This is a repository with the code for the ACL 2019 paper

Overview

The Story of Heads

This is the official repo for the following papers:

In this README, we discuss the ACL 2019 heads paper. Read the official blog post for the details!

For the contributions paper, go to the source_target_contributions folder.

Bibtex

@inproceedings{voita-etal-2019-analyzing,
    title = "Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned",
    author = "Voita, Elena  and
      Talbot, David  and
      Moiseev, Fedor  and
      Sennrich, Rico  and
      Titov, Ivan",
    booktitle = "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2019",
    address = "Florence, Italy",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/P19-1580",
    pages = "5797--5808",
}

Table of Contents

Introduction

In the paper, we:

  • evaluate the importance of attention heads in Transformer,

  • identify functions of the most important encoder heads,

  • prune the vast majority of attention heads in Transformer without seriously affecting quality using a method based on stochastic gates and a differentiable relaxation of the L0 penalty,

  • show which types of model attention are most sensitive to the number of attention heads and on which layers.

In this repo, we provide code and describe steps needed to reproduce our experiments with the L0 head pruning.

Pruning Attention Heads

In the standard Transformer, results of different attention heads in a layer are concatenated:

MultiHead(Q, K, V ) = Concat(head_i)W^O.

We modify the original Transformer architecture by multiplying the representation computed by each head_i by a scalar gate g_i:

MultiHead(Q, K, V ) = Concat(g_i * head_i)W^O.

Unlike usual gates, g_i are parameters specific to heads and are independent of the input (i.e. the sentence). As we would like to disable less important heads completely, we would ideally apply L0 regularization to the scalars g_i. The L0 norm equals the number of non-zero components and would push the model to switch off less important heads.

Unfortunately, the L0 norm is nondifferentiable and so cannot be directly incorporated as a regularization term in the objective function. Instead, we use a stochastic relaxation. Each gate g_i is a random variable drawn independently from a head-specific Hard Concrete distribution. The distributions have non-zero probability mass at 0 and 1; look at the illustration.

concrete_gif

We use the sum of the probabilities of heads being non-zero (L_C) as a stochastic relaxation of the non-differentiable L0 norm. The resulting training objective is:

L = L_xent + λ * L_C.

When applying the regularizer, we start from the converged model trained without the L_C penalty (i.e. the parameters are initialized with the parameters of the converged model) and then add the gates and continue training the full objective. By varying the coefficient λ in the optimized objective, we obtain models with different numbers of retained heads. Below is shown how the probabilities of encoder heads being completely closed (P(g_i)=0) change in training for different values of λ (pruning starts from a converged model). White color denotes P(g_i=0) = 1, which means that a head is completely removed from the model.

enc_head_gif

(Gif is for the model trained on EN-RU WMT. For other datasets, values of λ can be different.)

We observe that the model converges to solutions where gates are either almost completely closed or completely open. This means that at test time we can treat the model as a standard Transformer and use only a subset of heads.


Experiments

Requirements

Operating System: This implementation works on the most popular Linux distributions (tested on Ubuntu 14, 16). It will also likely to work on Mac OS. For other operating systems we recommend using Docker.

Hardware: The model can be trained on one or several GPUs. Training on CPU is also supported.

OpenMPI(optional): To train on several GPUs, you have to install OpenMPI. The code was tested on OpenMPI 3.1.2(download). See build instructions here.

Python: The code works with Python 3.5 and 3.6; we recommend using anaconda. Install the rest of python packages with pip install -r requirements.txt. If you haven't build OpenMPI, remove horovod from the list of requirements.

Data preprocessing

The model training config requires the data to be preprocessed, i.e. tokenized and bpeized.

Tokenization

Here is an example of how to tokenize (and lowercase) you data:

text_lines.en.tok ">
cat text_lines.en | moses-tokenizer en | python3 -c "import sys; print(sys.stdin.read().lower())" > text_lines.en.tok

For the OpenSubtitles18 dataset, you do not need this step since the data is already tokenized (you can just lowercase it).

BPE-ization

Learn BPE rules:

subword-nmt learn-bpe -s 32000 < text_lines.en.tok > bpe_rules.en

Apply BPE rules to your data:

/path_to_this_repo/lib/tools/apply_bpe.py  --bpe_rules ./bpe_rules.en  < text_lines.en.tok > text_lines.en.bpeized

Model training

In the scripts folder you can find files train_baseline.sh, train_concrete_heads.sh and train_fixed_alive_heads.sh with configs for training baseline, model with heads pruning using relaxation of the L0 penalty, and model with a fixed configuration of open and closed heads.

