An implementation of DeepMind's Relational Recurrent Neural Networks in PyTorch.

Overview

relational-rnn-pytorch

An implementation of DeepMind's Relational Recurrent Neural Networks (Santoro et al. 2018) in PyTorch.

Relational Memory Core (RMC) module is originally from official Sonnet implementation. However, currently they do not provide a full language modeling benchmark code.

This repo is a port of RMC with additional comments. It features a full-fledged word language modeling benchmark vs. traditional LSTM.

It supports any arbitrary word token-based text dataset, including WikiText-2 & WikiText-103.

Both RMC & LSTM models support adaptive softmax for much lower memory usage of large vocabulary dataset. RMC supports PyTorch's DataParallel, so you can easily experiment with a multi-GPU setup.

benchmark codes are hard-forked from official PyTorch word-language-model example

It also features an N-th farthest synthetic task from the paper (see below).

Requirements

PyTorch 0.4.1 or later (Tested on 1.0.0) & Python 3.6

Examples

python train_rmc.py --cuda for full training & test run of RMC with GPU.

python train_rmc.py --cuda --adaptivesoftmax --cutoffs 1000 5000 20000 if using large vocabulary dataset (like WikiText-103) to fit all the tensors in the VRAM.

python generate_rmc.py --cuda for generating sentences from the trained model.

python train_rnn.py --cuda for full training & test run of traditional RNN with GPU.

All default hyperparameters of RMC & LSTM are results from a two-week experiment using WikiText-2.

Data Preparation

Tested with WikiText-2 and WikiText-103. WikiText-2 is bundled.

Create a subfolder inside ./data and place word-level train.txt, valid.txt, and test.txt inside the subfolder.

Specify --data=(subfolder name) and you are good to go.

The code performs tokenization at the first training run, and the corpus is saved as pickle. The code will load the pickle file after the first run.

WikiText-2 Benchmark Results

Both RMC & LSTM have ~11M parameters. Please refer to the training code for details on hyperparameters.

Models Valid Perplexity Test Perplexity Forward pass ms/batch (TITAN Xp) Forward pass ms/batch (TITAN V)
LSTM (CuDNN) 111.31 105.56 26~27 40~41
LSTM (For Loop) Same as CuDNN Same as CuDNN 30~31 60~61
RMC 112.77 107.21 110~130 220~230

RMC can reach a comparable performance to LSTM (with heavy hyperparameter search), but it turns out that the RMC is very slow. The multi-head self-attention at every time step may be the culprit here. Using LSTMCell with for loop (which is more "fair" benchmark for RMC) slows down the forward pass, but it's still much faster.

Please also note that the hyperparameter for RMC is a worst-case scenario in terms of speed, because it used a single memory slot (as described in the paper) and did not benefit from a row-wise weight sharing from multi-slot memory.

Interesting to note here is that the speed is slower in TITAN V than TITAN Xp. The reason might be that the models are relatively small and the model calls small linear operations frequently.

Maybe TITAN Xp (~1,900Mhz unlocked CUDA clock speed vs. TITAN V's 1,335Mhz limit) benefits from these kind of workload. Or maybe TITAN V's CUDA kernel launch latency is higher for the ops in the model.

I'm not an expert in details of CUDA. Please share your results!

RMC Hyperparameter Search Results

Attention parameters tend to overfit the WikiText-2. reducing the hyperparmeters for attention (key_size) can combat the overfitting.

Applying dropout at the output logit before the softmax (like the LSTM one) helped preventing the overfitting.

embed & head size # heads attention MLP layers key size dropout at output memory slots test ppl
128 4 3 128 No 1 128.81
128 4 3 128 No 1 128.81
128 8 3 128 No 1 141.84
128 4 3 32 No 1 123.26
128 4 3 32 Yes 1 112.4
128 4 3 64 No 1 124.44
128 4 3 64 Yes 1 110.16
128 4 2 64 Yes 1 111.67
64 4 3 64 Yes 1 133.68
64 4 3 32 Yes 1 135.93
64 4 3 64 Yes 4 137.93
192 4 3 64 Yes 1 107.21
192 4 3 64 Yes 4 114.85
256 4 3 256 No 1 194.73
256 4 3 64 Yes 1 126.39

About WikiText-103

The original RMC paper presents WikiText-103 results with a larger model & batch size (6 Tesla P100, each with 64 batch size, so a total of 384. Ouch).

