewhat smaller (200M word) collection of older books that are public
domain.
[Common Crawl](http://commoncrawl.org/) is another very large collection of
text, but you will likely have to do substantial pre-processing and cleanup to
extract a usable corpus for pre-training BERT.
### Learning a new WordPiece vocabulary
This repository does not include code for *learning* a new WordPiece vocabulary.
The reason is that the code used in the paper was implemented in C++ with
dependencies on Google's internal libraries. For English, it is almost always
better to just start with our vocabulary and pre-trained models. For learning
vocabularies of other languages, there are a number of open source options
available. However, keep in mind that these are not compatible with our
`tokenization.py` library:
* [Google's SentencePiece library](https://github.com/google/sentencepiece)
* [tensor2tensor's WordPiece generation script](https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/data_generators/text_encoder_build_subword.py)
* [Rico Sennrich's Byte Pair Encoding library](https://github.com/rsennrich/subword-nmt)
## Using BERT in Colab
If you want to use BERT with [Colab](https://colab.research.google.com), you can
get started with the notebook
"[BERT FineTuning with Cloud TPUs](https://colab.research.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb)".
**At the time of this writing (October 31st, 2018), Colab users can access a
Cloud TPU completely for free.** Note: One per user, availability limited,
requires a Google Cloud Platform account with storage (although storage may be
purchased with free credit for signing up with GCP), and this capability may not
longer be available in the future. Click on the BERT Colab that was just linked
for more information.
## FAQ
#### Is this code compatible with Cloud TPUs? What about GPUs?
Yes, all of the code in this repository works out-of-the-box with CPU, GPU, and
Cloud TPU. However, GPU training is single-GPU only.
#### I am getting out-of-memory errors, what is wrong?
See the section on [out-of-memory issues](#out-of-memory-issues) for more
information.
#### Is there a PyTorch version available?
There is no official PyTorch implementation. However, NLP researchers from
HuggingFace made a
[PyTorch version of BERT available](https://github.com/huggingface/pytorch-pretrained-BERT)
which is compatible with our pre-trained checkpoints and is able to reproduce
our results. We were not involved in the creation or maintenance of the PyTorch
implementation so please direct any questions towards the authors of that
repository.
#### Is there a Chainer version available?
There is no official Chainer implementation. However, Sosuke Kobayashi made a
[Chainer version of BERT available](https://github.com/soskek/bert-chainer)
which is compatible with our pre-trained checkpoints and is able to reproduce
our results. We were not involved in the creation or maintenance of the Chainer
implementation so please direct any questions towards the authors of that
repository.
#### Will models in other languages be released?
Yes, we plan to release a multi-lingual BERT model in the near future. We cannot
make promises about exactly which languages will be included, but it will likely
be a single model which includes *most* of the languages which have a
significantly-sized Wikipedia.
#### Will models larger than `BERT-Large` be released?
So far we have not attempted to train anything larger than `BERT-Large`. It is
possible that we will release larger models if we are able to obtain significant
improvements.
#### What license is this library released under?
All code *and* models are released under the Apache 2.0 license. See the
`LICENSE` file for more information.
#### How do I cite BERT?
For now, cite [the Arxiv paper](https://arxiv.org/abs/1810.04805):
```
@article{devlin2018bert,
title={BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding},
author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
journal={arXiv preprint arXiv:1810.04805},
year={2018}
}
```
If we submit the paper to a conference or journal, we will update the BibTeX.
## Disclaimer
This is not an official Google product.
## Contact information
For help or issues using BERT, please submit a GitHub issue.
For personal communication related to BERT, please contact Jacob Devlin
(`jacobdevlin@google.com`), Ming-Wei Chang (`mingweichang@google.com`), or
Kenton Lee (`kentonl@google.com`).
================================================
FILE: __init__.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
================================================
FILE: create_pretraining_data.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Create masked LM/next sentence masked_lm TF examples for BERT."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import collections
import random
import tokenization
import tensorflow as tf
flags = tf.flags
FLAGS = flags.FLAGS
flags.DEFINE_string("input_file", None,
"Input raw text file (or comma-separated list of files).")
flags.DEFINE_string(
"output_file", None,
"Output TF example file (or comma-separated list of files).")
flags.DEFINE_string("vocab_file", None,
"The vocabulary file that the BERT model was trained on.")
flags.DEFINE_bool(
"do_lower_case", True,
"Whether to lower case the input text. Should be True for uncased "
"models and False for cased models.")
flags.DEFINE_bool(
"do_whole_word_mask", False,
"Whether to use whole word masking rather than per-WordPiece masking.")
flags.DEFINE_integer("max_seq_length", 128, "Maximum sequence length.")
flags.DEFINE_integer("max_predictions_per_seq", 20,
"Maximum number of masked LM predictions per sequence.")
flags.DEFINE_integer("random_seed", 12345, "Random seed for data generation.")
flags.DEFINE_integer(
"dupe_factor", 10,
"Number of times to duplicate the input data (with different masks).")
flags.DEFINE_float("masked_lm_prob", 0.15, "Masked LM probability.")
flags.DEFINE_float(
"short_seq_prob", 0.1,
"Probability of creating sequences which are shorter than the "
"maximum length.")
class TrainingInstance(object):
"""A single training instance (sentence pair)."""
def __init__(self, tokens, segment_ids, masked_lm_positions, masked_lm_labels,
is_random_next):
self.tokens = tokens
self.segment_ids = segment_ids
self.is_random_next = is_random_next
self.masked_lm_positions = masked_lm_positions
self.masked_lm_labels = masked_lm_labels
def __str__(self):
s = ""
s += "tokens: %s\n" % (" ".join(
[tokenization.printable_text(x) for x in self.tokens]))
s += "segment_ids: %s\n" % (" ".join([str(x) for x in self.segment_ids]))
s += "is_random_next: %s\n" % self.is_random_next
s += "masked_lm_positions: %s\n" % (" ".join(
[str(x) for x in self.masked_lm_positions]))
s += "masked_lm_labels: %s\n" % (" ".join(
[tokenization.printable_text(x) for x in self.masked_lm_labels]))
s += "\n"
return s
def __repr__(self):
return self.__str__()
def write_instance_to_example_files(instances, tokenizer, max_seq_length,
max_predictions_per_seq, output_files):
"""Create TF example files from `TrainingInstance`s."""
writers = []
for output_file in output_files:
writers.append(tf.python_io.TFRecordWriter(output_file))
writer_index = 0
total_written = 0
for (inst_index, instance) in enumerate(instances):
input_ids = tokenizer.convert_tokens_to_ids(instance.tokens)
input_mask = [1] * len(input_ids)
segment_ids = list(instance.segment_ids)
assert len(input_ids) <= max_seq_length
while len(input_ids) < max_seq_length:
input_ids.append(0)
input_mask.append(0)
segment_ids.append(0)
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
masked_lm_positions = list(instance.masked_lm_positions)
masked_lm_ids = tokenizer.convert_tokens_to_ids(instance.masked_lm_labels)
masked_lm_weights = [1.0] * len(masked_lm_ids)
while len(masked_lm_positions) < max_predictions_per_seq:
masked_lm_positions.append(0)
masked_lm_ids.append(0)
masked_lm_weights.append(0.0)
next_sentence_label = 1 if instance.is_random_next else 0
features = collections.OrderedDict()
features["input_ids"] = create_int_feature(input_ids)
features["input_mask"] = create_int_feature(input_mask)
features["segment_ids"] = create_int_feature(segment_ids)
features["masked_lm_positions"] = create_int_feature(masked_lm_positions)
features["masked_lm_ids"] = create_int_feature(masked_lm_ids)
features["masked_lm_weights"] = create_float_feature(masked_lm_weights)
features["next_sentence_labels"] = create_int_feature([next_sentence_label])
tf_example = tf.train.Example(features=tf.train.Features(feature=features))
writers[writer_index].write(tf_example.SerializeToString())
writer_index = (writer_index + 1) % len(writers)
total_written += 1
if inst_index < 20:
tf.logging.info("*** Example ***")
tf.logging.info("tokens: %s" % " ".join(
[tokenization.printable_text(x) for x in instance.tokens]))
for feature_name in features.keys():
feature = features[feature_name]
values = []
if feature.int64_list.value:
values = feature.int64_list.value
elif feature.float_list.value:
values = feature.float_list.value
tf.logging.info(
"%s: %s" % (feature_name, " ".join([str(x) for x in values])))
for writer in writers:
writer.close()
tf.logging.info("Wrote %d total instances", total_written)
def create_int_feature(values):
feature = tf.train.Feature(int64_list=tf.train.Int64List(value=list(values)))
return feature
def create_float_feature(values):
feature = tf.train.Feature(float_list=tf.train.FloatList(value=list(values)))
return feature
def create_training_instances(input_files, tokenizer, max_seq_length,
dupe_factor, short_seq_prob, masked_lm_prob,
max_predictions_per_seq, rng):
"""Create `TrainingInstance`s from raw text."""
all_documents = [[]]
# Input file format:
# (1) One sentence per line. These should ideally be actual sentences, not
# entire paragraphs or arbitrary spans of text. (Because we use the
# sentence boundaries for the "next sentence prediction" task).
# (2) Blank lines between documents. Document boundaries are needed so
# that the "next sentence prediction" task doesn't span between documents.
for input_file in input_files:
with tf.gfile.GFile(input_file, "r") as reader:
while True:
line = tokenization.convert_to_unicode(reader.readline())
if not line:
break
line = line.strip()
# Empty lines are used as document delimiters
if not line:
all_documents.append([])
tokens = tokenizer.tokenize(line)
if tokens:
all_documents[-1].append(tokens)
# Remove empty documents
all_documents = [x for x in all_documents if x]
rng.shuffle(all_documents)
vocab_words = list(tokenizer.vocab.keys())
instances = []
for _ in range(dupe_factor):
for document_index in range(len(all_documents)):
instances.extend(
create_instances_from_document(
all_documents, document_index, max_seq_length, short_seq_prob,
masked_lm_prob, max_predictions_per_seq, vocab_words, rng))
rng.shuffle(instances)
return instances
def create_instances_from_document(
all_documents, document_index, max_seq_length, short_seq_prob,
masked_lm_prob, max_predictions_per_seq, vocab_words, rng):
"""Creates `TrainingInstance`s for a single document."""
document = all_documents[document_index]
# Account for [CLS], [SEP], [SEP]
max_num_tokens = max_seq_length - 3
# We *usually* want to fill up the entire sequence since we are padding
# to `max_seq_length` anyways, so short sequences are generally wasted
# computation. However, we *sometimes*
# (i.e., short_seq_prob == 0.1 == 10% of the time) want to use shorter
# sequences to minimize the mismatch between pre-training and fine-tuning.
# The `target_seq_length` is just a rough target however, whereas
# `max_seq_length` is a hard limit.
target_seq_length = max_num_tokens
if rng.random() < short_seq_prob:
target_seq_length = rng.randint(2, max_num_tokens)
# We DON'T just concatenate all of the tokens from a document into a long
# sequence and choose an arbitrary split point because this would make the
# next sentence prediction task too easy. Instead, we split the input into
# segments "A" and "B" based on the actual "sentences" provided by the user
# input.
instances = []
current_chunk = []
current_length = 0
i = 0
while i < len(document):
segment = document[i]
current_chunk.append(segment)
current_length += len(segment)
if i == len(document) - 1 or current_length >= target_seq_length:
if current_chunk:
# `a_end` is how many segments from `current_chunk` go into the `A`
# (first) sentence.
a_end = 1
if len(current_chunk) >= 2:
a_end = rng.randint(1, len(current_chunk) - 1)
tokens_a = []
for j in range(a_end):
tokens_a.extend(current_chunk[j])
tokens_b = []
# Random next
is_random_next = False
if len(current_chunk) == 1 or rng.random() < 0.5:
is_random_next = True
target_b_length = target_seq_length - len(tokens_a)
# This should rarely go for more than one iteration for large
# corpora. However, just to be careful, we try to make sure that
# the random document is not the same as the document
# we're processing.
for _ in range(10):
random_document_index = rng.randint(0, len(all_documents) - 1)
if random_document_index != document_index:
break
random_document = all_documents[random_document_index]
random_start = rng.randint(0, len(random_document) - 1)
for j in range(random_start, len(random_document)):
tokens_b.extend(random_document[j])
if len(tokens_b) >= target_b_length:
break
# We didn't actually use these segments so we "put them back" so
# they don't go to waste.
num_unused_segments = len(current_chunk) - a_end
i -= num_unused_segments
# Actual next
else:
is_random_next = False
for j in range(a_end, len(current_chunk)):
tokens_b.extend(current_chunk[j])
truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng)
assert len(tokens_a) >= 1
assert len(tokens_b) >= 1
tokens = []
segment_ids = []
tokens.append("[CLS]")
segment_ids.append(0)
for token in tokens_a:
tokens.append(token)
segment_ids.append(0)
tokens.append("[SEP]")
segment_ids.append(0)
for token in tokens_b:
tokens.append(token)
segment_ids.append(1)
tokens.append("[SEP]")
segment_ids.append(1)
(tokens, masked_lm_positions,
masked_lm_labels) = create_masked_lm_predictions(
tokens, masked_lm_prob, max_predictions_per_seq, vocab_words, rng)
instance = TrainingInstance(
tokens=tokens,
segment_ids=segment_ids,
is_random_next=is_random_next,
masked_lm_positions=masked_lm_positions,
masked_lm_labels=masked_lm_labels)
instances.append(instance)
current_chunk = []
current_length = 0
i += 1
return instances
MaskedLmInstance = collections.namedtuple("MaskedLmInstance",
["index", "label"])
def create_masked_lm_predictions(tokens, masked_lm_prob,
max_predictions_per_seq, vocab_words, rng):
"""Creates the predictions for the masked LM objective."""
cand_indexes = []
for (i, token) in enumerate(tokens):
if token == "[CLS]" or token == "[SEP]":
continue
# Whole Word Masking means that if we mask all of the wordpieces
# corresponding to an original word. When a word has been split into
# WordPieces, the first token does not have any marker and any subsequence
# tokens are prefixed with ##. So whenever we see the ## token, we
# append it to the previous set of word indexes.
#
# Note that Whole Word Masking does *not* change the training code
# at all -- we still predict each WordPiece independently, softmaxed
# over the entire vocabulary.
if (FLAGS.do_whole_word_mask and len(cand_indexes) >= 1 and
token.startswith("##")):
cand_indexes[-1].append(i)
else:
cand_indexes.append([i])
rng.shuffle(cand_indexes)
output_tokens = list(tokens)
num_to_predict = min(max_predictions_per_seq,
max(1, int(round(len(tokens) * masked_lm_prob))))
masked_lms = []
covered_indexes = set()
for index_set in cand_indexes:
if len(masked_lms) >= num_to_predict:
break
# If adding a whole-word mask would exceed the maximum number of
# predictions, then just skip this candidate.
if len(masked_lms) + len(index_set) > num_to_predict:
continue
is_any_index_covered = False
for index in index_set:
if index in covered_indexes:
is_any_index_covered = True
break
if is_any_index_covered:
continue
for index in index_set:
covered_indexes.add(index)
masked_token = None
# 80% of the time, replace with [MASK]
if rng.random() < 0.8:
masked_token = "[MASK]"
else:
# 10% of the time, keep original
if rng.random() < 0.5:
masked_token = tokens[index]
# 10% of the time, replace with random word
else:
masked_token = vocab_words[rng.randint(0, len(vocab_words) - 1)]
output_tokens[index] = masked_token
masked_lms.append(MaskedLmInstance(index=index, label=tokens[index]))
assert len(masked_lms) <= num_to_predict
masked_lms = sorted(masked_lms, key=lambda x: x.index)
masked_lm_positions = []
masked_lm_labels = []
for p in masked_lms:
masked_lm_positions.append(p.index)
masked_lm_labels.append(p.label)
return (output_tokens, masked_lm_positions, masked_lm_labels)
def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng):
"""Truncates a pair of sequences to a maximum sequence length."""
