An implementation of the Pay Attention when Required transformer

Overview

Pay Attention when Required (PAR) Transformer-XL

An implementation of the Pay Attention when Required transformer from the paper: https://arxiv.org/pdf/2009.04534.pdf

alt text [source: Jonathan Kernes]

Quick overview

The Pay Attention when Required Transformer (Mandava, et. al. 2020) is just a regular transformer-XL (Dai et. al. 2019)[https://arxiv.org/pdf/1901.02860.pdf] , but the ratio of attention and dense layers has been optimized. This optimization is performed by allowing the network to choose which types of layer it prefers in each block of the network. The present implementation is not an exact replica of the author's efforts. Instead, we perform a simultaneous optimization procedure on both the model architecture and model parameters. The search is performed using a SuperNet, which is a sequential neural network composed of stochastic blocks, as shown in the figure below (taken from the paper. Please don't sue me!)

alt text [Source: Mandava et. al. 2020]

The key component is a Gumbel-Softmax layer [(Jang et al., 2016) and (Maddison et al., 2016). jang link: https://arxiv.org/pdf/1611.01144.pdf]. This layer is a continuous representation of a discrete sampling from a Categorical distribution, thereby allowing us to use gradients to learn parameters of a discrete distribution. (Recall a categorical is a distrbution over K states with kth state having probability pi_k, and we must have the normalization condition \sum_{i=1}^K pi_i = 1)

As the model learns, it is free to adjust both the usual model parameters, as well as its architecture search parameters pi, indicating the probability of choosing either

  1. Attention

  2. Dense

  3. Identity

for any given stochastic block. We perform simulated annealing: since the categorical distribution is approximated by a continuous representation, we get some scores like (0.02, 0.98, 0.02) for the probability of say sampling that state 2 is picked. The sharpness of this is set by a parameter \tau (the temperature), with a categorical distribution the limit tau-->0. Simulated annealing means we begin with tau=1 to let the model figure out what it wants, then slowly decrease tau so the distribution approaches a categorical.

All of this is implemented on the freely available wiki-text2 dataset.

Explanation of the main GIF: The main gif is the result of our experiments. It shows the pi distribution for each stochastic block of a 6 block SuperNet, as a function of training iterations. The number indicates the probability of the most likely layer type (darker means more probable). As you can see, the model learns to put attention in the beginning, and dense layers at the end.

Requirements

Usual ML stuff, if you have a conda environment, python 3+, TensorFlow 2+ you should be ok. You will need TensorFlow Text as well to handle the SentencePiece Tokenization

If you choose to run your own tokenizer (a flag option in data_utils for handling new text data), you will also need to download the SentencePiece package: https://github.com/google/sentencepiece

Data

The dataset used is Wiki-text2. We have provided a copy of this in the data folder, along with some preprocessed data for training. In order to reproduce this from scratch, run the shell script

./create_tfrecords.sh

This will download the wiki-text2 dataset from its source, then proceed to clean, batch, and write the data to a tfrecords file. The shell script calls build_data.py which offers more control over what type of data to generate. The general parameters you will want to tune are:

*batch_size *seq_len.

You can also supply your own dataset instead of the one provided. The underlying tokenizer uses sentencepiece (Kudo): https://github.com/google/sentencepiece, which works at the byte level and can handle any kind of input. Simply change the --input_text flag to your file, and set the desired --vocab_size.

Why do we need to specify the batch size? Transformer XL uses memory states to form a recurrent, long range network. After analyzing a particular sequence say [A,B] of the sequence [A,B,C,D], the results of [A,B] are fed into the [C,D] calculation with a stop gradient. Therefore, we must be sure that each datapoint follows chronologically from the previous one.

This is achieved by context batching (see data_utils.py function) where we break the entire dataset into batch_size segments, then pull in order one sequence from each batch at a time to form the dataset. Because of this, note that adding more shards to the data could result in a large loss (order of batch_size*seq_len*shards), as each shard will drop the remaining datapoint of size (batch_size*seq_len) to keep the tensor shapes.