To launch an experiment, do the following (example is for the heads pruning experiment):

mkdir exp_dir_name && cd exp_dir_name
cp the-story-of-heads_dir/scripts/train_concrete_heads.sh .
bash train_concrete_heads.sh

After that, checkpoints will be in the exp_dir_name/build/checkpoint directory, summary for tensorboard - in exp_dir_name/build/summary, translations of dev set for checkpoints (if specified; see below) in exp_dir_name/build/translations.


Notebooks: how to use a model

In the notebooks folder you can find notebooks showing how to deal with your trained model. From a notebook name it's content has to be clear, but I'll write this just in case.

1_Load_model_and_translate - how to load model and translate sentences;

2_Look_at_attention_maps - how to draw attention maps for encoder heads;

3_Look_which_heads_are_dead - if you are pruning heads, you might want to know which ended up dead; this notebook shows you how to do this.


Training config tour

Each training script has a thorough description of the parameters and explanation of the things you need to change for your experiment. Here we'll provide a tour of the config files and explain the parameters once again.

Data

First, you need to specify your directory with the the-story-of-heads repo, data directory and train/dev file names.

REPO_DIR="../" # insert the dir to the the-story-of-heads repo
DATA_DIR="../" # insert your datadir

NMT="${REPO_DIR}/scripts/nmt.py"

# path to preprocessed data (tokenized, bpe-ized)
train_src="${DATA_DIR}/train.src"
train_dst="${DATA_DIR}/train.dst"
dev_src="${DATA_DIR}/dev.src"
dev_dst="${DATA_DIR}/dev.dst"

After that, in the config you'll see the code for creating vocabularies from your data and shuffling the data.


Model

params=(
...
--model lib.task.seq2seq.models.transformer_head_gates.Model
...)

This is the Transformer model with extra options for attention head gates: stochastic, fixed or no extra parameters for the baseline. Model hyperparameters are split into groups:

  • main model hp,
  • minor model hp (probably you do not want to change them)
  • regularization and label smoothing
  • inference params (beam search with a beam of 4)
  • head gates parameters (for the baseline, nothing is here)

For the baseline, the parameters are as follows:

hp = {
     "num_layers": 6,
     "num_heads": 8,
     "ff_size": 2048,
     "ffn_type": "conv_relu",
     "hid_size": 512,
     "emb_size": 512,
     "res_steps": "nlda", 
    
     "rescale_emb": True,
     "inp_emb_bias": True,
     "normalize_out": True,
     "share_emb": False,
     "replace": 0,
    
     "relu_dropout": 0.1,
     "res_dropout": 0.1,
     "attn_dropout": 0.1,
     "label_smoothing": 0.1,
    
     "translator": "ingraph",
     "beam_size": 4,
     "beam_spread": 3,
     "len_alpha": 0.6,
     "attn_beta": 0,
    }

This set of parameters corresponds to Transformer-base (Vaswani et al., 2017).

To train the model with heads pruning, you need to specify the types of attention heads you want to prune. For encoder self-attention heads only,

    "concrete_heads": {"enc-self"},

and for all attention types, it's

    "concrete_heads": {"enc-self", "dec-self", "dec-enc"},

For fixed head configuration, specify gate values for each head:

     "alive_heads": {"enc-self": [[1,0,1,0,1,0,1,0],
                                  [1,1,1,1,1,1,1,1],
                                  [0,0,0,0,0,0,0,0],
                                  [1,1,1,0,0,1,0,0],
                                  [0,0,0,0,1,1,1,1],
                                  [0,0,1,1,0,0,1,1]],
                    },

In this case, only encoder self-attention heads will be masked. For all attention types, specify all gates:

     "alive_heads": {"enc-self": [[1,0,1,0,1,0,1,0],
                                  [1,1,1,1,1,1,1,1],
                                   ...
                                  [0,0,1,1,0,0,1,1]],
                     "dec-self": [[...],
                                   ...,
                                  [...]],
                     "dec-enc": [[...],
                                  ...,
                                 [...]],
                    },

Problem (loss function)

You need to set the training objective for you model. For the baseline and fixed head configuration, it's the standard cross-entropy loss with no extra options:

params=(
    ...
    --problem lib.task.seq2seq.problems.default.DefaultProblem
    --problem-opts '{}'
    ...)