while True:
total_length = len(tokens_a) + len(tokens_b)
if total_length <= max_num_tokens:
break
trunc_tokens = tokens_a if len(tokens_a) > len(tokens_b) else tokens_b
assert len(trunc_tokens) >= 1
# We want to sometimes truncate from the front and sometimes from the
# back to add more randomness and avoid biases.
if rng.random() < 0.5:
del trunc_tokens[0]
else:
trunc_tokens.pop()
def main(_):
tf.logging.set_verbosity(tf.logging.INFO)
tokenizer = tokenization.FullTokenizer(
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
input_files = []
for input_pattern in FLAGS.input_file.split(","):
input_files.extend(tf.gfile.Glob(input_pattern))
tf.logging.info("*** Reading from input files ***")
for input_file in input_files:
tf.logging.info(" %s", input_file)
rng = random.Random(FLAGS.random_seed)
instances = create_training_instances(
input_files, tokenizer, FLAGS.max_seq_length, FLAGS.dupe_factor,
FLAGS.short_seq_prob, FLAGS.masked_lm_prob, FLAGS.max_predictions_per_seq,
rng)
output_files = FLAGS.output_file.split(",")
tf.logging.info("*** Writing to output files ***")
for output_file in output_files:
tf.logging.info(" %s", output_file)
write_instance_to_example_files(instances, tokenizer, FLAGS.max_seq_length,
FLAGS.max_predictions_per_seq, output_files)
if __name__ == "__main__":
flags.mark_flag_as_required("input_file")
flags.mark_flag_as_required("output_file")
flags.mark_flag_as_required("vocab_file")
tf.app.run()
================================================
FILE: extract_features.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Extract pre-computed feature vectors from BERT."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import codecs
import collections
import json
import re
import modeling
import tokenization
import tensorflow as tf
flags = tf.flags
FLAGS = flags.FLAGS
flags.DEFINE_string("input_file", None, "")
flags.DEFINE_string("output_file", None, "")
flags.DEFINE_string("layers", "-1,-2,-3,-4", "")
flags.DEFINE_string(
"bert_config_file", None,
"The config json file corresponding to the pre-trained BERT model. "
"This specifies the model architecture.")
flags.DEFINE_integer(
"max_seq_length", 128,
"The maximum total input sequence length after WordPiece tokenization. "
"Sequences longer than this will be truncated, and sequences shorter "
"than this will be padded.")
flags.DEFINE_string(
"init_checkpoint", None,
"Initial checkpoint (usually from a pre-trained BERT model).")
flags.DEFINE_string("vocab_file", None,
"The vocabulary file that the BERT model was trained on.")
flags.DEFINE_bool(
"do_lower_case", True,
"Whether to lower case the input text. Should be True for uncased "
"models and False for cased models.")
flags.DEFINE_integer("batch_size", 32, "Batch size for predictions.")
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")
flags.DEFINE_string("master", None,
"If using a TPU, the address of the master.")
flags.DEFINE_integer(
"num_tpu_cores", 8,
"Only used if `use_tpu` is True. Total number of TPU cores to use.")
flags.DEFINE_bool(
"use_one_hot_embeddings", False,
"If True, tf.one_hot will be used for embedding lookups, otherwise "
"tf.nn.embedding_lookup will be used. On TPUs, this should be True "
"since it is much faster.")
class InputExample(object):
def __init__(self, unique_id, text_a, text_b):
self.unique_id = unique_id
self.text_a = text_a
self.text_b = text_b
class InputFeatures(object):
"""A single set of features of data."""
def __init__(self, unique_id, tokens, input_ids, input_mask, input_type_ids):
self.unique_id = unique_id
self.tokens = tokens
self.input_ids = input_ids
self.input_mask = input_mask
self.input_type_ids = input_type_ids
def input_fn_builder(features, seq_length):
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
all_unique_ids = []
all_input_ids = []
all_input_mask = []
all_input_type_ids = []
for feature in features:
all_unique_ids.append(feature.unique_id)
all_input_ids.append(feature.input_ids)
all_input_mask.append(feature.input_mask)
all_input_type_ids.append(feature.input_type_ids)
def input_fn(params):
"""The actual input function."""
batch_size = params["batch_size"]
num_examples = len(features)
# This is for demo purposes and does NOT scale to large data sets. We do
# not use Dataset.from_generator() because that uses tf.py_func which is
# not TPU compatible. The right way to load data is with TFRecordReader.
d = tf.data.Dataset.from_tensor_slices({
"unique_ids":
tf.constant(all_unique_ids, shape=[num_examples], dtype=tf.int32),
"input_ids":
tf.constant(
all_input_ids, shape=[num_examples, seq_length],
dtype=tf.int32),
"input_mask":
tf.constant(
all_input_mask,
shape=[num_examples, seq_length],
dtype=tf.int32),
"input_type_ids":
tf.constant(
all_input_type_ids,
shape=[num_examples, seq_length],
dtype=tf.int32),
})
d = d.batch(batch_size=batch_size, drop_remainder=False)
return d
return input_fn
def model_fn_builder(bert_config, init_checkpoint, layer_indexes, use_tpu,
use_one_hot_embeddings):
"""Returns `model_fn` closure for TPUEstimator."""
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
"""The `model_fn` for TPUEstimator."""
unique_ids = features["unique_ids"]
input_ids = features["input_ids"]
input_mask = features["input_mask"]
input_type_ids = features["input_type_ids"]
model = modeling.BertModel(
config=bert_config,
is_training=False,
input_ids=input_ids,
input_mask=input_mask,
token_type_ids=input_type_ids,
use_one_hot_embeddings=use_one_hot_embeddings)
if mode != tf.estimator.ModeKeys.PREDICT:
raise ValueError("Only PREDICT modes are supported: %s" % (mode))
tvars = tf.trainable_variables()
scaffold_fn = None
(assignment_map,
initialized_variable_names) = modeling.get_assignment_map_from_checkpoint(
tvars, init_checkpoint)
if use_tpu:
def tpu_scaffold():
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
return tf.train.Scaffold()
scaffold_fn = tpu_scaffold
else:
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
tf.logging.info("**** Trainable Variables ****")
for var in tvars:
init_string = ""
if var.name in initialized_variable_names:
init_string = ", *INIT_FROM_CKPT*"
tf.logging.info(" name = %s, shape = %s%s", var.name, var.shape,
init_string)
all_layers = model.get_all_encoder_layers()
predictions = {
"unique_id": unique_ids,
}
for (i, layer_index) in enumerate(layer_indexes):
predictions["layer_output_%d" % i] = all_layers[layer_index]
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
mode=mode, predictions=predictions, scaffold_fn=scaffold_fn)
return output_spec
return model_fn
def convert_examples_to_features(examples, seq_length, tokenizer):
"""Loads a data file into a list of `InputBatch`s."""
features = []
for (ex_index, example) in enumerate(examples):
tokens_a = tokenizer.tokenize(example.text_a)
tokens_b = None
if example.text_b:
tokens_b = tokenizer.tokenize(example.text_b)
if tokens_b:
# Modifies `tokens_a` and `tokens_b` in place so that the total
# length is less than the specified length.
# Account for [CLS], [SEP], [SEP] with "- 3"
_truncate_seq_pair(tokens_a, tokens_b, seq_length - 3)
else:
# Account for [CLS] and [SEP] with "- 2"
if len(tokens_a) > seq_length - 2:
tokens_a = tokens_a[0:(seq_length - 2)]
# The convention in BERT is:
# (a) For sequence pairs:
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
# (b) For single sequences:
# tokens: [CLS] the dog is hairy . [SEP]
# type_ids: 0 0 0 0 0 0 0
#
# Where "type_ids" are used to indicate whether this is the first
# sequence or the second sequence. The embedding vectors for `type=0` and
# `type=1` were learned during pre-training and are added to the wordpiece
# embedding vector (and position vector). This is not *strictly* necessary
# since the [SEP] token unambiguously separates the sequences, but it makes
# it easier for the model to learn the concept of sequences.
#
# For classification tasks, the first vector (corresponding to [CLS]) is
# used as as the "sentence vector". Note that this only makes sense because
# the entire model is fine-tuned.
tokens = []
input_type_ids = []
tokens.append("[CLS]")
input_type_ids.append(0)
for token in tokens_a:
tokens.append(token)
input_type_ids.append(0)
tokens.append("[SEP]")
input_type_ids.append(0)
if tokens_b:
for token in tokens_b:
tokens.append(token)
input_type_ids.append(1)
tokens.append("[SEP]")
input_type_ids.append(1)
input_ids = tokenizer.convert_tokens_to_ids(tokens)
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
input_mask = [1] * len(input_ids)
# Zero-pad up to the sequence length.
while len(input_ids) < seq_length:
input_ids.append(0)
input_mask.append(0)
input_type_ids.append(0)
assert len(input_ids) == seq_length
assert len(input_mask) == seq_length
assert len(input_type_ids) == seq_length
if ex_index < 5:
tf.logging.info("*** Example ***")
tf.logging.info("unique_id: %s" % (example.unique_id))
tf.logging.info("tokens: %s" % " ".join(
[tokenization.printable_text(x) for x in tokens]))
tf.logging.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
tf.logging.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
tf.logging.info(
"input_type_ids: %s" % " ".join([str(x) for x in input_type_ids]))
features.append(
InputFeatures(
unique_id=example.unique_id,
tokens=tokens,
input_ids=input_ids,
input_mask=input_mask,
input_type_ids=input_type_ids))
return features
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
"""Truncates a sequence pair in place to the maximum length."""
# This is a simple heuristic which will always truncate the longer sequence
# one token at a time. This makes more sense than truncating an equal percent
# of tokens from each, since if one sequence is very short then each token
# that's truncated likely contains more information than a longer sequence.
while True:
total_length = len(tokens_a) + len(tokens_b)
if total_length <= max_length:
break
if len(tokens_a) > len(tokens_b):
tokens_a.pop()
else:
tokens_b.pop()
def read_examples(input_file):
"""Read a list of `InputExample`s from an input file."""
examples = []
unique_id = 0
with tf.gfile.GFile(input_file, "r") as reader:
while True:
line = tokenization.convert_to_unicode(reader.readline())
if not line:
break
line = line.strip()
text_a = None
text_b = None
m = re.match(r"^(.*) \|\|\| (.*)$", line)
if m is None:
text_a = line
else:
text_a = m.group(1)
text_b = m.group(2)
examples.append(
InputExample(unique_id=unique_id, text_a=text_a, text_b=text_b))
unique_id += 1
return examples
def main(_):
tf.logging.set_verbosity(tf.logging.INFO)
layer_indexes = [int(x) for x in FLAGS.layers.split(",")]
bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)
tokenizer = tokenization.FullTokenizer(
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
run_config = tf.contrib.tpu.RunConfig(
master=FLAGS.master,
tpu_config=tf.contrib.tpu.TPUConfig(
num_shards=FLAGS.num_tpu_cores,
per_host_input_for_training=is_per_host))
examples = read_examples(FLAGS.input_file)
features = convert_examples_to_features(
examples=examples, seq_length=FLAGS.max_seq_length, tokenizer=tokenizer)
unique_id_to_feature = {}
for feature in features:
unique_id_to_feature[feature.unique_id] = feature
model_fn = model_fn_builder(
bert_config=bert_config,
init_checkpoint=FLAGS.init_checkpoint,
layer_indexes=layer_indexes,
use_tpu=FLAGS.use_tpu,
use_one_hot_embeddings=FLAGS.use_one_hot_embeddings)
# If TPU is not available, this will fall back to normal Estimator on CPU
# or GPU.
estimator = tf.contrib.tpu.TPUEstimator(
use_tpu=FLAGS.use_tpu,
model_fn=model_fn,
config=run_config,
predict_batch_size=FLAGS.batch_size)
input_fn = input_fn_builder(
features=features, seq_length=FLAGS.max_seq_length)
with codecs.getwriter("utf-8")(tf.gfile.Open(FLAGS.output_file,
"w")) as writer:
for result in estimator.predict(input_fn, yield_single_examples=True):
unique_id = int(result["unique_id"])
feature = unique_id_to_feature[unique_id]
output_json = collections.OrderedDict()
output_json["linex_index"] = unique_id
all_features = []
for (i, token) in enumerate(feature.tokens):
all_layers = []
for (j, layer_index) in enumerate(layer_indexes):
layer_output = result["layer_output_%d" % j]
layers = collections.OrderedDict()
layers["index"] = layer_index
layers["values"] = [
round(float(x), 6) for x in layer_output[i:(i + 1)].flat
]
all_layers.append(layers)
features = collections.OrderedDict()
features["token"] = token
features["layers"] = all_layers
all_features.append(features)
output_json["features"] = all_features
writer.write(json.dumps(output_json) + "\n")
if __name__ == "__main__":
flags.mark_flag_as_required("input_file")
flags.mark_flag_as_required("vocab_file")
flags.mark_flag_as_required("bert_config_file")
flags.mark_flag_as_required("init_checkpoint")
flags.mark_flag_as_required("output_file")
tf.app.run()
================================================
FILE: modeling.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""The main BERT model and related functions."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import collections
import copy
import json
import math
import re
import numpy as np
import six
import tensorflow as tf
class BertConfig(object):
"""Configuration for `BertModel`."""
def __init__(self,
vocab_size,
hidden_size=768,
num_hidden_layers=12,
num_attention_heads=12,
intermediate_size=3072,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=512,
type_vocab_size=16,
initializer_range=0.02):
"""Constructs BertConfig.
Args:
vocab_size: Vocabulary size of `inputs_ids` in `BertModel`.
hidden_size: Size of the encoder layers and the pooler layer.
num_hidden_layers: Number of hidden layers in the Transformer encoder.
num_attention_heads: Number of attention heads for each attention layer in
the Transformer encoder.
intermediate_size: The size of the "intermediate" (i.e., feed-forward)
layer in the Transformer encoder.
hidden_act: The non-linear activation function (function or string) in the
encoder and pooler.
hidden_dropout_prob: The dropout probability for all fully connected
layers in the embeddings, encoder, and pooler.
attention_probs_dropout_prob: The dropout ratio for the attention
probabilities.
max_position_embeddings: The maximum sequence length that this model might
ever be used with. Typically set this to something large just in case
(e.g., 512 or 1024 or 2048).
type_vocab_size: The vocabulary size of the `token_type_ids` passed into
`BertModel`.
initializer_range: The stdev of the truncated_normal_initializer for
initializing all weight matrices.
"""
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.hidden_act = hidden_act
self.intermediate_size = intermediate_size
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.type_vocab_size = type_vocab_size
self.initializer_range = initializer_range
@classmethod
def from_dict(cls, json_object):
"""Constructs a `BertConfig` from a Python dictionary of parameters."""
config = BertConfig(vocab_size=None)
for (key, value) in six.iteritems(json_object):
config.__dict__[key] = value
return config
@classmethod
def from_json_file(cls, json_file):
"""Constructs a `BertConfig` from a json file of parameters."""
with tf.gfile.GFile(json_file, "r") as reader:
text = reader.read()
return cls.from_dict(json.loads(text))
def to_dict(self):
"""Serializes this instance to a Python dictionary."""
output = copy.deepcopy(self.__dict__)
return output
def to_json_string(self):
"""Serializes this instance to a JSON string."""
return json.dumps(self.to_dict(), indent=2, sort_keys=True) + "\n"
class BertModel(object):
"""BERT model ("Bidirectional Encoder Representations from Transformers").
Example usage:
```python
# Already been converted into WordPiece token ids
input_ids = tf.constant([[31, 51, 99], [15, 5, 0]])
input_mask = tf.constant([[1, 1, 1], [1, 1, 0]])
token_type_ids = tf.constant([[0, 0, 1], [0, 2, 0]])
config = modeling.BertConfig(vocab_size=32000, hidden_size=512,
num_hidden_layers=8, num_attention_heads=6, intermediate_size=1024)
model = modeling.BertModel(config=config, is_training=True,
input_ids=input_ids, input_mask=input_mask, token_type_ids=token_type_ids)
label_embeddings = tf.get_variable(...)
pooled_output = model.get_pooled_output()
logits = tf.matmul(pooled_output, label_embeddings)
...