Addtional technical details

Per the original Transformer-XL, we also implement an adaptive softmax layer (Grave et. al. 2017, https://arxiv.org/abs/1609.04309) to deal with a potentially large number of outputs in the final dense layer. This implemenation is inspired by the TF 1.0 example at https://github.com/yangsaiyong/tf-adaptive-softmax-lstm-lm. To use the adaptive softmax, set the --cutoffs= flag in train.py. The cutoffs are the max values of each bin, and should NOT include the vocab size (i.e. the max cutoff of the final bin). If no cutoffs are specified, the model defaults to normal softmax.

For completeness, we have also provided a script optimal_cuts.py that determines the optimal cutoffs given a return space separated file of unigram probabilities (based on the assumptions of Grave et. al. regarding GPU computation complexity -- see the paper for details). The algorithm uses dynamic programming, but is quite slow at O(KN^2), for K cutoffs and N vocab words. In principle it's a one time cost to determine the cutoffs, but we are impatient and recommend to just play around with the cutoffs instead. See the script for flag details

Training and Benchmarks

The default model we use has memory length 16, feed-forward dimension 1024, attention dimension 128, and 6 stochastic blocks, with an adaptive softmax layer and 2 clusters. We trained on a colab GPU for 20 epochs, taking a total of 37 minutes. We use an Adam optimzer with cosine rate decay: an initial warmup of 4000 steps and a maximum learning rate of 1e-4, decaying to zero at the end of training. Our training benchmarks are:

Iteration (thousands) Train_perplexity Validation_perplexity Time
2.7k 163.9 114.4 1m 58s
8.5k 78.56 62.33 5m 37s
14.1k 65.71 51.88 9m 28s
28.3k 48.52 42.61 18m 40s
48.1k 41.85 39.57 31m 51s
56.5k 42.12 39.41 37m 14s

To train, simply run the shell script

./base_model.sh

adjusting the parameters as you see fit. The above model is the default configuration. To train in colab, simply open up the notebook "colab.ipynb" and follow the instructions. This is most easily done by going to [google.colab.com] and searching this repository in github. The benefit of colab, is it's easier to play around with the model after training.

While training, we have provided two ways to monitor the output

  1. A tensorboard log. The colab notebook takes care of running this for you. In the terminal, first create a 'logs' directory, then run the command tensorboard --logdir logs in a separate tab. This will open a port where you can view live plots of the learning rate, tau annealing, train/valid loss and perplexity.

  2. An output log saved to training_log.log. This will log the model summary, parameters, etc. as well as print out loss updates every 100 steps and save it to the log file.

Thanks for reading this far!

Enjoy! And thank you to the wonderful researchers that inspired this project.

If you would like to contribute, or have any comments questions concerns please open a pull request or email me directly.

I label phrases on a scale of five values: negative, somewhat negative, neutral, somewhat positive, positive

I label phrases on a scale of five values: negative, somewhat negative, neutral, somewhat positive, positive. Obstacles like sentence negation, sarcasm, terseness, language ambiguity, and many others

1 Jan 13, 2022
Nmt - TensorFlow Neural Machine Translation Tutorial

Neural Machine Translation (seq2seq) Tutorial Authors: Thang Luong, Eugene Brevdo, Rui Zhao (Google Research Blogpost, Github) This version of the tut

6.1k Dec 29, 2022
Residual2Vec: Debiasing graph embedding using random graphs

Residual2Vec: Debiasing graph embedding using random graphs This repository contains the code for S. Kojaku, J. Yoon, I. Constantino, and Y.-Y. Ahn, R

SADAMORI KOJAKU 5 Oct 12, 2022
Crowd sourced training data for Rasa NLU models

NLU Training Data Crowd-sourced training data for the development and testing of Rasa NLU models. If you're interested in grabbing some data feel free

Rasa 169 Dec 26, 2022
Implemented shortest-circuit disambiguation, maximum probability disambiguation, HMM-based lexical annotation and BiLSTM+CRF-based named entity recognition