For pruning heads, loss function is L = L_xent + λ * L_C.. You need to set another problem and specify the value of λ:

params=(
    ...
     --problem lib.task.seq2seq.problems.concrete.ConcreteProblem
     --problem-opts '{'"'"'concrete_coef'"'"': 0.1,}'
    ...)

Starting checkpoint

If you start model training from already trained model (for example, we start pruning heads from the trained baseline model), specify the initial checkpoint:

params=(
    ...
     --pre-init-model-checkpoint 'dir_to_your_trained_baseline_checkpoint.npz'
    ...)

You do not need this if you start from scratch.


Variables to optimize

If you want to freeze some sets of parameters in the model (for example, when pruning encoder heads we freeze the decoder parameters to ensure that heads functions do not move to the decoder), you have to specify which parameters you want to optimize. To optimize only encoder, add variables to --optimizer-opts:

params=(
    ...
    --optimizer-opts '{'"'"'beta1'"'"': 0.9, '"'"'beta2'"'"': 0.998,
                       '"'"'variables'"'"': ['"'"'mod/emb_inp*'"'"',
                                             '"'"'mod/enc*'"'"',],}'
    ...)

(Here beta1 and beta2 are parameters of the adam optimizer).


Batch size

It has been shown that Transformer’s performance depends heavily on a batch size (see for example Popel and Bojar, 2018), and we chose a large value of batch size to ensure that models show their best performance. In our experiments, each training batch contained a set of translation pairs containing approximately 16000 source tokens. This can be reached by using several of GPUs or by accumulating the gradients for several batches and then making an update. Our implementation enables both these options.

Batch size per one gpu is set like this:

params=(
    ...
     --batch-len 4000
    ...)

The effective batch size will be then batch-len * num_gpus. For example, with --batch-len 4000 and 4 gpus you would get the desirable batch size of 16000.

If you do not have several gpus (often, we don't have either :) ), you still have to have models of a proper quality. For this, accumulate the gradients for several batches and then make an update. Add average_grads: True and sync_every_steps: N to the optimizer options like this:

params=(
    ...
    --optimizer-opts '{'"'"'beta1'"'"': 0.9, '"'"'beta2'"'"': 0.998,
                       '"'"'sync_every_steps'"'"': 4,
                       '"'"'average_grads'"'"': True, }'
    ...)

The effective batch size will be then batch-len * sync_every_steps. For example, with --batch-len 4000 and sync_every_steps: 4 you would get the desirable batch size of 16000.


Other options

If you want to see dev BLEU score on your tensorboard:

params=(
    ...
      --translate-dev
      --translate-dev-every 2048
    ...)

Specify how often you want to save a checkpoint:

params=(
    ...
      --checkpoint-every-steps 2048
    ...)

Specify how often you want to score the dev set (eval loss values):

params=(
    ...
      --score-dev-every 256
    ...)

How many last checkpoints to keep:

params=(
    ...
       --keep-checkpoints-max 10
    ...)

Comments

  • lib.task.seq2seq.models.transformer_head_gates model enables you to train baseline as well as other versions, but if you want Transformer model without any modifications, you can find it here: lib.task.seq2seq.models.transformer.
Owner
PhD student at Edinburgh Uni and Amsterdam Uni, ex-research scientist at Yandex Research
"Graph Neural Controlled Differential Equations for Traffic Forecasting", AAAI 2022

Graph Neural Controlled Differential Equations for Traffic Forecasting Setup Python environment for STG-NCDE Install python environment $ conda env cr

Jeongwhan Choi 55 Dec 28, 2022
Code for EMNLP2021 paper "Allocating Large Vocabulary Capacity for Cross-lingual Language Model Pre-training"

VoCapXLM Code for EMNLP2021 paper Allocating Large Vocabulary Capacity for Cross-lingual Language Model Pre-training Environment DockerFile: dancingso

Bo Zheng 15 Jul 28, 2022
A boosting-based Multiple Instance Learning (MIL) package that includes MIL-Boost and MCIL-Boost

A boosting-based Multiple Instance Learning (MIL) package that includes MIL-Boost and MCIL-Boost

Jun-Yan Zhu 27 Aug 08, 2022
Files for a tutorial to train SegNet for road scenes using the CamVid dataset

SegNet and Bayesian SegNet Tutorial This repository contains all the files for you to complete the 'Getting Started with SegNet' and the 'Bayesian Seg

Alex Kendall 800 Dec 31, 2022
PyTorch implementations for our SIGGRAPH 2021 paper: Editable Free-viewpoint Video Using a Layered Neural Representation.

st-nerf We provide PyTorch implementations for our paper: Editable Free-viewpoint Video Using a Layered Neural Representation SIGGRAPH 2021 Jiakai Zha

Diplodocus 258 Jan 02, 2023
Use MATLAB to simulate the signal and extract features. Use PyTorch to build and train deep network to do spectrum sensing.