```
"""
def __init__(self,
config,
is_training,
input_ids,
input_mask=None,
token_type_ids=None,
use_one_hot_embeddings=False,
scope=None):
"""Constructor for BertModel.
Args:
config: `BertConfig` instance.
is_training: bool. true for training model, false for eval model. Controls
whether dropout will be applied.
input_ids: int32 Tensor of shape [batch_size, seq_length].
input_mask: (optional) int32 Tensor of shape [batch_size, seq_length].
token_type_ids: (optional) int32 Tensor of shape [batch_size, seq_length].
use_one_hot_embeddings: (optional) bool. Whether to use one-hot word
embeddings or tf.embedding_lookup() for the word embeddings.
scope: (optional) variable scope. Defaults to "bert".
Raises:
ValueError: The config is invalid or one of the input tensor shapes
is invalid.
"""
config = copy.deepcopy(config)
if not is_training:
config.hidden_dropout_prob = 0.0
config.attention_probs_dropout_prob = 0.0
input_shape = get_shape_list(input_ids, expected_rank=2)
batch_size = input_shape[0]
seq_length = input_shape[1]
if input_mask is None:
input_mask = tf.ones(shape=[batch_size, seq_length], dtype=tf.int32)
if token_type_ids is None:
token_type_ids = tf.zeros(shape=[batch_size, seq_length], dtype=tf.int32)
with tf.variable_scope(scope, default_name="bert"):
with tf.variable_scope("embeddings"):
# Perform embedding lookup on the word ids.
(self.embedding_output, self.embedding_table) = embedding_lookup(
input_ids=input_ids,
vocab_size=config.vocab_size,
embedding_size=config.hidden_size,
initializer_range=config.initializer_range,
word_embedding_name="word_embeddings",
use_one_hot_embeddings=use_one_hot_embeddings)
# Add positional embeddings and token type embeddings, then layer
# normalize and perform dropout.
self.embedding_output = embedding_postprocessor(
input_tensor=self.embedding_output,
use_token_type=True,
token_type_ids=token_type_ids,
token_type_vocab_size=config.type_vocab_size,
token_type_embedding_name="token_type_embeddings",
use_position_embeddings=True,
position_embedding_name="position_embeddings",
initializer_range=config.initializer_range,
max_position_embeddings=config.max_position_embeddings,
dropout_prob=config.hidden_dropout_prob)
with tf.variable_scope("encoder"):
# This converts a 2D mask of shape [batch_size, seq_length] to a 3D
# mask of shape [batch_size, seq_length, seq_length] which is used
# for the attention scores.
attention_mask = create_attention_mask_from_input_mask(
input_ids, input_mask)
# Run the stacked transformer.
# `sequence_output` shape = [batch_size, seq_length, hidden_size].
self.all_encoder_layers = transformer_model(
input_tensor=self.embedding_output,
attention_mask=attention_mask,
hidden_size=config.hidden_size,
num_hidden_layers=config.num_hidden_layers,
num_attention_heads=config.num_attention_heads,
intermediate_size=config.intermediate_size,
intermediate_act_fn=get_activation(config.hidden_act),
hidden_dropout_prob=config.hidden_dropout_prob,
attention_probs_dropout_prob=config.attention_probs_dropout_prob,
initializer_range=config.initializer_range,
do_return_all_layers=True)
self.sequence_output = self.all_encoder_layers[-1]
# The "pooler" converts the encoded sequence tensor of shape
# [batch_size, seq_length, hidden_size] to a tensor of shape
# [batch_size, hidden_size]. This is necessary for segment-level
# (or segment-pair-level) classification tasks where we need a fixed
# dimensional representation of the segment.
with tf.variable_scope("pooler"):
# We "pool" the model by simply taking the hidden state corresponding
# to the first token. We assume that this has been pre-trained
first_token_tensor = tf.squeeze(self.sequence_output[:, 0:1, :], axis=1)
self.pooled_output = tf.layers.dense(
first_token_tensor,
config.hidden_size,
activation=tf.tanh,
kernel_initializer=create_initializer(config.initializer_range))
def get_pooled_output(self):
return self.pooled_output
def get_sequence_output(self):
"""Gets final hidden layer of encoder.
Returns:
float Tensor of shape [batch_size, seq_length, hidden_size] corresponding
to the final hidden of the transformer encoder.
"""
return self.sequence_output
def get_all_encoder_layers(self):
return self.all_encoder_layers
def get_embedding_output(self):
"""Gets output of the embedding lookup (i.e., input to the transformer).
Returns:
float Tensor of shape [batch_size, seq_length, hidden_size] corresponding
to the output of the embedding layer, after summing the word
embeddings with the positional embeddings and the token type embeddings,
then performing layer normalization. This is the input to the transformer.
"""
return self.embedding_output
def get_embedding_table(self):
return self.embedding_table
def gelu(x):
"""Gaussian Error Linear Unit.
This is a smoother version of the RELU.
Original paper: https://arxiv.org/abs/1606.08415
Args:
x: float Tensor to perform activation.
Returns:
`x` with the GELU activation applied.
"""
cdf = 0.5 * (1.0 + tf.tanh(
(np.sqrt(2 / np.pi) * (x + 0.044715 * tf.pow(x, 3)))))
return x * cdf
def get_activation(activation_string):
"""Maps a string to a Python function, e.g., "relu" => `tf.nn.relu`.
Args:
activation_string: String name of the activation function.
Returns:
A Python function corresponding to the activation function. If
`activation_string` is None, empty, or "linear", this will return None.
If `activation_string` is not a string, it will return `activation_string`.
Raises:
ValueError: The `activation_string` does not correspond to a known
activation.
"""
# We assume that anything that"s not a string is already an activation
# function, so we just return it.
if not isinstance(activation_string, six.string_types):
return activation_string
if not activation_string:
return None
act = activation_string.lower()
if act == "linear":
return None
elif act == "relu":
return tf.nn.relu
elif act == "gelu":
return gelu
elif act == "tanh":
return tf.tanh
else:
raise ValueError("Unsupported activation: %s" % act)
def get_assignment_map_from_checkpoint(tvars, init_checkpoint):
"""Compute the union of the current variables and checkpoint variables."""
assignment_map = {}
initialized_variable_names = {}
name_to_variable = collections.OrderedDict()
for var in tvars:
name = var.name
m = re.match("^(.*):\\d+$", name)
if m is not None:
name = m.group(1)
name_to_variable[name] = var
init_vars = tf.train.list_variables(init_checkpoint)
assignment_map = collections.OrderedDict()
for x in init_vars:
(name, var) = (x[0], x[1])
if name not in name_to_variable:
continue
assignment_map[name] = name
initialized_variable_names[name] = 1
initialized_variable_names[name + ":0"] = 1
return (assignment_map, initialized_variable_names)
def dropout(input_tensor, dropout_prob):
"""Perform dropout.
Args:
input_tensor: float Tensor.
dropout_prob: Python float. The probability of dropping out a value (NOT of
*keeping* a dimension as in `tf.nn.dropout`).
Returns:
A version of `input_tensor` with dropout applied.
"""
if dropout_prob is None or dropout_prob == 0.0:
return input_tensor
output = tf.nn.dropout(input_tensor, 1.0 - dropout_prob)
return output
def layer_norm(input_tensor, name=None):
"""Run layer normalization on the last dimension of the tensor."""
return tf.contrib.layers.layer_norm(
inputs=input_tensor, begin_norm_axis=-1, begin_params_axis=-1, scope=name)
def layer_norm_and_dropout(input_tensor, dropout_prob, name=None):
"""Runs layer normalization followed by dropout."""
output_tensor = layer_norm(input_tensor, name)
output_tensor = dropout(output_tensor, dropout_prob)
return output_tensor
def create_initializer(initializer_range=0.02):
"""Creates a `truncated_normal_initializer` with the given range."""
return tf.truncated_normal_initializer(stddev=initializer_range)
def embedding_lookup(input_ids,
vocab_size,
embedding_size=128,
initializer_range=0.02,
word_embedding_name="word_embeddings",
use_one_hot_embeddings=False):
"""Looks up words embeddings for id tensor.
Args:
input_ids: int32 Tensor of shape [batch_size, seq_length] containing word
ids.
vocab_size: int. Size of the embedding vocabulary.
embedding_size: int. Width of the word embeddings.
initializer_range: float. Embedding initialization range.
word_embedding_name: string. Name of the embedding table.
use_one_hot_embeddings: bool. If True, use one-hot method for word
embeddings. If False, use `tf.gather()`.
Returns:
float Tensor of shape [batch_size, seq_length, embedding_size].
"""
# This function assumes that the input is of shape [batch_size, seq_length,
# num_inputs].
#
# If the input is a 2D tensor of shape [batch_size, seq_length], we
# reshape to [batch_size, seq_length, 1].
if input_ids.shape.ndims == 2:
input_ids = tf.expand_dims(input_ids, axis=[-1])
embedding_table = tf.get_variable(
name=word_embedding_name,
shape=[vocab_size, embedding_size],
initializer=create_initializer(initializer_range))
flat_input_ids = tf.reshape(input_ids, [-1])
if use_one_hot_embeddings:
one_hot_input_ids = tf.one_hot(flat_input_ids, depth=vocab_size)
output = tf.matmul(one_hot_input_ids, embedding_table)
else:
output = tf.gather(embedding_table, flat_input_ids)
input_shape = get_shape_list(input_ids)
output = tf.reshape(output,
input_shape[0:-1] + [input_shape[-1] * embedding_size])
return (output, embedding_table)
def embedding_postprocessor(input_tensor,
use_token_type=False,
token_type_ids=None,
token_type_vocab_size=16,
token_type_embedding_name="token_type_embeddings",
use_position_embeddings=True,
position_embedding_name="position_embeddings",
initializer_range=0.02,
max_position_embeddings=512,
dropout_prob=0.1):
"""Performs various post-processing on a word embedding tensor.
Args:
input_tensor: float Tensor of shape [batch_size, seq_length,
embedding_size].
use_token_type: bool. Whether to add embeddings for `token_type_ids`.
token_type_ids: (optional) int32 Tensor of shape [batch_size, seq_length].
Must be specified if `use_token_type` is True.
token_type_vocab_size: int. The vocabulary size of `token_type_ids`.
token_type_embedding_name: string. The name of the embedding table variable
for token type ids.
use_position_embeddings: bool. Whether to add position embeddings for the
position of each token in the sequence.
position_embedding_name: string. The name of the embedding table variable
for positional embeddings.
initializer_range: float. Range of the weight initialization.
max_position_embeddings: int. Maximum sequence length that might ever be
used with this model. This can be longer than the sequence length of
input_tensor, but cannot be shorter.
dropout_prob: float. Dropout probability applied to the final output tensor.
Returns:
float tensor with same shape as `input_tensor`.
Raises:
ValueError: One of the tensor shapes or input values is invalid.
"""
input_shape = get_shape_list(input_tensor, expected_rank=3)
batch_size = input_shape[0]
seq_length = input_shape[1]
width = input_shape[2]
output = input_tensor
if use_token_type:
if token_type_ids is None:
raise ValueError("`token_type_ids` must be specified if"
"`use_token_type` is True.")
token_type_table = tf.get_variable(
name=token_type_embedding_name,
shape=[token_type_vocab_size, width],
initializer=create_initializer(initializer_range))
# This vocab will be small so we always do one-hot here, since it is always
# faster for a small vocabulary.
flat_token_type_ids = tf.reshape(token_type_ids, [-1])
one_hot_ids = tf.one_hot(flat_token_type_ids, depth=token_type_vocab_size)
token_type_embeddings = tf.matmul(one_hot_ids, token_type_table)
token_type_embeddings = tf.reshape(token_type_embeddings,
[batch_size, seq_length, width])
output += token_type_embeddings
if use_position_embeddings:
assert_op = tf.assert_less_equal(seq_length, max_position_embeddings)
with tf.control_dependencies([assert_op]):
full_position_embeddings = tf.get_variable(
name=position_embedding_name,
shape=[max_position_embeddings, width],
initializer=create_initializer(initializer_range))
# Since the position embedding table is a learned variable, we create it
# using a (long) sequence length `max_position_embeddings`. The actual
# sequence length might be shorter than this, for faster training of
# tasks that do not have long sequences.
#
# So `full_position_embeddings` is effectively an embedding table
# for position [0, 1, 2, ..., max_position_embeddings-1], and the current
# sequence has positions [0, 1, 2, ... seq_length-1], so we can just
# perform a slice.
position_embeddings = tf.slice(full_position_embeddings, [0, 0],
[seq_length, -1])
num_dims = len(output.shape.as_list())
# Only the last two dimensions are relevant (`seq_length` and `width`), so
# we broadcast among the first dimensions, which is typically just
# the batch size.
position_broadcast_shape = []
for _ in range(num_dims - 2):
position_broadcast_shape.append(1)
position_broadcast_shape.extend([seq_length, width])
position_embeddings = tf.reshape(position_embeddings,
position_broadcast_shape)
output += position_embeddings
output = layer_norm_and_dropout(output, dropout_prob)
return output
def create_attention_mask_from_input_mask(from_tensor, to_mask):
"""Create 3D attention mask from a 2D tensor mask.
Args:
from_tensor: 2D or 3D Tensor of shape [batch_size, from_seq_length, ...].
to_mask: int32 Tensor of shape [batch_size, to_seq_length].
Returns:
float Tensor of shape [batch_size, from_seq_length, to_seq_length].
"""
from_shape = get_shape_list(from_tensor, expected_rank=[2, 3])
batch_size = from_shape[0]
from_seq_length = from_shape[1]
to_shape = get_shape_list(to_mask, expected_rank=2)
to_seq_length = to_shape[1]
to_mask = tf.cast(
tf.reshape(to_mask, [batch_size, 1, to_seq_length]), tf.float32)
# We don't assume that `from_tensor` is a mask (although it could be). We
# don't actually care if we attend *from* padding tokens (only *to* padding)
# tokens so we create a tensor of all ones.
#
# `broadcast_ones` = [batch_size, from_seq_length, 1]
broadcast_ones = tf.ones(
shape=[batch_size, from_seq_length, 1], dtype=tf.float32)
# Here we broadcast along two dimensions to create the mask.
mask = broadcast_ones * to_mask
return mask
def attention_layer(from_tensor,
to_tensor,
attention_mask=None,
num_attention_heads=1,
size_per_head=512,
query_act=None,
key_act=None,
value_act=None,
attention_probs_dropout_prob=0.0,
initializer_range=0.02,
do_return_2d_tensor=False,
batch_size=None,
from_seq_length=None,
to_seq_length=None):
"""Performs multi-headed attention from `from_tensor` to `to_tensor`.
This is an implementation of multi-headed attention based on "Attention
is all you Need". If `from_tensor` and `to_tensor` are the same, then
this is self-attention. Each timestep in `from_tensor` attends to the
corresponding sequence in `to_tensor`, and returns a fixed-with vector.
This function first projects `from_tensor` into a "query" tensor and
`to_tensor` into "key" and "value" tensors. These are (effectively) a list
of tensors of length `num_attention_heads`, where each tensor is of shape
[batch_size, seq_length, size_per_head].
Then, the query and key tensors are dot-producted and scaled. These are
softmaxed to obtain attention probabilities. The value tensors are then
interpolated by these probabilities, then concatenated back to a single
tensor and returned.
In practice, the multi-headed attention are done with transposes and
reshapes rather than actual separate tensors.
