Machine Learning

8-bit Numerical Formats for Deep Neural Networks

Given the current trend of increasing size and complexity of machine learning architectures, it has become of critical importance to identify new approaches to improve the computational efficiency of model training. In this context, we address the ad...

A BLOCK ORTHOGONALIZATION PROCEDURE WITH CONSTANT SYNCHRONIZATION REQUIREMENTS

We propose an alternative orthonormalization method that computes the orthonormal basis from the right singular vectors of a matrix. Its advantage are: a) all operations are matrix-matrix multiplications and thus cache-efficient, b) only one synchron...

Accelerated Proximal Stochastic Dual Coordinate Ascent for Regularized Loss Minimization

We introduce a proximal version of the stochastic dual coordinate ascent method and show how to accelerate the method using an inner-outer iteration procedure. We analyze the runtime of the framework and obtain rates that improve state-of-the-art res...

Accelerating Large-Scale Inference with Anisotropic Vector Quantization

Quantization based techniques are the current state-of-the-art for scaling maximum inner product search to massive databases. Traditional approaches to quantization aim to minimize the reconstruction error of the database points. Based on the observa...

Accuracy and stability of numerical algorithms, Second Edition.

From the Publisher: What is the most accurate way to sum floating point numbers? What are the advantages of IEEE arithmetic? How accurate is Gaussian elimination and what were the key breakthroughs in the development of error analysis for the method?...

Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour

Deep learning thrives with large neural networks and large datasets. However, larger networks and larger datasets result in longer training times that impede research and development progress. Distributed synchronous SGD offers a potential solution t...

A closed-form formula for the Kullback-Leibler divergence between Cauchy distributions

We report a closed-form expression for the Kullback-Leibler divergence between Cauchy distributions which involves the calculation of a novel definite integral. The formula shows that the Kullback-Leibler divergence between Cauchy densities is always...

ADADELTA: An Adaptive Learning Rate Method

We present a novel per-dimension learning rate method for gradient descent called ADADELTA. The method dynamically adapts over time using only first order information and has minimal computational overhead beyond vanilla stochastic gradient descent. ...

Adafactor: Adaptive Learning Rates with Sublinear Memory Cost

In several recently proposed stochastic optimization methods (e.g. RMSProp, Adam, Adadelta), parameter updates are scaled by the inverse square roots of exponential moving averages of squared past gradients. Maintaining these per-parameter second-mom...

Adam: A Method for Stochastic Optimization

We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little me...

Adaptive Subgradient Methods for Online Learning and Stochastic Optimization

We present a new family of subgradient methods that dynamically incorporate knowledge of the geometry of the data observed in earlier iterations to perform more informative gradient-based learning....

Ad click prediction: a view from the trenches.

Predicting ad click-through rates (CTR) is a massive-scale learning problem that is central to the multi-billion dollar online advertising industry. We present a selection of case studies and topics drawn from recent experiments in the setting of a d...

Adding vs. Averaging in Distributed Primal-Dual Optimization

Distributed optimization methods for large-scale machine learning suffer from a communication bottleneck. It is difficult to reduce this bottleneck while still efficiently and accurately aggregating partial work from different machines. In this paper...

Additive Powers-of-Two Quantization: An Efficient Non-uniform Discretization for Neural Networks

We propose Additive Powers-of-Two~(APoT) quantization, an efficient non-uniform quantization scheme for the bell-shaped and long-tailed distribution of weights and activations in neural networks. By constraining all quantization levels as the sum of ...

A guide to convolution arithmetic for deep learning

We introduce a guide to help deep learning practitioners understand and manipulate convolutional neural network architectures. The guide clarifies the relationship between various properties (input shape, kernel shape, zero padding, strides and outpu...

An Empirical Exploration of Recurrent Network Architectures.

This document examines the OData protocol as a new service oriented approach for distributed IT architectures. The main features of OData were compared with properties of well-established solutions like: REST, DCOM and Java RMI. OData's protocol is p...

A Robust and Efficient Implementation of LOBPCG.

Locally Optimal Block Preconditioned Conjugate Gradient (LOBPCG) is widely\nused to compute eigenvalues of large sparse symmetric matrices. The algorithm\ncan suffer from numerical instability if it is not implemented with care. This\nis especially p...

A Short Note about Kinetics-600

We describe an extension of the DeepMind Kinetics human action dataset from 400 classes, each with at least 400 video clips, to 600 classes, each with at least 600 video clips. In order to scale up the dataset we changed the data collection process s...

A simple method for generating gamma variables

We offer a procedure for generating a gamma variate as the cube of a suitably scaled normal variate. It is fast and simple, assuming one has a fast way to generate normal variables. In brief: generate a normal variate x and a uniform variate U until ...

A Theoretically Grounded Application of Dropout in Recurrent Neural Networks

Recurrent neural networks (RNNs) stand at the forefront of many recent developments in deep learning. Yet a major difficulty with these models is their tendency to overfit, with dropout shown to fail when applied to recurrent layers. Recent results a...

Attention Is All You Need

The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose ...

Automatic Cross-Replica Sharding of Weight Update in Data-Parallel Training

In data-parallel synchronous training of deep neural networks, different devices (replicas) run the same program with different partitions of the training batch, but weight update computation is repeated on all replicas, because the weights do not ha...

Auto-Vectorizing TensorFlow Graphs: Jacobians, Auto-Batching And Beyond

We propose a static loop vectorization optimization on top of high level dataflow IR used by frameworks like TensorFlow. A new statically vectorized parallel-for abstraction is provided on top of TensorFlow, and used for applications ranging from aut...

