Official PyTorch implementation of Time-aware Large Kernel (TaLK) Convolutions (ICML 2020)

Vasileios Lioutas

Last update: Dec 7, 2022

Related tags

Overview

Time-aware Large Kernel (TaLK) Convolutions (Lioutas et al., 2020)

This repository contains the source code, pre-trained models, as well as instructions to reproduce results for our paper Time-aware Large Kernel Convolutions (ICML 2020).

TaLK Convolutions is a sequence modeling method that uses an adaptive convolution operation that learns to predict the size of a summation kernel instead of using a fixed-sized learnable kernel matrix. It utilizes a fast parallelized implementation of the summed-area table, also known as the integral image operation, to efficiently calculate the convolution output that uses the summation kernel. We generate relative offsets for each timestep of the input sequence, which are used to adaptively expand the size of the summation kernel conditioned on the input. This method yields a time complexity of O(n), effectively making the sequence encoding process linear to the number of tokens.

Video Presentation:

Citation:

@inproceedings{lioutas2020timeaware,
    author={Vasileios Lioutas and Yuhong Guo},
    title={Time-aware Large Kernel Convolutions},
    booktitle={Proceedings of the 37th International Conference on Machine Learning (ICML)},
    year={2020}
}

Setup

Requirements

PyTorch version >= 1.3.1
fairseq version >= 0.10.1
Python version >= 3.6
CUDA >= 10.1
NVIDIA's apex library (for mixed-precision training)

Clone this repository

git clone https://github.com/lioutasb/TaLKConvolutions.git
cd TaLKConvolutions

Efficient CUDA Kernels

In order to support the parallelization of TaLK Convolutions, we have developed our own CUDA primitives. To install the kernels, use the commands below. We tested compiling the kernels using CUDA 10.1 but if a future CUDA release does not work, please feel free to open an issue.

cd talkconv/talkconv_module/
python setup.py install

We are welcoming contributions from experienced CUDA developers regarding making the CUDA kernels more efficient.

Translation

Pre-trained models

Dataset	Model	Prepared test set
IWSLT14 German-English	download (.pt)	IWSLT14 test: download (.zip)
WMT16 English-German	download (.pt)	newstest2014: download (.zip)
WMT14 English-French	download (.pt)	newstest2014: download (.zip)

Preprocessing the training datasets

Please follow the instructions https://github.com/pytorch/fairseq/blob/master/examples/translation/README.md to preprocess the data.

IWSLT14 De-En

Training and evaluating TaLK Convolutions on a single GPU:

# Training
SAVE="checkpoints/talkconv_iwslt_deen"
mkdir -p $SAVE

CUDA_VISIBLE_DEVICES=0 \
fairseq-train data-bin/iwslt14.tokenized.de-en \
    --user-dir talkconv/talkconv_fairseq \
    --arch talkconv_iwslt_de_en \
    --optimizer adam  --fp16 --lr 0.0005 \
    --source-lang de --target-lang en --max-tokens 4000 \
    --min-lr '1e-09' --weight-decay 0.0001 \
    --criterion label_smoothed_cross_entropy --label-smoothing 0.1 \
    --lr-scheduler inverse_sqrt \
    --dropout 0.3 --attention-dropout 0.1 --weight-dropout 0.1  \
    --max-update 85000 --warmup-updates 4000 --warmup-init-lr '1e-07' \
    --adam-betas '(0.9, 0.98)' --left-pad-source "False" --max-epoch 52 --seed 1024 \
    --save-dir $SAVE 

python utils/average_checkpoints.py --inputs $SAVE \
    --num-epoch-checkpoints 10 --output "${SAVE}/model.pt"

# Evaluation
fairseq-generate data-bin/iwslt14.tokenized.de-en --user-dir talkconv/talkconv_fairseq \
    --path "${SAVE}/model.pt" \
    --batch-size 128 --beam 5 --remove-bpe --lenpen 1.6 --gen-subset test --quiet

WMT16 En-De

Training and evaluating TaLK Convolutions on WMT16 En-De using cosine scheduler on one machine with 8 NVIDIA GPUs:

# Training
SAVE="checkpoints/talkconv_wmt_ende_big"
mkdir -p $SAVE

python -m torch.distributed.launch --nproc_per_node 8 fairseq-train \
    data-bin/wmt16_en_de_bpe32k --fp16 --log-interval 100 --no-progress-bar --distributed-no-spawn \
    --user-dir talkconv/talkconv_fairseq \
    --max-update 30243 --share-all-embeddings --optimizer adam \
    --adam-betas '(0.9, 0.98)' --clip-norm 0.0 --weight-decay 0.0 \
    --criterion label_smoothed_cross_entropy --label-smoothing 0.1 \
    --min-lr 1e-09 --update-freq 16 \
    --ddp-backend=no_c10d --max-tokens 3584 \
    --lr-scheduler cosine --warmup-init-lr 1e-7 --warmup-updates 10000 \
    --lr-shrink 1 --max-lr 0.001 --lr 1e-7 --min-lr 1e-9 --warmup-init-lr 1e-07 \
    --t-mult 1 --lr-period-updates 20000 \
    --arch talkconv_wmt_en_de_big \
    --save-dir $SAVE