Args:
from_tensor: float Tensor of shape [batch_size, from_seq_length,
from_width].
to_tensor: float Tensor of shape [batch_size, to_seq_length, to_width].
attention_mask: (optional) int32 Tensor of shape [batch_size,
from_seq_length, to_seq_length]. The values should be 1 or 0. The
attention scores will effectively be set to -infinity for any positions in
the mask that are 0, and will be unchanged for positions that are 1.
num_attention_heads: int. Number of attention heads.
size_per_head: int. Size of each attention head.
query_act: (optional) Activation function for the query transform.
key_act: (optional) Activation function for the key transform.
value_act: (optional) Activation function for the value transform.
attention_probs_dropout_prob: (optional) float. Dropout probability of the
attention probabilities.
initializer_range: float. Range of the weight initializer.
do_return_2d_tensor: bool. If True, the output will be of shape [batch_size
* from_seq_length, num_attention_heads * size_per_head]. If False, the
output will be of shape [batch_size, from_seq_length, num_attention_heads
* size_per_head].
batch_size: (Optional) int. If the input is 2D, this might be the batch size
of the 3D version of the `from_tensor` and `to_tensor`.
from_seq_length: (Optional) If the input is 2D, this might be the seq length
of the 3D version of the `from_tensor`.
to_seq_length: (Optional) If the input is 2D, this might be the seq length
of the 3D version of the `to_tensor`.
Returns:
float Tensor of shape [batch_size, from_seq_length,
num_attention_heads * size_per_head]. (If `do_return_2d_tensor` is
true, this will be of shape [batch_size * from_seq_length,
num_attention_heads * size_per_head]).
Raises:
ValueError: Any of the arguments or tensor shapes are invalid.
"""
def transpose_for_scores(input_tensor, batch_size, num_attention_heads,
seq_length, width):
output_tensor = tf.reshape(
input_tensor, [batch_size, seq_length, num_attention_heads, width])
output_tensor = tf.transpose(output_tensor, [0, 2, 1, 3])
return output_tensor
from_shape = get_shape_list(from_tensor, expected_rank=[2, 3])
to_shape = get_shape_list(to_tensor, expected_rank=[2, 3])
if len(from_shape) != len(to_shape):
raise ValueError(
"The rank of `from_tensor` must match the rank of `to_tensor`.")
if len(from_shape) == 3:
batch_size = from_shape[0]
from_seq_length = from_shape[1]
to_seq_length = to_shape[1]
elif len(from_shape) == 2:
if (batch_size is None or from_seq_length is None or to_seq_length is None):
raise ValueError(
"When passing in rank 2 tensors to attention_layer, the values "
"for `batch_size`, `from_seq_length`, and `to_seq_length` "
"must all be specified.")
# Scalar dimensions referenced here:
# B = batch size (number of sequences)
# F = `from_tensor` sequence length
# T = `to_tensor` sequence length
# N = `num_attention_heads`
# H = `size_per_head`
from_tensor_2d = reshape_to_matrix(from_tensor)
to_tensor_2d = reshape_to_matrix(to_tensor)
# `query_layer` = [B*F, N*H]
query_layer = tf.layers.dense(
from_tensor_2d,
num_attention_heads * size_per_head,
activation=query_act,
name="query",
kernel_initializer=create_initializer(initializer_range))
# `key_layer` = [B*T, N*H]
key_layer = tf.layers.dense(
to_tensor_2d,
num_attention_heads * size_per_head,
activation=key_act,
name="key",
kernel_initializer=create_initializer(initializer_range))
# `value_layer` = [B*T, N*H]
value_layer = tf.layers.dense(
to_tensor_2d,
num_attention_heads * size_per_head,
activation=value_act,
name="value",
kernel_initializer=create_initializer(initializer_range))
# `query_layer` = [B, N, F, H]
query_layer = transpose_for_scores(query_layer, batch_size,
num_attention_heads, from_seq_length,
size_per_head)
# `key_layer` = [B, N, T, H]
key_layer = transpose_for_scores(key_layer, batch_size, num_attention_heads,
to_seq_length, size_per_head)
# Take the dot product between "query" and "key" to get the raw
# attention scores.
# `attention_scores` = [B, N, F, T]
attention_scores = tf.matmul(query_layer, key_layer, transpose_b=True)
attention_scores = tf.multiply(attention_scores,
1.0 / math.sqrt(float(size_per_head)))
if attention_mask is not None:
# `attention_mask` = [B, 1, F, T]
attention_mask = tf.expand_dims(attention_mask, axis=[1])
# Since attention_mask is 1.0 for positions we want to attend and 0.0 for
# masked positions, this operation will create a tensor which is 0.0 for
# positions we want to attend and -10000.0 for masked positions.
adder = (1.0 - tf.cast(attention_mask, tf.float32)) * -10000.0
# Since we are adding it to the raw scores before the softmax, this is
# effectively the same as removing these entirely.
attention_scores += adder
# Normalize the attention scores to probabilities.
# `attention_probs` = [B, N, F, T]
attention_probs = tf.nn.softmax(attention_scores)
# This is actually dropping out entire tokens to attend to, which might
# seem a bit unusual, but is taken from the original Transformer paper.
attention_probs = dropout(attention_probs, attention_probs_dropout_prob)
# `value_layer` = [B, T, N, H]
value_layer = tf.reshape(
value_layer,
[batch_size, to_seq_length, num_attention_heads, size_per_head])
# `value_layer` = [B, N, T, H]
value_layer = tf.transpose(value_layer, [0, 2, 1, 3])
# `context_layer` = [B, N, F, H]
context_layer = tf.matmul(attention_probs, value_layer)
# `context_layer` = [B, F, N, H]
context_layer = tf.transpose(context_layer, [0, 2, 1, 3])
if do_return_2d_tensor:
# `context_layer` = [B*F, N*H]
context_layer = tf.reshape(
context_layer,
[batch_size * from_seq_length, num_attention_heads * size_per_head])
else:
# `context_layer` = [B, F, N*H]
context_layer = tf.reshape(
context_layer,
[batch_size, from_seq_length, num_attention_heads * size_per_head])
return context_layer
def transformer_model(input_tensor,
attention_mask=None,
hidden_size=768,
num_hidden_layers=12,
num_attention_heads=12,
intermediate_size=3072,
intermediate_act_fn=gelu,
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
initializer_range=0.02,
do_return_all_layers=False):
"""Multi-headed, multi-layer Transformer from "Attention is All You Need".
This is almost an exact implementation of the original Transformer encoder.
See the original paper:
https://arxiv.org/abs/1706.03762
Also see:
https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/models/transformer.py
Args:
input_tensor: float Tensor of shape [batch_size, seq_length, hidden_size].
attention_mask: (optional) int32 Tensor of shape [batch_size, seq_length,
seq_length], with 1 for positions that can be attended to and 0 in
positions that should not be.
hidden_size: int. Hidden size of the Transformer.
num_hidden_layers: int. Number of layers (blocks) in the Transformer.
num_attention_heads: int. Number of attention heads in the Transformer.
intermediate_size: int. The size of the "intermediate" (a.k.a., feed
forward) layer.
intermediate_act_fn: function. The non-linear activation function to apply
to the output of the intermediate/feed-forward layer.
hidden_dropout_prob: float. Dropout probability for the hidden layers.
attention_probs_dropout_prob: float. Dropout probability of the attention
probabilities.
initializer_range: float. Range of the initializer (stddev of truncated
normal).
do_return_all_layers: Whether to also return all layers or just the final
layer.
Returns:
float Tensor of shape [batch_size, seq_length, hidden_size], the final
hidden layer of the Transformer.
Raises:
ValueError: A Tensor shape or parameter is invalid.
"""
if hidden_size % num_attention_heads != 0:
raise ValueError(
"The hidden size (%d) is not a multiple of the number of attention "
"heads (%d)" % (hidden_size, num_attention_heads))
attention_head_size = int(hidden_size / num_attention_heads)
input_shape = get_shape_list(input_tensor, expected_rank=3)
batch_size = input_shape[0]
seq_length = input_shape[1]
input_width = input_shape[2]
# The Transformer performs sum residuals on all layers so the input needs
# to be the same as the hidden size.
if input_width != hidden_size:
raise ValueError("The width of the input tensor (%d) != hidden size (%d)" %
(input_width, hidden_size))
# We keep the representation as a 2D tensor to avoid re-shaping it back and
# forth from a 3D tensor to a 2D tensor. Re-shapes are normally free on
# the GPU/CPU but may not be free on the TPU, so we want to minimize them to
# help the optimizer.
prev_output = reshape_to_matrix(input_tensor)
all_layer_outputs = []
for layer_idx in range(num_hidden_layers):
with tf.variable_scope("layer_%d" % layer_idx):
layer_input = prev_output
with tf.variable_scope("attention"):
attention_heads = []
with tf.variable_scope("self"):
attention_head = attention_layer(
from_tensor=layer_input,
to_tensor=layer_input,
attention_mask=attention_mask,
num_attention_heads=num_attention_heads,
size_per_head=attention_head_size,
attention_probs_dropout_prob=attention_probs_dropout_prob,
initializer_range=initializer_range,
do_return_2d_tensor=True,
batch_size=batch_size,
from_seq_length=seq_length,
to_seq_length=seq_length)
attention_heads.append(attention_head)
attention_output = None
if len(attention_heads) == 1:
attention_output = attention_heads[0]
else:
# In the case where we have other sequences, we just concatenate
# them to the self-attention head before the projection.
attention_output = tf.concat(attention_heads, axis=-1)
# Run a linear projection of `hidden_size` then add a residual
# with `layer_input`.
with tf.variable_scope("output"):
attention_output = tf.layers.dense(
attention_output,
hidden_size,
kernel_initializer=create_initializer(initializer_range))
attention_output = dropout(attention_output, hidden_dropout_prob)
attention_output = layer_norm(attention_output + layer_input)
# The activation is only applied to the "intermediate" hidden layer.
with tf.variable_scope("intermediate"):
intermediate_output = tf.layers.dense(
attention_output,
intermediate_size,
activation=intermediate_act_fn,
kernel_initializer=create_initializer(initializer_range))
# Down-project back to `hidden_size` then add the residual.
with tf.variable_scope("output"):
layer_output = tf.layers.dense(
intermediate_output,
hidden_size,
kernel_initializer=create_initializer(initializer_range))
layer_output = dropout(layer_output, hidden_dropout_prob)
layer_output = layer_norm(layer_output + attention_output)
prev_output = layer_output
all_layer_outputs.append(layer_output)
if do_return_all_layers:
final_outputs = []
for layer_output in all_layer_outputs:
final_output = reshape_from_matrix(layer_output, input_shape)
final_outputs.append(final_output)
return final_outputs
else:
final_output = reshape_from_matrix(prev_output, input_shape)
return final_output
def get_shape_list(tensor, expected_rank=None, name=None):
"""Returns a list of the shape of tensor, preferring static dimensions.
Args:
tensor: A tf.Tensor object to find the shape of.
expected_rank: (optional) int. The expected rank of `tensor`. If this is
specified and the `tensor` has a different rank, and exception will be
thrown.
name: Optional name of the tensor for the error message.
Returns:
A list of dimensions of the shape of tensor. All static dimensions will
be returned as python integers, and dynamic dimensions will be returned
as tf.Tensor scalars.
"""
if name is None:
name = tensor.name
if expected_rank is not None:
assert_rank(tensor, expected_rank, name)
shape = tensor.shape.as_list()
non_static_indexes = []
for (index, dim) in enumerate(shape):
if dim is None:
non_static_indexes.append(index)
if not non_static_indexes:
return shape
dyn_shape = tf.shape(tensor)
for index in non_static_indexes:
shape[index] = dyn_shape[index]
return shape
def reshape_to_matrix(input_tensor):
"""Reshapes a >= rank 2 tensor to a rank 2 tensor (i.e., a matrix)."""
ndims = input_tensor.shape.ndims
if ndims < 2:
raise ValueError("Input tensor must have at least rank 2. Shape = %s" %
(input_tensor.shape))
if ndims == 2:
return input_tensor
width = input_tensor.shape[-1]
output_tensor = tf.reshape(input_tensor, [-1, width])
return output_tensor
def reshape_from_matrix(output_tensor, orig_shape_list):
"""Reshapes a rank 2 tensor back to its original rank >= 2 tensor."""
if len(orig_shape_list) == 2:
return output_tensor
output_shape = get_shape_list(output_tensor)
orig_dims = orig_shape_list[0:-1]
width = output_shape[-1]
return tf.reshape(output_tensor, orig_dims + [width])
def assert_rank(tensor, expected_rank, name=None):
"""Raises an exception if the tensor rank is not of the expected rank.
Args:
tensor: A tf.Tensor to check the rank of.
expected_rank: Python integer or list of integers, expected rank.
name: Optional name of the tensor for the error message.
Raises:
ValueError: If the expected shape doesn't match the actual shape.
"""
if name is None:
name = tensor.name
expected_rank_dict = {}
if isinstance(expected_rank, six.integer_types):
expected_rank_dict[expected_rank] = True
else:
for x in expected_rank:
expected_rank_dict[x] = True
actual_rank = tensor.shape.ndims
if actual_rank not in expected_rank_dict:
scope_name = tf.get_variable_scope().name
raise ValueError(
"For the tensor `%s` in scope `%s`, the actual rank "
"`%d` (shape = %s) is not equal to the expected rank `%s`" %
(name, scope_name, actual_rank, str(tensor.shape), str(expected_rank)))
================================================
FILE: modeling_test.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import collections
import json
import random
import re
import modeling
import six
import tensorflow as tf
class BertModelTest(tf.test.TestCase):
class BertModelTester(object):
def __init__(self,
parent,
batch_size=13,
seq_length=7,
is_training=True,
use_input_mask=True,
use_token_type_ids=True,
vocab_size=99,
hidden_size=32,
num_hidden_layers=5,
num_attention_heads=4,
intermediate_size=37,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=512,
type_vocab_size=16,
initializer_range=0.02,
scope=None):
self.parent = parent
self.batch_size = batch_size
self.seq_length = seq_length
self.is_training = is_training
self.use_input_mask = use_input_mask
self.use_token_type_ids = use_token_type_ids
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.type_vocab_size = type_vocab_size
self.initializer_range = initializer_range
self.scope = scope
def create_model(self):
input_ids = BertModelTest.ids_tensor([self.batch_size, self.seq_length],
self.vocab_size)
input_mask = None
if self.use_input_mask:
input_mask = BertModelTest.ids_tensor(
[self.batch_size, self.seq_length], vocab_size=2)
token_type_ids = None
if self.use_token_type_ids:
token_type_ids = BertModelTest.ids_tensor(
[self.batch_size, self.seq_length], self.type_vocab_size)
config = modeling.BertConfig(
vocab_size=self.vocab_size,
hidden_size=self.hidden_size,
num_hidden_layers=self.num_hidden_layers,
num_attention_heads=self.num_attention_heads,
intermediate_size=self.intermediate_size,
hidden_act=self.hidden_act,
hidden_dropout_prob=self.hidden_dropout_prob,
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
max_position_embeddings=self.max_position_embeddings,
type_vocab_size=self.type_vocab_size,
initializer_range=self.initializer_range)
model = modeling.BertModel(
config=config,
is_training=self.is_training,
input_ids=input_ids,
input_mask=input_mask,
token_type_ids=token_type_ids,
scope=self.scope)
outputs = {
"embedding_output": model.get_embedding_output(),
"sequence_output": model.get_sequence_output(),
"pooled_output": model.get_pooled_output(),
"all_encoder_layers": model.get_all_encoder_layers(),
}
return outputs
def check_output(self, result):
self.parent.assertAllEqual(
result["embedding_output"].shape,
[self.batch_size, self.seq_length, self.hidden_size])
self.parent.assertAllEqual(
result["sequence_output"].shape,
[self.batch_size, self.seq_length, self.hidden_size])
self.parent.assertAllEqual(result["pooled_output"].shape,
[self.batch_size, self.hidden_size])
def test_default(self):
self.run_tester(BertModelTest.BertModelTester(self))
def test_config_to_json_string(self):
config = modeling.BertConfig(vocab_size=99, hidden_size=37)
obj = json.loads(config.to_json_string())
self.assertEqual(obj["vocab_size"], 99)
self.assertEqual(obj["hidden_size"], 37)
def run_tester(self, tester):
with self.test_session() as sess:
ops = tester.create_model()
init_op = tf.group(tf.global_variables_initializer(),
tf.local_variables_initializer())
sess.run(init_op)
output_result = sess.run(ops)
tester.check_output(output_result)
self.assert_all_tensors_reachable(sess, [init_op, ops])
@classmethod
def ids_tensor(cls, shape, vocab_size, rng=None, name=None):
"""Creates a random int32 tensor of the shape within the vocab size."""
