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
WSDM2022 Challenge - Large scale temporal graph link prediction

WSDM 2022 Large-scale Temporal Graph Link Prediction - Baseline and Initial Test Set WSDM Cup Website link Link to this challenge This branch offers A

Deep Graph Library 34 Dec 29, 2022
Kaggle-titanic - A tutorial for Kaggle's Titanic: Machine Learning from Disaster competition. Demonstrates basic data munging, analysis, and visualization techniques. Shows examples of supervised machine learning techniques.

Kaggle-titanic This is a tutorial in an IPython Notebook for the Kaggle competition, Titanic Machine Learning From Disaster. The goal of this reposito

Andrew Conti 800 Dec 15, 2022
A Shading-Guided Generative Implicit Model for Shape-Accurate 3D-Aware Image Synthesis

A Shading-Guided Generative Implicit Model for Shape-Accurate 3D-Aware Image Synthesis Project Page | Paper A Shading-Guided Generative Implicit Model

Xingang Pan 115 Dec 18, 2022
SWA Object Detection

SWA Object Detection This project hosts the scripts for training SWA object detectors, as presented in our paper: @article{zhang2020swa, title={SWA

237 Nov 28, 2022
Code for reproducing experiments in "Improved Training of Wasserstein GANs"

Improved Training of Wasserstein GANs Code for reproducing experiments in "Improved Training of Wasserstein GANs". Prerequisites Python, NumPy, Tensor

Ishaan Gulrajani 2.2k Jan 01, 2023
CNN Based Meta-Learning for Noisy Image Classification and Template Matching

CNN Based Meta-Learning for Noisy Image Classification and Template Matching Introduction This master thesis used a few-shot meta learning approach to

Kumar Manas 2 Dec 09, 2021
Code for "Multi-Time Attention Networks for Irregularly Sampled Time Series", ICLR 2021.

Multi-Time Attention Networks (mTANs) This repository contains the PyTorch implementation for the paper Multi-Time Attention Networks for Irregularly

The Laboratory for Robust and Efficient Machine Learning 68 Dec 17, 2022
A framework for using LSTMs to detect anomalies in multivariate time series data. Includes spacecraft anomaly data and experiments from the Mars Science Laboratory and SMAP missions.

Telemanom (v2.0) v2.0 updates: Vectorized operations via numpy Object-oriented restructure, improved organization Merge branches into single branch fo

Kyle Hundman 844 Dec 28, 2022
A deep-learning pipeline for segmentation of ambiguous microscopic images.

Welcome to Official repository of deepflash2 - a deep-learning pipeline for segmentation of ambiguous microscopic images. Quick Start in 30 seconds se

Matthias Griebel 39 Dec 19, 2022
Computer Vision Script to recognize first person motion, developed as final project for the course "Machine Learning and Deep Learning"

Overview of The Code BaseColab/MLDL_FPAR.pdf: it contains the full explanation of our work Base Colab: it contains the base colab used to perform all

Simone Papicchio 4 Jul 16, 2022
Code for "Steerable Pyramid Transform Enables Robust Left Ventricle Quantification"

Code for "Steerable Pyramid Transform Enables Robust Left Ventricle Quantification" This is an end-to-end framework for accurate and robust left ventr

2 Jul 09, 2022
List of all dependencies affected by node-ipc malicious commit

node-ipc-dependencies-list List of all dependencies affected by node-ipc malicious commit as of 17/3/2022 - 19/3/2022 (timestamp) Please improve upon

99 Oct 15, 2022
J.A.R.V.I.S is an AI virtual assistant made in python.

J.A.R.V.I.S is an AI virtual assistant made in python. Running JARVIS Without Python To run JARVIS without python: 1. Head over to our installation pa

somePythonProgrammer 16 Dec 29, 2022
From Canonical Correlation Analysis to Self-supervised Graph Neural Networks

Code for CCA-SSG model proposed in the NeurIPS 2021 paper From Canonical Correlation Analysis to Self-supervised Graph Neural Networks.

Hengrui Zhang 44 Nov 27, 2022
Implementation of "A MLP-like Architecture for Dense Prediction"

A MLP-like Architecture for Dense Prediction (arXiv) Updates (22/07/2021) Initial release. Model Zoo We provide CycleMLP models pretrained on ImageNet

Shoufa Chen 244 Dec 27, 2022
Quickly and easily create / train a custom DeepDream model

Dream-Creator This project aims to simplify the process of creating a custom DeepDream model by using pretrained GoogleNet models and custom image dat

55 Dec 27, 2022
Fast Axiomatic Attribution for Neural Networks (NeurIPS*2021)

Fast Axiomatic Attribution for Neural Networks This is the official repository accompanying the NeurIPS 2021 paper: R. Hesse, S. Schaub-Meyer, and S.

Visual Inference Lab @TU Darmstadt 11 Nov 21, 2022
The implementation of the paper "HIST: A Graph-based Framework for Stock Trend Forecasting via Mining Concept-Oriented Shared Information".

The HIST framework for stock trend forecasting The implementation of the paper "HIST: A Graph-based Framework for Stock Trend Forecasting via Mining C

Wentao Xu 110 Dec 27, 2022
Bottleneck Transformers for Visual Recognition

Bottleneck Transformers for Visual Recognition Experiments Model Params (M) Acc (%) ResNet50 baseline (ref) 23.5M 93.62 BoTNet-50 18.8M 95.11% BoTNet-

Myeongjun Kim 236 Jan 03, 2023
Underwater image enhancement

LANet Our work proposes an adaptive learning attention network (LANet) to solve the problem of color casts and low illumination in underwater images.

LiuShiBen 7 Sep 14, 2022