# Checkpoint averaging
python utilss/average_checkpoints.py --inputs $SAVE \
    --num-epoch-checkpoints 10 --output "${SAVE}/model.pt"

# Evaluation on newstest2014
CUDA_VISIBLE_DEVICES=0 \
fairseq-generate data-bin/wmt16_en_de_bpe32k --user-dir talkconv/talkconv_fairseq \
  --path "${SAVE}/model.pt" \
  --batch-size 128 --beam 4 --remove-bpe --lenpen 0.35 --gen-subset test > wmt14_gen_ende.txt 

bash utils/compound_split_bleu.sh wmt14_gen_ende.txt

WMT14 En-Fr

Training and evaluating TaLK Convolutions on WMT14 En-Fr using cosine scheduler on one machine with 8 NVIDIA GPUs:

# Training
SAVE="checkpoints/talkconv_wmt_enfr_big"
mkdir -p $SAVE
python -m torch.distributed.launch --nproc_per_node 8 fairseq-train \
    data-bin/wmt14_en_fr --fp16 --log-interval 100 --no-progress-bar --distributed-no-spawn \
    --user-dir talkconv/talkconv_fairseq \
    --max-update 80000 --share-all-embeddings --optimizer adam \
    --adam-betas '(0.9, 0.98)' --clip-norm 0.0 --weight-decay 0.0 \
    --criterion label_smoothed_cross_entropy --label-smoothing 0.1 \
    --min-lr 1e-09 --update-freq 32 \
    --ddp-backend=no_c10d --max-tokens 1800 \
    --lr-scheduler cosine --warmup-init-lr 1e-7 --warmup-updates 10000 \
    --lr-shrink 1 --max-lr 0.001 --lr 1e-7 --min-lr 1e-9 --warmup-init-lr 1e-07 \
    --t-mult 1 --lr-period-updates 70000 \
    --arch talkconv_wmt_en_fr_big \
    --save-dir $SAVE

# Checkpoint averaging
python utils/average_checkpoints.py --inputs $SAVE \
    --num-epoch-checkpoints 10 --output "${SAVE}/model.pt"

# Evaluation
CUDA_VISIBLE_DEVICES=0 \
fairseq-generate data-bin/wmt14_en_fr --user-dir talkconv/talkconv_fairseq \
    --path "${SAVE}/model.pt" \
    --batch-size 128 --beam 6 --remove-bpe --lenpen 0.65 --gen-subset test --quiet

License

This project is MIT-licensed. The license applies to the pre-trained models as well.

IndoBERTweet is the first large-scale pretrained model for Indonesian Twitter. Published at EMNLP 2021 (main conference)

IndoBERTweet 🐦 🇮🇩 1. Paper Fajri Koto, Jey Han Lau, and Timothy Baldwin. IndoBERTweet: A Pretrained Language Model for Indonesian Twitter with Effe

40 Nov 30, 2022

WIT (Wikipedia-based Image Text) Dataset is a large multimodal multilingual dataset comprising 37M+ image-text sets with 11M+ unique images across 100+ languages.

740 Dec 24, 2022

BMInf (Big Model Inference) is a low-resource inference package for large-scale pretrained language models (PLMs).

377 Jan 2, 2023

An easy-to-use Python module that helps you to extract the BERT embeddings for a large text dataset (Bengali/English) efficiently.

37 Sep 5, 2022

Code for text augmentation method leveraging large-scale language models

HyperMix Code for our paper GPT3Mix and conducting classification experiments using GPT-3 prompt-based data augmentation. Getting Started Installing P

47 Dec 20, 2022

Tools for curating biomedical training data for large-scale language modeling

242 Dec 25, 2022

The model is designed to train a single and large neural network in order to predict correct translation by reading the given sentence.

Neural Machine Translation communication system The model is basically direct to convert one source language to another targeted language using encode

7 Sep 22, 2022

Large-scale Knowledge Graph Construction with Prompting

Large-scale Knowledge Graph Construction with Prompting across tasks (predictive and generative), and modalities (language, image, vision + language, etc.)

161 Dec 28, 2022

Guide to using pre-trained large language models of source code

Large Models of Source Code I occasionally train and publicly release large neural language models on programs, including PolyCoder. Here, I describe

947 Dec 28, 2022

Comments

how to understand the following line of code?

const scalar_t S_output = ((1.0 - alpha_right)input[ind_floor_right][batchIdx][rIdx] + alpha_rightinput[ind_ceil_right][batchIdx][rIdx]) - (alpha_left*((ind_floor_left-1 < 0)?static_cast<scalar_t>(0.0):input[ind_floor_left-1][batchIdx][rIdx]) + (1.0 - alpha_left)*((ind_ceil_left-1 < 0)?static_cast<scalar_t>(0.0):input[ind_ceil_left-1][batchIdx][rIdx]));

How to understand the following code? ((ind_floor_left-1 < 0)?static_cast<scalar_t>(0.0):input[ind_floor_left-1][batchIdx][rIdx] (ind_ceil_left-1 < 0)?static_cast<scalar_t>(0.0):input[ind_ceil_left-1][batchIdx][rIdx])

opened by gaopengcuhk 1
Question about Encoder_attn in decoder layer

It seems that in the decoder, both self attention and TaLK has been used. Have you ever try to replace Self Attention(Co attention between encoder input) with TaLK?

opened by gaopengcuhk 1