if rng is None:
rng = random.Random()
total_dims = 1
for dim in shape:
total_dims *= dim
values = []
for _ in range(total_dims):
values.append(rng.randint(0, vocab_size - 1))
return tf.constant(value=values, dtype=tf.int32, shape=shape, name=name)
def assert_all_tensors_reachable(self, sess, outputs):
"""Checks that all the tensors in the graph are reachable from outputs."""
graph = sess.graph
ignore_strings = [
"^.*/assert_less_equal/.*$",
"^.*/dilation_rate$",
"^.*/Tensordot/concat$",
"^.*/Tensordot/concat/axis$",
"^testing/.*$",
]
ignore_regexes = [re.compile(x) for x in ignore_strings]
unreachable = self.get_unreachable_ops(graph, outputs)
filtered_unreachable = []
for x in unreachable:
do_ignore = False
for r in ignore_regexes:
m = r.match(x.name)
if m is not None:
do_ignore = True
if do_ignore:
continue
filtered_unreachable.append(x)
unreachable = filtered_unreachable
self.assertEqual(
len(unreachable), 0, "The following ops are unreachable: %s" %
(" ".join([x.name for x in unreachable])))
@classmethod
def get_unreachable_ops(cls, graph, outputs):
"""Finds all of the tensors in graph that are unreachable from outputs."""
outputs = cls.flatten_recursive(outputs)
output_to_op = collections.defaultdict(list)
op_to_all = collections.defaultdict(list)
assign_out_to_in = collections.defaultdict(list)
for op in graph.get_operations():
for x in op.inputs:
op_to_all[op.name].append(x.name)
for y in op.outputs:
output_to_op[y.name].append(op.name)
op_to_all[op.name].append(y.name)
if str(op.type) == "Assign":
for y in op.outputs:
for x in op.inputs:
assign_out_to_in[y.name].append(x.name)
assign_groups = collections.defaultdict(list)
for out_name in assign_out_to_in.keys():
name_group = assign_out_to_in[out_name]
for n1 in name_group:
assign_groups[n1].append(out_name)
for n2 in name_group:
if n1 != n2:
assign_groups[n1].append(n2)
seen_tensors = {}
stack = [x.name for x in outputs]
while stack:
name = stack.pop()
if name in seen_tensors:
continue
seen_tensors[name] = True
if name in output_to_op:
for op_name in output_to_op[name]:
if op_name in op_to_all:
for input_name in op_to_all[op_name]:
if input_name not in stack:
stack.append(input_name)
expanded_names = []
if name in assign_groups:
for assign_name in assign_groups[name]:
expanded_names.append(assign_name)
for expanded_name in expanded_names:
if expanded_name not in stack:
stack.append(expanded_name)
unreachable_ops = []
for op in graph.get_operations():
is_unreachable = False
all_names = [x.name for x in op.inputs] + [x.name for x in op.outputs]
for name in all_names:
if name not in seen_tensors:
is_unreachable = True
if is_unreachable:
unreachable_ops.append(op)
return unreachable_ops
@classmethod
def flatten_recursive(cls, item):
"""Flattens (potentially nested) a tuple/dictionary/list to a list."""
output = []
if isinstance(item, list):
output.extend(item)
elif isinstance(item, tuple):
output.extend(list(item))
elif isinstance(item, dict):
for (_, v) in six.iteritems(item):
output.append(v)
else:
return [item]
flat_output = []
for x in output:
flat_output.extend(cls.flatten_recursive(x))
return flat_output
if __name__ == "__main__":
tf.test.main()
================================================
FILE: multilingual.md
================================================
## Models
There are two multilingual models currently available. We do not plan to release
more single-language models, but we may release `BERT-Large` versions of these
two in the future:
* **[`BERT-Base, Multilingual Cased (New, recommended)`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Base, Multilingual Uncased (Orig, not recommended)`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)**:
102 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
parameters
**The `Multilingual Cased (New)` model also fixes normalization issues in many
languages, so it is recommended in languages with non-Latin alphabets (and is
often better for most languages with Latin alphabets). When using this model,
make sure to pass `--do_lower_case=false` to `run_pretraining.py` and other
scripts.**
See the [list of languages](#list-of-languages) that the Multilingual model
supports. The Multilingual model does include Chinese (and English), but if your
fine-tuning data is Chinese-only, then the Chinese model will likely produce
better results.
## Results
To evaluate these systems, we use the
[XNLI dataset](https://github.com/facebookresearch/XNLI) dataset, which is a
version of [MultiNLI](https://www.nyu.edu/projects/bowman/multinli/) where the
dev and test sets have been translated (by humans) into 15 languages. Note that
the training set was *machine* translated (we used the translations provided by
XNLI, not Google NMT). For clarity, we only report on 6 languages below:
<!-- mdformat off(no table) -->
| System | English | Chinese | Spanish | German | Arabic | Urdu |
| --------------------------------- | -------- | -------- | -------- | -------- | -------- | -------- |
| XNLI Baseline - Translate Train | 73.7 | 67.0 | 68.8 | 66.5 | 65.8 | 56.6 |
| XNLI Baseline - Translate Test | 73.7 | 68.3 | 70.7 | 68.7 | 66.8 | 59.3 |
| BERT - Translate Train Cased | **81.9** | **76.6** | **77.8** | **75.9** | **70.7** | 61.6 |
| BERT - Translate Train Uncased | 81.4 | 74.2 | 77.3 | 75.2 | 70.5 | 61.7 |
| BERT - Translate Test Uncased | 81.4 | 70.1 | 74.9 | 74.4 | 70.4 | **62.1** |
| BERT - Zero Shot Uncased | 81.4 | 63.8 | 74.3 | 70.5 | 62.1 | 58.3 |
<!-- mdformat on -->
The first two rows are baselines from the XNLI paper and the last three rows are
our results with BERT.
**Translate Train** means that the MultiNLI training set was machine translated
from English into the foreign language. So training and evaluation were both
done in the foreign language. Unfortunately, training was done on
machine-translated data, so it is impossible to quantify how much of the lower
accuracy (compared to English) is due to the quality of the machine translation
vs. the quality of the pre-trained model.
**Translate Test** means that the XNLI test set was machine translated from the
foreign language into English. So training and evaluation were both done on
English. However, test evaluation was done on machine-translated English, so the
accuracy depends on the quality of the machine translation system.
**Zero Shot** means that the Multilingual BERT system was fine-tuned on English
MultiNLI, and then evaluated on the foreign language XNLI test. In this case,
machine translation was not involved at all in either the pre-training or
fine-tuning.
Note that the English result is worse than the 84.2 MultiNLI baseline because
this training used Multilingual BERT rather than English-only BERT. This implies
that for high-resource languages, the Multilingual model is somewhat worse than
a single-language model. However, it is not feasible for us to train and
maintain dozens of single-language models. Therefore, if your goal is to maximize
performance with a language other than English or Chinese, you might find it
beneficial to run pre-training for additional steps starting from our
Multilingual model on data from your language of interest.
Here is a comparison of training Chinese models with the Multilingual
`BERT-Base` and Chinese-only `BERT-Base`:
System | Chinese
----------------------- | -------
XNLI Baseline | 67.0
BERT Multilingual Model | 74.2
BERT Chinese-only Model | 77.2
Similar to English, the single-language model does 3% better than the
Multilingual model.
## Fine-tuning Example
The multilingual model does **not** require any special consideration or API
changes. We did update the implementation of `BasicTokenizer` in
`tokenization.py` to support Chinese character tokenization, so please update if
you forked it. However, we did not change the tokenization API.
To test the new models, we did modify `run_classifier.py` to add support for the
[XNLI dataset](https://github.com/facebookresearch/XNLI). This is a 15-language
version of MultiNLI where the dev/test sets have been human-translated, and the
training set has been machine-translated.
To run the fine-tuning code, please download the
[XNLI dev/test set](https://www.nyu.edu/projects/bowman/xnli/XNLI-1.0.zip) and the
[XNLI machine-translated training set](https://www.nyu.edu/projects/bowman/xnli/XNLI-MT-1.0.zip)
and then unpack both .zip files into some directory `$XNLI_DIR`.
To run fine-tuning on XNLI. The language is hard-coded into `run_classifier.py`
(Chinese by default), so please modify `XnliProcessor` if you want to run on
another language.
This is a large dataset, so this will training will take a few hours on a GPU
(or about 30 minutes on a Cloud TPU). To run an experiment quickly for
debugging, just set `num_train_epochs` to a small value like `0.1`.
```shell
export BERT_BASE_DIR=/path/to/bert/chinese_L-12_H-768_A-12 # or multilingual_L-12_H-768_A-12
export XNLI_DIR=/path/to/xnli
python run_classifier.py \
--task_name=XNLI \
--do_train=true \
--do_eval=true \
--data_dir=$XNLI_DIR \
--vocab_file=$BERT_BASE_DIR/vocab.txt \
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
--max_seq_length=128 \
--train_batch_size=32 \
--learning_rate=5e-5 \
--num_train_epochs=2.0 \
--output_dir=/tmp/xnli_output/
```
With the Chinese-only model, the results should look something like this:
```
***** Eval results *****
eval_accuracy = 0.774116
eval_loss = 0.83554
global_step = 24543
loss = 0.74603
```
## Details
### Data Source and Sampling
The languages chosen were the
[top 100 languages with the largest Wikipedias](https://meta.wikimedia.org/wiki/List_of_Wikipedias).
The entire Wikipedia dump for each language (excluding user and talk pages) was
taken as the training data for each language
However, the size of the Wikipedia for a given language varies greatly, and
therefore low-resource languages may be "under-represented" in terms of the
neural network model (under the assumption that languages are "competing" for
limited model capacity to some extent). At the same time, we also don't want
to overfit the model by performing thousands of epochs over a tiny Wikipedia
for a particular language.
To balance these two factors, we performed exponentially smoothed weighting of
the data during pre-training data creation (and WordPiece vocab creation). In
other words, let's say that the probability of a language is *P(L)*, e.g.,
*P(English) = 0.21* means that after concatenating all of the Wikipedias
together, 21% of our data is English. We exponentiate each probability by some
factor *S* and then re-normalize, and sample from that distribution. In our case
we use *S=0.7*. So, high-resource languages like English will be under-sampled,
and low-resource languages like Icelandic will be over-sampled. E.g., in the
original distribution English would be sampled 1000x more than Icelandic, but
after smoothing it's only sampled 100x more.
### Tokenization
For tokenization, we use a 110k shared WordPiece vocabulary. The word counts are
weighted the same way as the data, so low-resource languages are upweighted by
some factor. We intentionally do *not* use any marker to denote the input
language (so that zero-shot training can work).
Because Chinese (and Japanese Kanji and Korean Hanja) does not have whitespace
characters, we add spaces around every character in the
[CJK Unicode range](https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_\(Unicode_block\))
before applying WordPiece. This means that Chinese is effectively
character-tokenized. Note that the CJK Unicode block only includes
Chinese-origin characters and does *not* include Hangul Korean or
Katakana/Hiragana Japanese, which are tokenized with whitespace+WordPiece like
all other languages.
For all other languages, we apply the
[same recipe as English](https://github.com/google-research/bert#tokenization):
(a) lower casing+accent removal, (b) punctuation splitting, (c) whitespace
tokenization. We understand that accent markers have substantial meaning in some
languages, but felt that the benefits of reducing the effective vocabulary make
up for this. Generally the strong contextual models of BERT should make up for
any ambiguity introduced by stripping accent markers.
### List of Languages
The multilingual model supports the following languages. These languages were
chosen because they are the top 100 languages with the largest Wikipedias:
* Afrikaans
* Albanian
* Arabic
* Aragonese
* Armenian
* Asturian
* Azerbaijani
* Bashkir
* Basque
* Bavarian
* Belarusian
* Bengali
* Bishnupriya Manipuri
* Bosnian
* Breton
* Bulgarian
* Burmese
* Catalan
* Cebuano
* Chechen
* Chinese (Simplified)
* Chinese (Traditional)
* Chuvash
* Croatian
* Czech
* Danish
* Dutch
* English
* Estonian
* Finnish
* French
* Galician
* Georgian
* German
* Greek
* Gujarati
* Haitian
* Hebrew
* Hindi
* Hungarian
* Icelandic
* Ido
* Indonesian
* Irish
* Italian
* Japanese
* Javanese
* Kannada
* Kazakh
* Kirghiz
* Korean
* Latin
* Latvian
* Lithuanian
* Lombard
* Low Saxon
* Luxembourgish
* Macedonian
* Malagasy
* Malay
* Malayalam
* Marathi
* Minangkabau
* Nepali
* Newar
* Norwegian (Bokmal)
* Norwegian (Nynorsk)
* Occitan
* Persian (Farsi)
* Piedmontese
* Polish
* Portuguese
* Punjabi
* Romanian
* Russian
* Scots
* Serbian
* Serbo-Croatian
* Sicilian
* Slovak
* Slovenian
* South Azerbaijani
* Spanish
* Sundanese
* Swahili
* Swedish
* Tagalog
* Tajik
* Tamil
* Tatar
* Telugu
* Turkish
* Ukrainian
* Urdu
* Uzbek
* Vietnamese
* Volapük
* Waray-Waray
* Welsh
* West Frisian
* Western Punjabi
* Yoruba
The **Multilingual Cased (New)** release contains additionally **Thai** and
**Mongolian**, which were not included in the original release.
================================================
FILE: optimization.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Functions and classes related to optimization (weight updates)."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import re
import tensorflow as tf
def create_optimizer(loss, init_lr, num_train_steps, num_warmup_steps, use_tpu):
"""Creates an optimizer training op."""
global_step = tf.train.get_or_create_global_step()
learning_rate = tf.constant(value=init_lr, shape=[], dtype=tf.float32)
# Implements linear decay of the learning rate.
learning_rate = tf.train.polynomial_decay(
learning_rate,
global_step,
num_train_steps,
end_learning_rate=0.0,
power=1.0,
cycle=False)
# Implements linear warmup. I.e., if global_step < num_warmup_steps, the
# learning rate will be `global_step/num_warmup_steps * init_lr`.
if num_warmup_steps:
global_steps_int = tf.cast(global_step, tf.int32)
warmup_steps_int = tf.constant(num_warmup_steps, dtype=tf.int32)
global_steps_float = tf.cast(global_steps_int, tf.float32)
warmup_steps_float = tf.cast(warmup_steps_int, tf.float32)
warmup_percent_done = global_steps_float / warmup_steps_float
warmup_learning_rate = init_lr * warmup_percent_done
is_warmup = tf.cast(global_steps_int < warmup_steps_int, tf.float32)
learning_rate = (
(1.0 - is_warmup) * learning_rate + is_warmup * warmup_learning_rate)
# It is recommended that you use this optimizer for fine tuning, since this
# is how the model was trained (note that the Adam m/v variables are NOT
# loaded from init_checkpoint.)
optimizer = AdamWeightDecayOptimizer(
learning_rate=learning_rate,
weight_decay_rate=0.01,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-6,
exclude_from_weight_decay=["LayerNorm", "layer_norm", "bias"])
if use_tpu:
optimizer = tf.contrib.tpu.CrossShardOptimizer(optimizer)
tvars = tf.trainable_variables()
grads = tf.gradients(loss, tvars)
# This is how the model was pre-trained.
(grads, _) = tf.clip_by_global_norm(grads, clip_norm=1.0)
train_op = optimizer.apply_gradients(
zip(grads, tvars), global_step=global_step)
# Normally the global step update is done inside of `apply_gradients`.
# However, `AdamWeightDecayOptimizer` doesn't do this. But if you use
# a different optimizer, you should probably take this line out.
new_global_step = global_step + 1
train_op = tf.group(train_op, [global_step.assign(new_global_step)])
return train_op
class AdamWeightDecayOptimizer(tf.train.Optimizer):
"""A basic Adam optimizer that includes "correct" L2 weight decay."""