Using a full softmax easily blows up the VRAM. Using --adaptivesoftmax is highly recommended. If using --adaptivesoftmax, --cutoffs should be properly provided. Please refer to the original API description

I don't have such hardware and my resource is too limited to do the experiments. Benchmark result, or any other contributions are very welcome!

Nth Farthest Task

The objective of the task is: Given k randomly labelled (from 1 to k) D-dimensional vectors, identify which is the Nth farthest vector from vector M. (The answer is an integer from 1 to k.)

The specific task in the paper is: given 8 labelled 16-dimensional vectors, which is the Nth farthest vector from vector M? The vectors are labelled randomly so the model has to recognise that the Mth vector is the vector labelled as M as opposed to the vector in the Mth position in the input.

The input to the model comprises 8 40-dimensional vectors for each example. Each of these 40-dimensional vectors is structured like this:

[(vector 1) (label: which vector is it, from 1 to 8, one-hot encoded) (N, one-hot encoded) (M, one-hot encoded)] 

Example

python train_nth_farthest.py --cuda for training and testing on the Nth Farthest Task with GPU(s).

This uses the RelationalMemory class in relational_rnn_general.py, which is a version of relational_rnn_models.py without the language-modelling specific code.

Please refer totrain_nth_farthest.py for details on hyperparameter values. These are taken from Appendix A1 in the paper and from the Sonnet implementation when the hyperparameter values are not given in the paper.

Note: new examples are generated per epoch as in the Sonnet implementation. This seems to be consistent with the paper, which does not specify the number of examples used.

Experiment results

The model has been trained with a single TITAN Xp GPU for forever until it reaches 91% test accuracy. Below are the results with 3 independent runs:

The model does break the 25% barrier if trained long enough, but the wall clock time is roughly over 2~3x longer than those reported in the paper.

TODO

Experiment with different hyperparameters

Owner
Sang-gil Lee
Ph.D. student in ML/AI @ Seoul National University, South Korea. I do deep learning for sequence & generative models.
Sang-gil Lee
The authors' implementation of Unsupervised Adversarial Learning of 3D Human Pose from 2D Joint Locations

Unsupervised Adversarial Learning of 3D Human Pose from 2D Joint Locations This is the authors' implementation of Unsupervised Adversarial Learning of

Dwango Media Village 140 Dec 07, 2022
Adds timm pretrained backbone to pytorch's FasterRcnn model

Operating Systems Lab (ETCS-352) Experiments for Operating Systems Lab (ETCS-352) performed by me in 2021 at uni. All codes are written by me except t

Mriganka Nath 12 Dec 03, 2022
BLEURT is a metric for Natural Language Generation based on transfer learning.

BLEURT: a Transfer Learning-Based Metric for Natural Language Generation BLEURT is an evaluation metric for Natural Language Generation. It takes a pa

Google Research 492 Jan 05, 2023
code for generating data set ES-ImageNet with corresponding training code

es-imagenet-master code for generating data set ES-ImageNet with corresponding training code dataset generator some codes of ODG algorithm The variabl

Ordinarabbit 18 Dec 25, 2022
A high-level Python library for Quantum Natural Language Processing

lambeq About lambeq is a toolkit for quantum natural language processing (QNLP). Documentation: https://cqcl.github.io/lambeq/ Getting started Prerequ

Cambridge Quantum 315 Jan 01, 2023
Code to reproduce the results for Compositional Attention