Implemented shortest-circuit disambiguation, maximum probability disambiguation, HMM-based lexical annotation and BiLSTM+CRF-based named entity recognition

0 Feb 13, 2022
A fast Text-to-Speech (TTS) model. Work well for English, Mandarin/Chinese, Japanese, Korean, Russian and Tibetan (so far). 快速语音合成模型,适用于英语、普通话/中文、日语、韩语、俄语和藏语(当前已测试)。

简体中文 | English 并行语音合成 [TOC] 新进展 2021/04/20 合并 wavegan 分支到 main 主分支,删除 wavegan 分支! 2021/04/13 创建 encoder 分支用于开发语音风格迁移模块! 2021/04/13 softdtw 分支 支持使用 Sof

Atomicoo 161 Dec 19, 2022
gaiic2021-track3-小布助手对话短文本语义匹配复赛rank3、决赛rank4

决赛答辩已经过去一段时间了,我们队伍ac milan最终获得了复赛第3,决赛第4的成绩。在此首先感谢一些队友的carry~ 经过2个多月的比赛,学习收获了很多,也认识了很多大佬,在这里记录一下自己的参赛体验和学习收获。

102 Dec 19, 2022
WikiPron - a command-line tool and Python API for mining multilingual pronunciation data from Wiktionary

WikiPron WikiPron is a command-line tool and Python API for mining multilingual pronunciation data from Wiktionary, as well as a database of pronuncia

213 Jan 01, 2023
Conditional probing: measuring usable information beyond a baseline

Conditional probing: measuring usable information beyond a baseline

John Hewitt 20 Dec 15, 2022
The Classical Language Toolkit

Notice: This Git branch (dev) contains the CLTK's upcoming major release (v. 1.0.0). See https://github.com/cltk/cltk/tree/master and https://docs.clt

Classical Language Toolkit 754 Jan 09, 2023
Trex is a tool to match semantically similar functions based on transfer learning.

Trex is a tool to match semantically similar functions based on transfer learning.

62 Dec 28, 2022
Tool to add main subject to items on Wikidata using a WMFs CirrusSearch for named entity recognition or a manually supplied list of QIDs

ItemSubjector Tool made to add main subject statements to items based on the title using a home-brewed CirrusSearch-based Named Entity Recognition alg

Dennis Priskorn 9 Nov 17, 2022
Sploitus - Command line search tool for sploitus.com. Think searchsploit, but with more POCs

Sploitus Command line search tool for sploitus.com. Think searchsploit, but with

watchdog2000 5 Mar 07, 2022
Code for our paper "Transfer Learning for Sequence Generation: from Single-source to Multi-source" in ACL 2021.

TRICE: a task-agnostic transferring framework for multi-source sequence generation This is the source code of our work Transfer Learning for Sequence

THUNLP-MT 9 Jun 27, 2022
The Easy-to-use Dialogue Response Selection Toolkit for Researchers

The Easy-to-use Dialogue Response Selection Toolkit for Researchers

GMFTBY 32 Nov 13, 2022
中文空间语义理解评测

中文空间语义理解评测 最新消息 2021-04-10 🚩 排行榜发布: Leaderboard 2021-04-05 基线系统发布: SpaCE2021-Baseline 2021-04-05 开放数据提交: 提交结果 2021-04-01 开放报名: 我要报名 2021-04-01 数据集 pa

40 Jan 04, 2023
Shared code for training sentence embeddings with Flax / JAX

flax-sentence-embeddings This repository will be used to share code for the Flax / JAX community event to train sentence embeddings on 1B+ training pa

Nils Reimers 23 Dec 30, 2022
Behavioral Testing of Clinical NLP Models

Behavioral Testing of Clinical NLP Models This repository contains code for testing the behavior of clinical prediction models based on patient letter

Betty van Aken 2 Sep 20, 2022
Twitter-NLP-Analysis - Twitter Natural Language Processing Analysis

Twitter-NLP-Analysis Business Problem I got last @turk_politika 3000 tweets with

Çağrı Karadeniz 7 Mar 12, 2022