def __init__(self,
learning_rate,
weight_decay_rate=0.0,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-6,
exclude_from_weight_decay=None,
name="AdamWeightDecayOptimizer"):
"""Constructs a AdamWeightDecayOptimizer."""
super(AdamWeightDecayOptimizer, self).__init__(False, name)
self.learning_rate = learning_rate
self.weight_decay_rate = weight_decay_rate
self.beta_1 = beta_1
self.beta_2 = beta_2
self.epsilon = epsilon
self.exclude_from_weight_decay = exclude_from_weight_decay
def apply_gradients(self, grads_and_vars, global_step=None, name=None):
"""See base class."""
assignments = []
for (grad, param) in grads_and_vars:
if grad is None or param is None:
continue
param_name = self._get_variable_name(param.name)
m = tf.get_variable(
name=param_name + "/adam_m",
shape=param.shape.as_list(),
dtype=tf.float32,
trainable=False,
initializer=tf.zeros_initializer())
v = tf.get_variable(
name=param_name + "/adam_v",
shape=param.shape.as_list(),
dtype=tf.float32,
trainable=False,
initializer=tf.zeros_initializer())
# Standard Adam update.
next_m = (
tf.multiply(self.beta_1, m) + tf.multiply(1.0 - self.beta_1, grad))
next_v = (
tf.multiply(self.beta_2, v) + tf.multiply(1.0 - self.beta_2,
tf.square(grad)))
update = next_m / (tf.sqrt(next_v) + self.epsilon)
# Just adding the square of the weights to the loss function is *not*
# the correct way of using L2 regularization/weight decay with Adam,
# since that will interact with the m and v parameters in strange ways.
#
# Instead we want ot decay the weights in a manner that doesn't interact
# with the m/v parameters. This is equivalent to adding the square
# of the weights to the loss with plain (non-momentum) SGD.
if self._do_use_weight_decay(param_name):
update += self.weight_decay_rate * param
update_with_lr = self.learning_rate * update
next_param = param - update_with_lr
assignments.extend(
[param.assign(next_param),
m.assign(next_m),
v.assign(next_v)])
return tf.group(*assignments, name=name)
def _do_use_weight_decay(self, param_name):
"""Whether to use L2 weight decay for `param_name`."""
if not self.weight_decay_rate:
return False
if self.exclude_from_weight_decay:
for r in self.exclude_from_weight_decay:
if re.search(r, param_name) is not None:
return False
return True
def _get_variable_name(self, param_name):
"""Get the variable name from the tensor name."""
m = re.match("^(.*):\\d+$", param_name)
if m is not None:
param_name = m.group(1)
return param_name
================================================
FILE: optimization_test.py
================================================
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import optimization
import tensorflow as tf
class OptimizationTest(tf.test.TestCase):
def test_adam(self):
with self.test_session() as sess:
w = tf.get_variable(
"w",
shape=[3],
initializer=tf.constant_initializer([0.1, -0.2, -0.1]))
x = tf.constant([0.4, 0.2, -0.5])
loss = tf.reduce_mean(tf.square(x - w))
tvars = tf.trainable_variables()
grads = tf.gradients(loss, tvars)
global_step = tf.train.get_or_create_global_step()
optimizer = optimization.AdamWeightDecayOptimizer(learning_rate=0.2)
train_op = optimizer.apply_gradients(zip(grads, tvars), global_step)
init_op = tf.group(tf.global_variables_initializer(),
tf.local_variables_initializer())
sess.run(init_op)
for _ in range(100):
sess.run(train_op)
w_np = sess.run(w)
self.assertAllClose(w_np.flat, [0.4, 0.2, -0.5], rtol=1e-2, atol=1e-2)
if __name__ == "__main__":
tf.test.main()
================================================
FILE: predicting_movie_reviews_with_bert_on_tf_hub.ipynb
================================================
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "Predicting Movie Reviews with BERT on TF Hub.ipynb",
"version": "0.3.2",
"provenance": [],
"collapsed_sections": []
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"metadata": {
"id": "j0a4mTk9o1Qg",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"# Copyright 2019 Google Inc.\n",
"\n",
"# Licensed under the Apache License, Version 2.0 (the \"License\");\n",
"# you may not use this file except in compliance with the License.\n",
"# You may obtain a copy of the License at\n",
"\n",
"# http://www.apache.org/licenses/LICENSE-2.0\n",
"\n",
"# Unless required by applicable law or agreed to in writing, software\n",
"# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
"# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
"# See the License for the specific language governing permissions and\n",
"# limitations under the License."
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "dCpvgG0vwXAZ",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"#Predicting Movie Review Sentiment with BERT on TF Hub"
]
},
{
"metadata": {
"id": "xiYrZKaHwV81",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"If you’ve been following Natural Language Processing over the past year, you’ve probably heard of BERT: Bidirectional Encoder Representations from Transformers. It’s a neural network architecture designed by Google researchers that’s totally transformed what’s state-of-the-art for NLP tasks, like text classification, translation, summarization, and question answering.\n",
"\n",
"Now that BERT's been added to [TF Hub](https://www.tensorflow.org/hub) as a loadable module, it's easy(ish) to add into existing Tensorflow text pipelines. In an existing pipeline, BERT can replace text embedding layers like ELMO and GloVE. Alternatively, [finetuning](http://wiki.fast.ai/index.php/Fine_tuning) BERT can provide both an accuracy boost and faster training time in many cases.\n",
"\n",
"Here, we'll train a model to predict whether an IMDB movie review is positive or negative using BERT in Tensorflow with tf hub. Some code was adapted from [this colab notebook](https://colab.sandbox.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb). Let's get started!"
]
},
{
"metadata": {
"id": "hsZvic2YxnTz",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"from sklearn.model_selection import train_test_split\n",
"import pandas as pd\n",
"import tensorflow as tf\n",
"import tensorflow_hub as hub\n",
"from datetime import datetime"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "cp5wfXDx5SPH",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"In addition to the standard libraries we imported above, we'll need to install BERT's python package."
]
},
{
"metadata": {
"id": "jviywGyWyKsA",
"colab_type": "code",
"outputId": "166f3005-d219-404f-b201-2a0b75480360",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 51
}
},
"cell_type": "code",
"source": [
"!pip install bert-tensorflow"
],
"execution_count": 38,
"outputs": [
{
"output_type": "stream",
"text": [
"Requirement already satisfied: bert-tensorflow in /usr/local/lib/python3.6/dist-packages (1.0.1)\n",
"Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from bert-tensorflow) (1.11.0)\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "hhbGEfwgdEtw",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"import bert\n",
"from bert import run_classifier\n",
"from bert import optimization\n",
"from bert import tokenization"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "KVB3eOcjxxm1",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Below, we'll set an output directory location to store our model output and checkpoints. This can be a local directory, in which case you'd set OUTPUT_DIR to the name of the directory you'd like to create. If you're running this code in Google's hosted Colab, the directory won't persist after the Colab session ends.\n",
"\n",
"Alternatively, if you're a GCP user, you can store output in a GCP bucket. To do that, set a directory name in OUTPUT_DIR and the name of the GCP bucket in the BUCKET field.\n",
"\n",
"Set DO_DELETE to rewrite the OUTPUT_DIR if it exists. Otherwise, Tensorflow will load existing model checkpoints from that directory (if they exist)."
]
},
{
"metadata": {
"id": "US_EAnICvP7f",
"colab_type": "code",
"outputId": "7780a032-31d4-4794-e6aa-664a5d2ae7dd",
"cellView": "form",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"cell_type": "code",
"source": [
"# Set the output directory for saving model file\n",
"# Optionally, set a GCP bucket location\n",
"\n",
"OUTPUT_DIR = 'OUTPUT_DIR_NAME'#@param {type:\"string\"}\n",
"#@markdown Whether or not to clear/delete the directory and create a new one\n",
"DO_DELETE = False #@param {type:\"boolean\"}\n",
"#@markdown Set USE_BUCKET and BUCKET if you want to (optionally) store model output on GCP bucket.\n",
"USE_BUCKET = True #@param {type:\"boolean\"}\n",
"BUCKET = 'BUCKET_NAME' #@param {type:\"string\"}\n",
"\n",
"if USE_BUCKET:\n",
" OUTPUT_DIR = 'gs://{}/{}'.format(BUCKET, OUTPUT_DIR)\n",
" from google.colab import auth\n",
" auth.authenticate_user()\n",
"\n",
"if DO_DELETE:\n",
" try:\n",
" tf.gfile.DeleteRecursively(OUTPUT_DIR)\n",
" except:\n",
" # Doesn't matter if the directory didn't exist\n",
" pass\n",
"tf.gfile.MakeDirs(OUTPUT_DIR)\n",
"print('***** Model output directory: {} *****'.format(OUTPUT_DIR))\n"
],
"execution_count": 40,
"outputs": [
{
"output_type": "stream",
"text": [
"***** Model output directory: gs://bert-tfhub/aclImdb_v1 *****\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "pmFYvkylMwXn",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"#Data"
]
},
{
"metadata": {
"id": "MC_w8SRqN0fr",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"First, let's download the dataset, hosted by Stanford. The code below, which downloads, extracts, and imports the IMDB Large Movie Review Dataset, is borrowed from [this Tensorflow tutorial](https://www.tensorflow.org/hub/tutorials/text_classification_with_tf_hub)."
]
},
{
"metadata": {
"id": "fom_ff20gyy6",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"from tensorflow import keras\n",
"import os\n",
"import re\n",
"\n",
"# Load all files from a directory in a DataFrame.\n",
"def load_directory_data(directory):\n",
" data = {}\n",
" data[\"sentence\"] = []\n",
" data[\"sentiment\"] = []\n",
" for file_path in os.listdir(directory):\n",
" with tf.gfile.GFile(os.path.join(directory, file_path), \"r\") as f:\n",
" data[\"sentence\"].append(f.read())\n",
" data[\"sentiment\"].append(re.match(\"\\d+_(\\d+)\\.txt\", file_path).group(1))\n",
" return pd.DataFrame.from_dict(data)\n",
"\n",
"# Merge positive and negative examples, add a polarity column and shuffle.\n",
"def load_dataset(directory):\n",
" pos_df = load_directory_data(os.path.join(directory, \"pos\"))\n",
" neg_df = load_directory_data(os.path.join(directory, \"neg\"))\n",
" pos_df[\"polarity\"] = 1\n",
" neg_df[\"polarity\"] = 0\n",
" return pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True)\n",
"\n",
"# Download and process the dataset files.\n",
"def download_and_load_datasets(force_download=False):\n",
" dataset = tf.keras.utils.get_file(\n",
" fname=\"aclImdb.tar.gz\", \n",
" origin=\"http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\", \n",
" extract=True)\n",
" \n",
" train_df = load_dataset(os.path.join(os.path.dirname(dataset), \n",
" \"aclImdb\", \"train\"))\n",
" test_df = load_dataset(os.path.join(os.path.dirname(dataset), \n",
" \"aclImdb\", \"test\"))\n",
" \n",
" return train_df, test_df\n"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "2abfwdn-g135",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"train, test = download_and_load_datasets()"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "XA8WHJgzhIZf",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"To keep training fast, we'll take a sample of 5000 train and test examples, respectively."
]
},
{
"metadata": {
"id": "lw_F488eixTV",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"train = train.sample(5000)\n",
"test = test.sample(5000)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "prRQM8pDi8xI",
"colab_type": "code",
"outputId": "34445cb8-2be0-4379-fdbc-7794091f6049",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"cell_type": "code",
"source": [
"train.columns"
],
"execution_count": 44,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"Index(['sentence', 'sentiment', 'polarity'], dtype='object')"
]
},
"metadata": {
"tags": []
},
"execution_count": 44
}
]
},
{
"metadata": {
"id": "sfRnHSz3iSXz",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"For us, our input data is the 'sentence' column and our label is the 'polarity' column (0, 1 for negative and positive, respecitvely)"
]
},
{
"metadata": {
"id": "IuMOGwFui4it",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"DATA_COLUMN = 'sentence'\n",
"LABEL_COLUMN = 'polarity'\n",
"# label_list is the list of labels, i.e. True, False or 0, 1 or 'dog', 'cat'\n",
"label_list = [0, 1]"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "V399W0rqNJ-Z",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"#Data Preprocessing\n",
"We'll need to transform our data into a format BERT understands. This involves two steps. First, we create `InputExample`'s using the constructor provided in the BERT library.\n",
"\n",
"- `text_a` is the text we want to classify, which in this case, is the `Request` field in our Dataframe. \n",
"- `text_b` is used if we're training a model to understand the relationship between sentences (i.e. is `text_b` a translation of `text_a`? Is `text_b` an answer to the question asked by `text_a`?). This doesn't apply to our task, so we can leave `text_b` blank.\n",
"- `label` is the label for our example, i.e. True, False"
]
},
{
"metadata": {
"id": "p9gEt5SmM6i6",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"# Use the InputExample class from BERT's run_classifier code to create examples from the data\n",
"train_InputExamples = train.apply(lambda x: bert.run_classifier.InputExample(guid=None, # Globally unique ID for bookkeeping, unused in this example\n",
" text_a = x[DATA_COLUMN], \n",
" text_b = None, \n",
" label = x[LABEL_COLUMN]), axis = 1)\n",
"\n",
"test_InputExamples = test.apply(lambda x: bert.run_classifier.InputExample(guid=None, \n",
" text_a = x[DATA_COLUMN], \n",
" text_b = None, \n",
" label = x[LABEL_COLUMN]), axis = 1)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "SCZWZtKxObjh",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Next, we need to preprocess our data so that it matches the data BERT was trained on. For this, we'll need to do a couple of things (but don't worry--this is also included in the Python library):\n",
"\n",
"\n",
"1. Lowercase our text (if we're using a BERT lowercase model)\n",
"2. Tokenize it (i.e. \"sally says hi\" -> [\"sally\", \"says\", \"hi\"])\n",
"3. Break words into WordPieces (i.e. \"calling\" -> [\"call\", \"##ing\"])\n",
"4. Map our words to indexes using a vocab file that BERT provides\n",
"5. Add special \"CLS\" and \"SEP\" tokens (see the [readme](https://github.com/google-research/bert))\n",
"6. Append \"index\" and \"segment\" tokens to each input (see the [BERT paper](https://arxiv.org/pdf/1810.04805.pdf))\n",
"\n",
"Happily, we don't have to worry about most of these details.\n",
"\n",
"\n"
]
},
{
"metadata": {
"id": "qMWiDtpyQSoU",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"To start, we'll need to load a vocabulary file and lowercasing information directly from the BERT tf hub module:"
]
},
{
"metadata": {
"id": "IhJSe0QHNG7U",
"colab_type": "code",
"outputId": "20b28cc7-3cb3-4ce6-bfff-a7847ce3bbaa",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"cell_type": "code",
"source": [
"# This is a path to an uncased (all lowercase) version of BERT\n",
"BERT_MODEL_HUB = \"https://tfhub.dev/google/bert_uncased_L-12_H-768_A-12/1\"\n",
"\n",
"def create_tokenizer_from_hub_module():\n",
" \"\"\"Get the vocab file and casing info from the Hub module.\"\"\"\n",
" with tf.Graph().as_default():\n",
" bert_module = hub.Module(BERT_MODEL_HUB)\n",
" tokenization_info = bert_module(signature=\"tokenization_info\", as_dict=True)\n",
" with tf.Session() as sess:\n",
" vocab_file, do_lower_case = sess.run([tokenization_info[\"vocab_file\"],\n",
" tokenization_info[\"do_lower_case\"]])\n",
" \n",
" return bert.tokenization.FullTokenizer(\n",
" vocab_file=vocab_file, do_lower_case=do_lower_case)\n",
"\n",
"tokenizer = create_tokenizer_from_hub_module()"
],
"execution_count": 47,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Saver not created because there are no variables in the graph to restore\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "z4oFkhpZBDKm",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Great--we just learned that the BERT model we're using expects lowercase data (that's what stored in tokenization_info[\"do_lower_case\"]) and we also loaded BERT's vocab file. We also created a tokenizer, which breaks words into word pieces:"
]
},
{
"metadata": {
"id": "dsBo6RCtQmwx",
"colab_type": "code",
"outputId": "9af8c917-90ec-4fe9-897b-79dc89ca88e1",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 221
}
},
"cell_type": "code",
"source": [
"tokenizer.tokenize(\"This here's an example of using the BERT tokenizer\")"
],
"execution_count": 48,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"['this',\n",
" 'here',\n",
" \"'\",\n",
" 's',\n",
" 'an',\n",
" 'example',\n",
" 'of',\n",
" 'using',\n",
" 'the',\n",
" 'bert',\n",
" 'token',\n",
" '##izer']"
]
},
"metadata": {
"tags": []
},
"execution_count": 48
}
]
},
{
"metadata": {
"id": "0OEzfFIt6GIc",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Using our tokenizer, we'll call `run_classifier.convert_examples_to_features` on our InputExamples to convert them into features BERT understands."