Compositional-Attention This repository contains the official implementation for the paper Compositional Attention: Disentangling Search and Retrieval

Sarthak Mittal 58 Nov 30, 2022
Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"

Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis Abstract: This work targets at using a general deep lea

163 Dec 14, 2022
(AAAI 2021) Progressive One-shot Human Parsing

End-to-end One-shot Human Parsing This is the official repository for our two papers: Progressive One-shot Human Parsing (AAAI 2021) End-to-end One-sh

54 Dec 30, 2022
Python scripts to detect faces in Python with the BlazeFace Tensorflow Lite models

Python scripts to detect faces using Python with the BlazeFace Tensorflow Lite models. Tested on Windows 10, Tensorflow 2.4.0 (Python 3.8).

Ibai Gorordo 46 Nov 17, 2022
[CVPRW 2022] Attentions Help CNNs See Better: Attention-based Hybrid Image Quality Assessment Network

Attention Helps CNN See Better: Hybrid Image Quality Assessment Network [CVPRW 2022] Code for Hybrid Image Quality Assessment Network [paper] [code] T

IIGROUP 49 Dec 11, 2022
Official Implementation of SimIPU: Simple 2D Image and 3D Point Cloud Unsupervised Pre-Training for Spatial-Aware Visual Representations

Official Implementation of SimIPU SimIPU: Simple 2D Image and 3D Point Cloud Unsupervised Pre-Training for Spatial-Aware Visual Representations Since

Zhyever 37 Dec 01, 2022
OpenMMLab's Next Generation Video Understanding Toolbox and Benchmark

Introduction English | 简体中文 MMAction2 is an open-source toolbox for video understanding based on PyTorch. It is a part of the OpenMMLab project. The m

OpenMMLab 2.7k Jan 07, 2023
DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.

DeepSpeed+Megatron trained the world's most powerful language model: MT-530B DeepSpeed is hiring, come join us! DeepSpeed is a deep learning optimizat

Microsoft 8.4k Dec 28, 2022
Code for paper "Do Language Models Have Beliefs? Methods for Detecting, Updating, and Visualizing Model Beliefs"

This is the codebase for the paper: Do Language Models Have Beliefs? Methods for Detecting, Updating, and Visualizing Model Beliefs Directory Structur

Peter Hase 19 Aug 21, 2022
Compositional and Parameter-Efficient Representations for Large Knowledge Graphs

NodePiece - Compositional and Parameter-Efficient Representations for Large Knowledge Graphs NodePiece is a "tokenizer" for reducing entity vocabulary

Michael Galkin 107 Jan 04, 2023
Deep Multimodal Neural Architecture Search

MMNas: Deep Multimodal Neural Architecture Search This repository corresponds to the PyTorch implementation of the MMnas for visual question answering

Vision and Language Group@ MIL 23 Dec 21, 2022
PyG (PyTorch Geometric) - A library built upon PyTorch to easily write and train Graph Neural Networks (GNNs)

PyG (PyTorch Geometric) is a library built upon PyTorch to easily write and train Graph Neural Networks (GNNs) for a wide range of applications related to structured data.

PyG 16.5k Jan 08, 2023
Python版OpenCVのTracking APIのサンプルです。DaSiamRPNアルゴリズムまで対応しています。

OpenCV-Object-Tracker-Sample Python版OpenCVのTracking APIのサンプルです。   Requirement opencv-contrib-python 4.5.3.56 or later Algorithm 2021/07/16時点でOpenCVには以

KazuhitoTakahashi 36 Jan 01, 2023
Image data augmentation scheduler for albumentations transforms

albu_scheduler Scheduler for albumentations transforms based on PyTorch schedulers interface Usage TransformMultiStepScheduler import albumentations a

19 Aug 04, 2021
Official implementation of "GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators" (NeurIPS 2020)

GS-WGAN This repository contains the implementation for GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators (NeurIPS

46 Nov 09, 2022