Deep-Learning-based-Spectrum-Sensing Use MATLAB to simulate the signal and extract features. Use PyTorch to build and train deep network to do spectru

10 Dec 14, 2022
Code for Transformers Solve Limited Receptive Field for Monocular Depth Prediction

Official PyTorch code for Transformers Solve Limited Receptive Field for Monocular Depth Prediction. Guanglei Yang, Hao Tang, Mingli Ding, Nicu Sebe,

stanley 152 Dec 16, 2022
PyTorch implementation for the paper Pseudo Numerical Methods for Diffusion Models on Manifolds

Pseudo Numerical Methods for Diffusion Models on Manifolds (PNDM) This repo is the official PyTorch implementation for the paper Pseudo Numerical Meth

Luping Liu (刘路平) 196 Jan 05, 2023
Official implementation for CVPR 2021 paper: Adaptive Class Suppression Loss for Long-Tail Object Detection

Adaptive Class Suppression Loss for Long-Tail Object Detection This repo is the official implementation for CVPR 2021 paper: Adaptive Class Suppressio

CASIA-IVA-Lab 67 Dec 04, 2022
All the code and files related to the MI-Lab of UE19CS305 course in sem 5

Machine-Intelligence-Lab-CS305 The compilation of all the code an drelated files from MI-Lab UE19CS305 (of batch 2019-2023) offered by PES University

Arvind Krishna 3 Nov 10, 2022
​ This is the Pytorch implementation of Progressive Attentional Manifold Alignment.

PAMA This is the Pytorch implementation of Progressive Attentional Manifold Alignment. Requirements python 3.6 pytorch 1.2.0+ PIL, numpy, matplotlib C

98 Nov 15, 2022
Language-Driven Semantic Segmentation

Language-driven Semantic Segmentation (LSeg) The repo contains official PyTorch Implementation of paper Language-driven Semantic Segmentation. Authors

Intelligent Systems Lab Org 416 Jan 03, 2023
make ASCII Art by Deep Learning

DeepAA This is convolutional neural networks generating ASCII art. This repository is under construction. This work is accepted by NIPS 2017 Workshop,

OsciiArt 1.4k Dec 28, 2022
The lightweight PyTorch wrapper for high-performance AI research. Scale your models, not the boilerplate.

The lightweight PyTorch wrapper for high-performance AI research. Scale your models, not the boilerplate. Website • Key Features • How To Use • Docs •

Pytorch Lightning 21.1k Jan 01, 2023
Invariant Causal Prediction for Block MDPs

MISA Abstract Generalization across environments is critical to the successful application of reinforcement learning algorithms to real-world challeng

Meta Research 41 Sep 17, 2022
Project of 'TBEFN: A Two-branch Exposure-fusion Network for Low-light Image Enhancement '

TBEFN: A Two-branch Exposure-fusion Network for Low-light Image Enhancement Codes for TMM20 paper "TBEFN: A Two-branch Exposure-fusion Network for Low

KUN LU 31 Nov 06, 2022
Unofficial PyTorch implementation of Fastformer based on paper "Fastformer: Additive Attention Can Be All You Need"."

Fastformer-PyTorch Unofficial PyTorch implementation of Fastformer based on paper Fastformer: Additive Attention Can Be All You Need. Usage : import t

Hong-Jia Chen 126 Dec 06, 2022
OneShot Learning-based hotword detection.

EfficientWord-Net Hotword detection based on one-shot learning Home assistants require special phrases called hotwords to get activated (eg:"ok google

ANT-BRaiN 102 Dec 25, 2022
Official implementation of "Accelerating Reinforcement Learning with Learned Skill Priors", Pertsch et al., CoRL 2020

Accelerating Reinforcement Learning with Learned Skill Priors [Project Website] [Paper] Karl Pertsch1, Youngwoon Lee1, Joseph Lim1 1CLVR Lab, Universi

Cognitive Learning for Vision and Robotics (CLVR) lab @ USC 134 Dec 06, 2022
Revealing and Protecting Labels in Distributed Training

Revealing and Protecting Labels in Distributed Training

Google Interns 0 Nov 09, 2022