]
},
{
"metadata": {
"id": "LL5W8gEGRTAf",
"colab_type": "code",
"outputId": "65001dda-155b-48fc-b5fc-1e4cabc8dfbf",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1261
}
},
"cell_type": "code",
"source": [
"# We'll set sequences to be at most 128 tokens long.\n",
"MAX_SEQ_LENGTH = 128\n",
"# Convert our train and test features to InputFeatures that BERT understands.\n",
"train_features = bert.run_classifier.convert_examples_to_features(train_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)\n",
"test_features = bert.run_classifier.convert_examples_to_features(test_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)"
],
"execution_count": 49,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Writing example 0 of 5000\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] i ' m watching this on the sci - fi channel right now . it ' s so horrible i can ' t stop watching it ! i ' m a video ##grapher and this movie makes me sad . i feel bad for anyone associated with this movie . some of the camera work is good . most is very questionable . there are a few decent actors in the flick . too bad they ' re surrounded by what must have been the director ' s relatives . that ' s the only way they could have been qualified to be in a movie ! music was a little better than the acting . if you get around to watching this i hope it [SEP]\n",
"INFO:tensorflow:input_ids: 101 1045 1005 1049 3666 2023 2006 1996 16596 1011 10882 3149 2157 2085 1012 2009 1005 1055 2061 9202 1045 2064 1005 1056 2644 3666 2009 999 1045 1005 1049 1037 2678 18657 1998 2023 3185 3084 2033 6517 1012 1045 2514 2919 2005 3087 3378 2007 2023 3185 1012 2070 1997 1996 4950 2147 2003 2204 1012 2087 2003 2200 21068 1012 2045 2024 1037 2261 11519 5889 1999 1996 17312 1012 2205 2919 2027 1005 2128 5129 2011 2054 2442 2031 2042 1996 2472 1005 1055 9064 1012 2008 1005 1055 1996 2069 2126 2027 2071 2031 2042 4591 2000 2022 1999 1037 3185 999 2189 2001 1037 2210 2488 2084 1996 3772 1012 2065 2017 2131 2105 2000 3666 2023 1045 3246 2009 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 0 (id = 0)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] i have been a fan of pushing dai ##sies since the very beginning . it is wonderful ##ly thought up , and bryan fuller has the most remarkable ideas for this show . < br / > < br / > it is unbelievable on how much tv has been needing a creative , original show like pushing dai ##sies . it is a huge relief to see a show , that is unlike the rest , where as , if you compared it to some of the newer shows , such as scrub ##s and house , you would see the similarities , and it does get ted ##ious at moments to see shows so close in identity . < br / > < br [SEP]\n",
"INFO:tensorflow:input_ids: 101 1045 2031 2042 1037 5470 1997 6183 18765 14625 2144 1996 2200 2927 1012 2009 2003 6919 2135 2245 2039 1010 1998 8527 12548 2038 1996 2087 9487 4784 2005 2023 2265 1012 1026 7987 1013 1028 1026 7987 1013 1028 2009 2003 23653 2006 2129 2172 2694 2038 2042 11303 1037 5541 1010 2434 2265 2066 6183 18765 14625 1012 2009 2003 1037 4121 4335 2000 2156 1037 2265 1010 2008 2003 4406 1996 2717 1010 2073 2004 1010 2065 2017 4102 2009 2000 2070 1997 1996 10947 3065 1010 2107 2004 18157 2015 1998 2160 1010 2017 2052 2156 1996 12319 1010 1998 2009 2515 2131 6945 6313 2012 5312 2000 2156 3065 2061 2485 1999 4767 1012 1026 7987 1013 1028 1026 7987 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 1 (id = 1)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] this movie starts out promising ##ly , with an early scene in which frank morgan advises against gary cooper ' s marriage to his daughter , anita louise . frank morgan , playing an una ##bas ##hed gold - digger , loudly complain ##s to cooper about his perceived pen ##ury at the hands of his family - including his daughter , anita louise . i am a fan of all 3 actors . frank morgan is ( to my mind ) a hollywood treasure , cooper a legend , and louise a very lovely , versatile and under - appreciated actress seldom seen in the leading role . i also have nothing against teresa wright , and while not blessed with great range , she [SEP]\n",
"INFO:tensorflow:input_ids: 101 2023 3185 4627 2041 10015 2135 1010 2007 2019 2220 3496 1999 2029 3581 5253 25453 2114 5639 6201 1005 1055 3510 2000 2010 2684 1010 12918 8227 1012 3581 5253 1010 2652 2019 14477 22083 9072 2751 1011 28661 1010 9928 17612 2015 2000 6201 2055 2010 8690 7279 13098 2012 1996 2398 1997 2010 2155 1011 2164 2010 2684 1010 12918 8227 1012 1045 2572 1037 5470 1997 2035 1017 5889 1012 3581 5253 2003 1006 2000 2026 2568 1007 1037 5365 8813 1010 6201 1037 5722 1010 1998 8227 1037 2200 8403 1010 22979 1998 2104 1011 12315 3883 15839 2464 1999 1996 2877 2535 1012 1045 2036 2031 2498 2114 12409 6119 1010 1998 2096 2025 10190 2007 2307 2846 1010 2016 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 0 (id = 0)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] i was over ##taken by the emotion . un ##for ##get ##table rendering of a wartime story which is unknown to most people . the performances were fault ##less and outstanding . [SEP]\n",
"INFO:tensorflow:input_ids: 101 1045 2001 2058 25310 2011 1996 7603 1012 4895 29278 18150 10880 14259 1997 1037 12498 2466 2029 2003 4242 2000 2087 2111 1012 1996 4616 2020 6346 3238 1998 5151 1012 102 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 1 (id = 1)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] soldier blue is a movie with pre ##tension ##s : pre ##tension ##s to be some sort of profound statement on man ' s inhuman ##ity to man , on the white man ' s exploitation of and brutality towards indigenous peoples ; a biting , un ##fl ##in ##ching and sar ##don ##ic commentary on the horrors of vietnam . well , sorry , but it fails mis ##era ##bly to be any of those things . what soldier blue actually is is per ##nic ##ious , tri ##te , badly made , dish ##ones ##t rubbish . < br / > < br / > another reviewer here hit the nail on the head in saying that it appears to be a hybrid of [SEP]\n",
"INFO:tensorflow:input_ids: 101 5268 2630 2003 1037 3185 2007 3653 29048 2015 1024 3653 29048 2015 2000 2022 2070 4066 1997 13769 4861 2006 2158 1005 1055 29582 3012 2000 2158 1010 2006 1996 2317 2158 1005 1055 14427 1997 1998 24083 2875 6284 7243 1025 1037 12344 1010 4895 10258 2378 8450 1998 18906 5280 2594 8570 2006 1996 22812 1997 5148 1012 2092 1010 3374 1010 2021 2009 11896 28616 6906 6321 2000 2022 2151 1997 2216 2477 1012 2054 5268 2630 2941 2003 2003 2566 8713 6313 1010 13012 2618 1010 6649 2081 1010 9841 21821 2102 29132 1012 1026 7987 1013 1028 1026 7987 1013 1028 2178 12027 2182 2718 1996 13774 2006 1996 2132 1999 3038 2008 2009 3544 2000 2022 1037 8893 1997 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 0 (id = 0)\n",
"INFO:tensorflow:Writing example 0 of 5000\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] i just watched this today on tv . it was on abc ' s sunday afternoon movie . < br / > < br / > this wasn ' t a very good movie , but for a low budget independent film like this , it was okay . there is some suspense in it , but there are so many bad qualities that really bring the movie down . the script is pretty lame , and the plot elements aren ' t very realistic , such as the way a 911 operator would laugh and hang up when someone is reporting a murder . i don ' t know what the writer was thinking when they came up with that idea , but it isn [SEP]\n",
"INFO:tensorflow:input_ids: 101 1045 2074 3427 2023 2651 2006 2694 1012 2009 2001 2006 5925 1005 1055 4465 5027 3185 1012 1026 7987 1013 1028 1026 7987 1013 1028 2023 2347 1005 1056 1037 2200 2204 3185 1010 2021 2005 1037 2659 5166 2981 2143 2066 2023 1010 2009 2001 3100 1012 2045 2003 2070 23873 1999 2009 1010 2021 2045 2024 2061 2116 2919 11647 2008 2428 3288 1996 3185 2091 1012 1996 5896 2003 3492 20342 1010 1998 1996 5436 3787 4995 1005 1056 2200 12689 1010 2107 2004 1996 2126 1037 19989 6872 2052 4756 1998 6865 2039 2043 2619 2003 7316 1037 4028 1012 1045 2123 1005 1056 2113 2054 1996 3213 2001 3241 2043 2027 2234 2039 2007 2008 2801 1010 2021 2009 3475 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 0 (id = 0)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] from hardly alien sounding lasers , to an elementary school style shuttle crash , \" night ##be ##ast \" is better classified as a far ##cic ##al mix of fake blood and bare chest . the almost pornographic style of the film seems to be a failed attempt to recover from a lack of co ##hesive or effective story . the acting however is not nearly as beast ##ly , many of the young , aspiring , actors ad ##mir ##ably showcase a hidden talent . particularly don lei ##fer ##t and jamie ze ##mare ##l , who shed a well needed sha ##rd of light on this otherwise terrible film . night ##be ##ast would have never shown up on set had he known the [SEP]\n",
"INFO:tensorflow:input_ids: 101 2013 6684 7344 9391 23965 1010 2000 2019 4732 2082 2806 10382 5823 1010 1000 2305 4783 14083 1000 2003 2488 6219 2004 1037 2521 19053 2389 4666 1997 8275 2668 1998 6436 3108 1012 1996 2471 26932 2806 1997 1996 2143 3849 2000 2022 1037 3478 3535 2000 8980 2013 1037 3768 1997 2522 21579 2030 4621 2466 1012 1996 3772 2174 2003 2025 3053 2004 6841 2135 1010 2116 1997 1996 2402 1010 22344 1010 5889 4748 14503 8231 13398 1037 5023 5848 1012 3391 2123 26947 7512 2102 1998 6175 27838 24376 2140 1010 2040 8328 1037 2092 2734 21146 4103 1997 2422 2006 2023 4728 6659 2143 1012 2305 4783 14083 2052 2031 2196 3491 2039 2006 2275 2018 2002 2124 1996 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 0 (id = 0)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] here we have the in ##imi ##table charlie chaplin for ##sa ##king his slap ##stick past to tackle the serious subject of anti - semi ##tism , and into ##ler ##ance in general . he portrays two characters - the sweet , innocent jewish barber - a war veteran , and the ravi ##ng and ruthless dictator , aden ##oid h ##yn ##kel . the jewish ghetto in this country is not safe for long , due to the w ##him ##s of h ##yn ##kel and his armed thugs , who routinely rough up its residents , or leave them alone , dependent upon his mood that day or week . the barber is among them , but is befriended by his former commanding officer [SEP]\n",
"INFO:tensorflow:input_ids: 101 2182 2057 2031 1996 1999 27605 10880 4918 23331 2005 3736 6834 2010 14308 21354 2627 2000 11147 1996 3809 3395 1997 3424 1011 4100 17456 1010 1998 2046 3917 6651 1999 2236 1012 2002 17509 2048 3494 1011 1996 4086 1010 7036 3644 13362 1011 1037 2162 8003 1010 1998 1996 16806 3070 1998 18101 21237 1010 16298 9314 1044 6038 11705 1012 1996 3644 17276 1999 2023 2406 2003 2025 3647 2005 2146 1010 2349 2000 1996 1059 14341 2015 1997 1044 6038 11705 1998 2010 4273 24106 1010 2040 19974 5931 2039 2049 3901 1010 2030 2681 2068 2894 1010 7790 2588 2010 6888 2008 2154 2030 2733 1012 1996 13362 2003 2426 2068 1010 2021 2003 23386 2011 2010 2280 7991 2961 102\n",
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
"INFO:tensorflow:segment_ids: 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 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\n",
"INFO:tensorflow:label: 1 (id = 1)\n",
"INFO:tensorflow:*** Example ***\n",
"INFO:tensorflow:guid: None\n",
"INFO:tensorflow:tokens: [CLS] i really hated this movie and it ' s the first movie written by stephen king that i didn ' t finish . i was truly disappointed , it was the worst crap i ' ve ever seen . what were you thinking making three hours out of it ? it may have a quite good story , but actors ? no . suspense ? no . romance ? no . horror ? no . it didn ' t have anything . < br / > < br
gitextract_6m8hx9o7/ ├── .gitignore ├── CONTRIBUTING.md ├── LICENSE ├── README.md ├── __init__.py ├── create_pretraining_data.py ├── extract_features.py ├── modeling.py ├── modeling_test.py ├── multilingual.md ├── optimization.py ├── optimization_test.py ├── predicting_movie_reviews_with_bert_on_tf_hub.ipynb ├── requirements.txt ├── run_classifier.py ├── run_classifier_with_tfhub.py ├── run_pretraining.py ├── run_squad.py ├── sample_text.txt ├── tokenization.py └── tokenization_test.py
SYMBOL INDEX (185 symbols across 12 files)
FILE: create_pretraining_data.py
class TrainingInstance (line 68) | class TrainingInstance(object):
method __init__ (line 71) | def __init__(self, tokens, segment_ids, masked_lm_positions, masked_lm...
method __str__ (line 79) | def __str__(self):
method __repr__ (line 92) | def __repr__(self):
function write_instance_to_example_files (line 96) | def write_instance_to_example_files(instances, tokenizer, max_seq_length,
function create_int_feature (line 169) | def create_int_feature(values):
function create_float_feature (line 174) | def create_float_feature(values):
function create_training_instances (line 179) | def create_training_instances(input_files, tokenizer, max_seq_length,
function create_instances_from_document (line 223) | def create_instances_from_document(
function create_masked_lm_predictions (line 342) | def create_masked_lm_predictions(tokens, masked_lm_prob,
function truncate_seq_pair (line 418) | def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng):
function main (line 436) | def main(_):
FILE: extract_features.py
class InputExample (line 81) | class InputExample(object):
method __init__ (line 83) | def __init__(self, unique_id, text_a, text_b):
class InputFeatures (line 89) | class InputFeatures(object):
method __init__ (line 92) | def __init__(self, unique_id, tokens, input_ids, input_mask, input_typ...
