Lingvo

What is it?

Lingvo is a framework for building neural networks in Tensorflow, particularly sequence models.

A list of publications using Lingvo can be found here.

Quick start

Docker

The easiest way to get started is to use the provided Docker script. If instead you want to install it directly on your machine, skip to the section below.

First, install docker. Then, the following commands should give you a working shell with Lingvo installed.

LINGVO_DIR="/tmp/lingvo"  # (change to the cloned lingvo directory, e.g. "$HOME/lingvo")
LINGVO_DEVICE="gpu"  # (Leave empty to build and run CPU only docker)
sudo docker build --tag tensorflow:lingvo $(test "$LINGVO_DEVICE" = "gpu" && echo "--build-arg base_image=nvidia/cuda:10.0-cudnn7-runtime-ubuntu16.04") - < ${LINGVO_DIR}/docker/dev.dockerfile
sudo docker run --rm $(test "$LINGVO_DEVICE" = "gpu" && echo "--runtime=nvidia") -it -v ${LINGVO_DIR}:/tmp/lingvo -v ${HOME}/.gitconfig:/home/${USER}/.gitconfig:ro -p 6006:6006 -p 8888:8888 --name lingvo tensorflow:lingvo bash
bazel test -c opt //lingvo:trainer_test //lingvo:models_test

Installing directly

This is an alternative to using Docker as described in the section above.

The prerequisites are:

a TensorFlow installation (for now tf-nightly is required),
a C++ compiler (only g++ 4.8 is officially supported), and
the bazel build system.

Refer to docker/dev.dockerfile for more specific details.

Running the MNIST image model

Preparing the input data

mkdir -p /tmp/mnist
bazel run -c opt //lingvo/tools:keras2ckpt -- --dataset=mnist --out=/tmp/mnist/mnist

You will get the following files in /tmp/mnist:

mnist.data-00000-of-00001: 53MB.
mnist.index: 241 bytes.

Running the model

To run the trainer in single-machine mode, use

bazel build -c opt //lingvo:trainer
bazel-bin/lingvo/trainer --run_locally=cpu --mode=sync --model=image.mnist.LeNet5 --logdir=/tmp/mnist/log --logtostderr

After a few seconds, the training accuracy should reach 85% at step 100, as seen in the following line.

INFO:tensorflow:step:   100 accuracy:0.85546875 log_pplx:0.46025506 loss:0.46025506 num_preds:256 num_samples_in_batch:256

The artifacts will be produced in /tmp/mnist/log/control:

params.txt: hyper-parameters.
model_analysis.txt: model sizes for each layer.
train.pbtxt: the training tf.GraphDef.
events.*: a tensorboard events file.

In the /tmp/mnist/log/train directory, one will obtain:

ckpt-*: the checkpoint files.
checkpoint: a text file containing information about the checkpoint files.

Running the machine translation model

To run a more elaborate model, you'll need a cluster with GPUs. Please refer to lingvo/tasks/mt/README.md for more information.

Current models

Automatic Speech Recogition

Image

image.mnist.LeNet5³

Language Modelling

lm.one_billion_wds.WordLevelOneBwdsSimpleSampledSoftmax⁴

Machine Translation

[1]: Listen, Attend and Spell. William Chan, Navdeep Jaitly, Quoc V. Le, and Oriol Vinyals. ICASSP 2016.

[2]: End-to-end Continuous Speech Recognition using Attention-based Recurrent NN: First Results. Jan Chorowski, Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. arXiv 2014.

[3]: Gradient-based learning applied to document recognition. Yann LeCun, Leon Bottou, Yoshua Bengio, and Patrick Haffner. IEEE 1998.

[4]: Exploring the Limits of Language Modeling. Rafal Jozefowicz, Oriol Vinyals, Mike Schuster, Noam Shazeer, and Yonghui Wu. arXiv, 2016.

[5]: The Best of Both Worlds: Combining Recent Advances in Neural Machine Translation. Mia X. Chen, Orhan Firat, Ankur Bapna, Melvin Johnson, Wolfgang Macherey, George Foster, Llion Jones, Mike Schuster, Noam Shazeer, Niki Parmar, Ashish Vaswani, Jakob Uszkoreit, Lukasz Kaiser, Zhifeng Chen, Yonghui Wu, and Macduff Hughes. ACL 2018.

References

API Docs

Please cite this paper when referencing Lingvo.

@misc{shen2019lingvo,
    title={Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling},
    author={Jonathan Shen and Patrick Nguyen and Yonghui Wu and Zhifeng Chen and others},
    year={2019},
    eprint={1902.08295},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

newhousewhite/lingvo