function input_fn_builder (line 100) | def input_fn_builder(features, seq_length):
function model_fn_builder (line 148) | def model_fn_builder(bert_config, init_checkpoint, layer_indexes, use_tpu,
function convert_examples_to_features (line 210) | def convert_examples_to_features(examples, seq_length, tokenizer):
function _truncate_seq_pair (line 302) | def _truncate_seq_pair(tokens_a, tokens_b, max_length):
function read_examples (line 319) | def read_examples(input_file):
function main (line 343) | def main(_):
FILE: modeling.py
class BertConfig (line 31) | class BertConfig(object):
method __init__ (line 34) | def __init__(self,
method from_dict (line 83) | def from_dict(cls, json_object):
method from_json_file (line 91) | def from_json_file(cls, json_file):
method to_dict (line 97) | def to_dict(self):
method to_json_string (line 102) | def to_json_string(self):
class BertModel (line 107) | class BertModel(object):
method __init__ (line 131) | def __init__(self,
method get_pooled_output (line 234) | def get_pooled_output(self):
method get_sequence_output (line 237) | def get_sequence_output(self):
method get_all_encoder_layers (line 246) | def get_all_encoder_layers(self):
method get_embedding_output (line 249) | def get_embedding_output(self):
method get_embedding_table (line 260) | def get_embedding_table(self):
function gelu (line 264) | def gelu(x):
function get_activation (line 280) | def get_activation(activation_string):
function get_assignment_map_from_checkpoint (line 317) | def get_assignment_map_from_checkpoint(tvars, init_checkpoint):
function dropout (line 344) | def dropout(input_tensor, dropout_prob):
function layer_norm (line 362) | def layer_norm(input_tensor, name=None):
function layer_norm_and_dropout (line 368) | def layer_norm_and_dropout(input_tensor, dropout_prob, name=None):
function create_initializer (line 375) | def create_initializer(initializer_range=0.02):
function embedding_lookup (line 380) | def embedding_lookup(input_ids,
function embedding_postprocessor (line 428) | def embedding_postprocessor(input_tensor,
function create_attention_mask_from_input_mask (line 524) | def create_attention_mask_from_input_mask(from_tensor, to_mask):
function attention_layer (line 558) | def attention_layer(from_tensor,
function transformer_model (line 754) | def transformer_model(input_tensor,
function get_shape_list (line 895) | def get_shape_list(tensor, expected_rank=None, name=None):
function reshape_to_matrix (line 932) | def reshape_to_matrix(input_tensor):
function reshape_from_matrix (line 946) | def reshape_from_matrix(output_tensor, orig_shape_list):
function assert_rank (line 959) | def assert_rank(tensor, expected_rank, name=None):
FILE: modeling_test.py
class BertModelTest (line 29) | class BertModelTest(tf.test.TestCase):
class BertModelTester (line 31) | class BertModelTester(object):
method __init__ (line 33) | def __init__(self,
method create_model (line 71) | def create_model(self):
method check_output (line 114) | def check_output(self, result):
method test_default (line 126) | def test_default(self):
method test_config_to_json_string (line 129) | def test_config_to_json_string(self):
method run_tester (line 135) | def run_tester(self, tester):
method ids_tensor (line 147) | def ids_tensor(cls, shape, vocab_size, rng=None, name=None):
method assert_all_tensors_reachable (line 162) | def assert_all_tensors_reachable(self, sess, outputs):
method get_unreachable_ops (line 194) | def get_unreachable_ops(cls, graph, outputs):
method flatten_recursive (line 257) | def flatten_recursive(cls, item):
FILE: optimization.py
function create_optimizer (line 25) | def create_optimizer(loss, init_lr, num_train_steps, num_warmup_steps, u...
class AdamWeightDecayOptimizer (line 87) | class AdamWeightDecayOptimizer(tf.train.Optimizer):
method __init__ (line 90) | def __init__(self,
method apply_gradients (line 108) | def apply_gradients(self, grads_and_vars, global_step=None, name=None):
method _do_use_weight_decay (line 159) | def _do_use_weight_decay(self, param_name):
method _get_variable_name (line 169) | def _get_variable_name(self, param_name):
FILE: optimization_test.py
class OptimizationTest (line 23) | class OptimizationTest(tf.test.TestCase):
method test_adam (line 25) | def test_adam(self):
FILE: run_classifier.py
class InputExample (line 127) | class InputExample(object):
method __init__ (line 130) | def __init__(self, guid, text_a, text_b=None, label=None):
class PaddingInputExample (line 148) | class PaddingInputExample(object):
class InputFeatures (line 161) | class InputFeatures(object):
method __init__ (line 164) | def __init__(self,
class DataProcessor (line 177) | class DataProcessor(object):
method get_train_examples (line 180) | def get_train_examples(self, data_dir):
method get_dev_examples (line 184) | def get_dev_examples(self, data_dir):
method get_test_examples (line 188) | def get_test_examples(self, data_dir):
method get_labels (line 192) | def get_labels(self):
method _read_tsv (line 197) | def _read_tsv(cls, input_file, quotechar=None):
class XnliProcessor (line 207) | class XnliProcessor(DataProcessor):
method __init__ (line 210) | def __init__(self):
method get_train_examples (line 213) | def get_train_examples(self, data_dir):
method get_dev_examples (line 232) | def get_dev_examples(self, data_dir):
method get_labels (line 250) | def get_labels(self):
class MnliProcessor (line 255) | class MnliProcessor(DataProcessor):
method get_train_examples (line 258) | def get_train_examples(self, data_dir):
method get_dev_examples (line 263) | def get_dev_examples(self, data_dir):
method get_test_examples (line 269) | def get_test_examples(self, data_dir):
method get_labels (line 274) | def get_labels(self):
method _create_examples (line 278) | def _create_examples(self, lines, set_type):
class MrpcProcessor (line 296) | class MrpcProcessor(DataProcessor):
method get_train_examples (line 299) | def get_train_examples(self, data_dir):
method get_dev_examples (line 304) | def get_dev_examples(self, data_dir):
method get_test_examples (line 309) | def get_test_examples(self, data_dir):
method get_labels (line 314) | def get_labels(self):
method _create_examples (line 318) | def _create_examples(self, lines, set_type):
class ColaProcessor (line 336) | class ColaProcessor(DataProcessor):
method get_train_examples (line 339) | def get_train_examples(self, data_dir):
method get_dev_examples (line 344) | def get_dev_examples(self, data_dir):
method get_test_examples (line 349) | def get_test_examples(self, data_dir):
method get_labels (line 354) | def get_labels(self):
method _create_examples (line 358) | def _create_examples(self, lines, set_type):
function convert_single_example (line 377) | def convert_single_example(ex_index, example, label_list, max_seq_length,
function file_based_convert_examples_to_features (line 479) | def file_based_convert_examples_to_features(
function file_based_input_fn_builder (line 509) | def file_based_input_fn_builder(input_file, seq_length, is_training,
function _truncate_seq_pair (line 557) | def _truncate_seq_pair(tokens_a, tokens_b, max_length):
function create_model (line 574) | def create_model(bert_config, is_training, input_ids, input_mask, segmen...
function model_fn_builder (line 619) | def model_fn_builder(bert_config, num_labels, init_checkpoint, learning_...
function input_fn_builder (line 713) | def input_fn_builder(features, seq_length, is_training, drop_remainder):
function convert_examples_to_features (line 767) | def convert_examples_to_features(examples, label_list, max_seq_length,
function main (line 783) | def main(_):
FILE: run_classifier_with_tfhub.py
function create_model (line 37) | def create_model(is_training, input_ids, input_mask, segment_ids, labels,
function model_fn_builder (line 87) | def model_fn_builder(num_labels, learning_rate, num_train_steps,
function create_tokenizer_from_hub_module (line 146) | def create_tokenizer_from_hub_module(bert_hub_module_handle):
function main (line 158) | def main(_):
FILE: run_pretraining.py
function model_fn_builder (line 109) | def model_fn_builder(bert_config, init_checkpoint, learning_rate,
function get_masked_lm_output (line 240) | def get_masked_lm_output(bert_config, input_tensor, output_weights, posi...
function get_next_sentence_output (line 285) | def get_next_sentence_output(bert_config, input_tensor, labels):
function gather_indexes (line 308) | def gather_indexes(sequence_tensor, positions):
function input_fn_builder (line 324) | def input_fn_builder(input_files,
function _decode_record (line 391) | def _decode_record(record, name_to_features):
function main (line 406) | def main(_):
FILE: run_squad.py
class SquadExample (line 157) | class SquadExample(object):
method __init__ (line 163) | def __init__(self,
method __str__ (line 179) | def __str__(self):
method __repr__ (line 182) | def __repr__(self):
class InputFeatures (line 197) | class InputFeatures(object):
method __init__ (line 200) | def __init__(self,
function read_squad_examples (line 227) | def read_squad_examples(input_file, is_training):
function convert_examples_to_features (line 309) | def convert_examples_to_features(examples, tokenizer, max_seq_length,
function _improve_answer_span (line 476) | def _improve_answer_span(doc_tokens, input_start, input_end, tokenizer,
function _check_is_max_context (line 513) | def _check_is_max_context(doc_spans, cur_span_index, position):
function create_model (line 550) | def create_model(bert_config, is_training, input_ids, input_mask, segmen...
function model_fn_builder (line 590) | def model_fn_builder(bert_config, init_checkpoint, learning_rate,
function input_fn_builder (line 687) | def input_fn_builder(input_file, seq_length, is_training, drop_remainder):
function write_predictions (line 741) | def write_predictions(all_examples, all_features, all_results, n_best_size,
function get_final_text (line 927) | def get_final_text(pred_text, orig_text, do_lower_case):
function _get_best_indexes (line 1023) | def _get_best_indexes(logits, n_best_size):
function _compute_softmax (line 1035) | def _compute_softmax(scores):
class FeatureWriter (line 1058) | class FeatureWriter(object):
method __init__ (line 1061) | def __init__(self, filename, is_training):
method process_feature (line 1067) | def process_feature(self, feature):
method close (line 1093) | def close(self):
function validate_flags_or_throw (line 1097) | def validate_flags_or_throw(bert_config):
function main (line 1126) | def main(_):
FILE: tokenization.py
function validate_case_matches_checkpoint (line 28) | def validate_case_matches_checkpoint(do_lower_case, init_checkpoint):
function convert_to_unicode (line 78) | def convert_to_unicode(text):
function printable_text (line 98) | def printable_text(text):
function load_vocab (line 121) | def load_vocab(vocab_file):
function convert_by_vocab (line 136) | def convert_by_vocab(vocab, items):
function convert_tokens_to_ids (line 144) | def convert_tokens_to_ids(vocab, tokens):
function convert_ids_to_tokens (line 148) | def convert_ids_to_tokens(inv_vocab, ids):
function whitespace_tokenize (line 152) | def whitespace_tokenize(text):
class FullTokenizer (line 161) | class FullTokenizer(object):
method __init__ (line 164) | def __init__(self, vocab_file, do_lower_case=True):
method tokenize (line 170) | def tokenize(self, text):
method convert_tokens_to_ids (line 178) | def convert_tokens_to_ids(self, tokens):
method convert_ids_to_tokens (line 181) | def convert_ids_to_tokens(self, ids):
class BasicTokenizer (line 185) | class BasicTokenizer(object):
method __init__ (line 188) | def __init__(self, do_lower_case=True):
method tokenize (line 196) | def tokenize(self, text):
method _run_strip_accents (line 220) | def _run_strip_accents(self, text):
method _run_split_on_punc (line 231) | def _run_split_on_punc(self, text):
method _tokenize_chinese_chars (line 251) | def _tokenize_chinese_chars(self, text):
method _is_chinese_char (line 264) | def _is_chinese_char(self, cp):
method _clean_text (line 286) | def _clean_text(self, text):
class WordpieceTokenizer (line 300) | class WordpieceTokenizer(object):
method __init__ (line 303) | def __init__(self, vocab, unk_token="[UNK]", max_input_chars_per_word=...
method tokenize (line 308) | def tokenize(self, text):
function _is_whitespace (line 362) | def _is_whitespace(char):
function _is_control (line 374) | def _is_control(char):
function _is_punctuation (line 386) | def _is_punctuation(char):
FILE: tokenization_test.py
class TokenizationTest (line 26) | class TokenizationTest(tf.test.TestCase):
method test_full_tokenizer (line 28) | def test_full_tokenizer(self):
method test_chinese (line 51) | def test_chinese(self):
method test_basic_tokenizer_lower (line 58) | def test_basic_tokenizer_lower(self):
method test_basic_tokenizer_no_lower (line 66) | def test_basic_tokenizer_no_lower(self):
method test_wordpiece_tokenizer (line 73) | def test_wordpiece_tokenizer(self):
method test_convert_tokens_to_ids (line 93) | def test_convert_tokens_to_ids(self):
method test_is_whitespace (line 107) | def test_is_whitespace(self):
method test_is_control (line 117) | def test_is_control(self):
method test_is_punctuation (line 126) | def test_is_punctuation(self):
Condensed preview — 21 files, each showing path, character count, and a content snippet. Download the .json file or copy for the full structured content (378K chars).
[
{
"path": ".gitignore",
"chars": 1361,
"preview": "# Initially taken from Github's Python gitignore file\n\n# Byte-compiled / optimized / DLL files\n__pycache__/\n*.py[cod]\n*$"
},
{
"path": "CONTRIBUTING.md",
"chars": 1323,
"preview": "# How to Contribute\n\nBERT needs to maintain permanent compatibility with the pre-trained model files,\nso we do not plan "
},
{
"path": "LICENSE",
"chars": 11358,
"preview": "\n Apache License\n Version 2.0, January 2004\n "
},
{
"path": "README.md",
"chars": 50505,
"preview": "# BERT\n\n**\\*\\*\\*\\*\\* New March 11th, 2020: Smaller BERT Models \\*\\*\\*\\*\\***\n\nThis is a release of 24 smaller BERT models"
},
{
"path": "__init__.py",
"chars": 616,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "create_pretraining_data.py",
"chars": 16475,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "extract_features.py",
"chars": 13898,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "modeling.py",
"chars": 37922,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "modeling_test.py",
"chars": 9191,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "multilingual.md",
"chars": 11241,
"preview": "## Models\n\nThere are two multilingual models currently available. We do not plan to release\nmore single-language models,"
},
{
"path": "optimization.py",
"chars": 6258,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "optimization_test.py",
"chars": 1721,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "predicting_movie_reviews_with_bert_on_tf_hub.ipynb",
"chars": 66478,
"preview": "{\n \"nbformat\": 4,\n \"nbformat_minor\": 0,\n \"metadata\": {\n \"colab\": {\n \"name\": \"Predicting Movie Reviews with BE"
},
{
"path": "requirements.txt",
"chars": 110,
"preview": "tensorflow >= 1.11.0 # CPU Version of TensorFlow.\n# tensorflow-gpu >= 1.11.0 # GPU version of TensorFlow.\n"
},
{
"path": "run_classifier.py",
"chars": 34783,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "run_classifier_with_tfhub.py",
"chars": 11426,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "run_pretraining.py",
"chars": 18667,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "run_squad.py",
"chars": 46532,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "sample_text.txt",
"chars": 4364,
"preview": "This text is included to make sure Unicode is handled properly: 力加勝北区ᴵᴺᵀᵃছজটডণত\nText should be one-sentence-per-line, wi"
},
{
"path": "tokenization.py",
"chars": 12257,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
},
{
"path": "tokenization_test.py",
"chars": 4589,
"preview": "# coding=utf-8\n# Copyright 2018 The Google AI Language Team Authors.\n#\n# Licensed under the Apache License, Version 2.0 "
}
]
About this extraction
This page contains the full source code of the google-research/bert GitHub repository, extracted and formatted as plain text for AI agents and large language models (LLMs). The extraction includes 21 files (352.6 KB), approximately 97.2k tokens, and a symbol index with 185 extracted functions, classes, methods, constants, and types. Use this with OpenClaw, Claude, ChatGPT, Cursor, Windsurf, or any other AI tool that accepts text input. You can copy the full output to your clipboard or download it as a .txt file.
Extracted by GitExtract — free GitHub repo to text converter for AI. Built by Nikandr Surkov.