Averaging Weights Leads to Wider Optima and Better Generalization

Deep neural networks are typically trained by optimizing a loss function with an SGD variant, in conjunction with a decaying learning rate, until convergence. We show that simple averaging of multiple points along the trajectory of SGD, with a cyclic...

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful...

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations fro...

Breadth-First Pipeline Parallelism

We introduce Breadth-First Pipeline Parallelism, a novel training schedule which optimizes the combination of pipeline and data parallelism. Breadth-First Pipeline Parallelism lowers training time, cost and memory usage by combining a high GPU utiliz...

Categorical Reparameterization with Gumbel-Softmax

Categorical variables are a natural choice for representing discrete structure in the world. However, stochastic neural networks rarely use categorical latent variables due to the inability to backpropagate through samples. In this work, we present a...

Chakra: Advancing Performance Benchmarking and Co-design using Standardized Execution Traces

Benchmarking and co-design are essential for driving optimizations and innovation around ML models, ML software, and next-generation hardware. Full workload benchmarks, e.g. MLPerf, play an essential role in enabling fair comparison across different ...

Conditional Noise-Contrastive Estimation of Unnormalised Models

Many parametric statistical models are not properly normalised and only specified up to an intractable partition function, which renders parameter estimation difficult. Examples of unnormalised models are Gibbs distributions, Markov random fields, an...

Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks.

Many real-world sequence learning tasks require the prediction of sequences of labels from noisy, unsegmented input data. In speech recognition, for example, an acoustic signal is transcribed into words or sub-word units. Recurrent neural networks (R...

Continuously Differentiable Exponential Linear Units

Exponential Linear Units (ELUs) are a useful rectifier for constructing deep learning architectures, as they may speed up and otherwise improve learning by virtue of not have vanishing gradients and by having mean activations near zero. However, the ...

Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

The goal of precipitation nowcasting is to predict the future rainfall intensity in a local region over a relatively short period of time. Very few previous studies have examined this crucial and challenging weather forecasting problem from the machi...

Cyclical Learning Rates for Training Neural Networks

It is known that the learning rate is the most important hyper-parameter to tune for training deep neural networks. This paper describes a new method for setting the learning rate, named cyclical learning rates, which practically eliminates the need ...

Data-Free Quantization Through Weight Equalization and Bias Correction

We introduce a data-free quantization method for deep neural networks that does not require fine-tuning or hyperparameter selection. It achieves near-original model performance on common computer vision architectures and tasks. 8-bit fixed-point quan...

Deconvolutional networks

Building robust low and mid-level image representations, beyond edge primitives, is a long-standing goal in vision. Many existing feature detectors spatially pool edge information which destroys cues such as edge intersections, parallelism and symmet...

Decoupled Weight Decay Regularization

L$_2$ regularization and weight decay regularization are equivalent for standard stochastic gradient descent (when rescaled by the learning rate), but as we demonstrate this is \emph{not} the case for adaptive gradient algorithms, such as Adam. While...

Deep Residual Learning for Image Recognition

Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual function...

DeepSeek-V3 Technical Report

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) an...

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification.

Rectified activation units (rectifiers) are essential for state-of-the-art neural networks. In this work, we study rectifier neural networks for image classification from two aspects. First, we propose a Parametric Rectified Linear Unit (PReLU) that ...

Demystifying Why Local Aggregation Helps: Convergence Analysis of Hierarchical SGD

Hierarchical SGD (H-SGD) has emerged as a new distributed SGD algorithm for multi-level communication networks. In H-SGD, before each global aggregation, workers send their updated local models to local servers for aggregations. Despite recent resear...

Differentiation of the Cholesky decomposition

We review strategies for differentiating matrix-based computations, and derive symbolic and algorithmic update rules for differentiating expressions containing the Cholesky decomposition. We recommend new `blocked' algorithms, based on differentiatin...

Distributed Representations of Words and Phrases and their Compositionality

The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships. In this paper we present several ...

Don't Use Large Mini-Batches, Use Local SGD

Mini-batch stochastic gradient methods (SGD) are state of the art for distributed training of deep neural networks. Drastic increases in the mini-batch sizes have lead to key efficiency and scalability gains in recent years. However, progress faces a...

DoRA: Weight-Decomposed Low-Rank Adaptation

Among the widely used parameter-efficient fine-tuning (PEFT) methods, LoRA and its variants have gained considerable popularity because of avoiding additional inference costs. However, there still often exists an accuracy gap between these methods an...

Dynamical Isometry and a Mean Field Theory of CNNs: How to Train 10, 000-Layer Vanilla Convolutional Neural Networks.

In recent years, electroencephalogram (EEG) e-motion recognition has been becoming an emerging field in artificial intelligence area, which can reflect the relation between emotional states and brain activity. In this paper, we designed a novel archi...

Dynamic Control Flow in Large-Scale Machine Learning

Many recent machine learning models rely on fine-grained dynamic control flow for training and inference. In particular, models based on recurrent neural networks and on reinforcement learning depend on recurrence relations, data-dependent conditiona...

Efficient BackProp

Efficient ConvBN Blocks for Transfer Learning and Beyond

Convolution-BatchNorm (ConvBN) blocks are integral components in various computer vision tasks and other domains. A ConvBN block can operate in three modes: Train, Eval, and Deploy. While the Train mode is indispensable for training models from scrat...

Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LM

Large language models have led to state-of-the-art accuracies across a range of tasks. However, training these models efficiently is challenging for two reasons: a) GPU memory capacity is limited, making it impossible to fit large models on even a mu...

Efficient Learning using Forward-Backward Splitting.

In the wake of the sacramental crisis Asbury established a pattern of relentless travel by horseback across the continent that defined the church for decades to come. He visited New York City, which had been cut off by the war, in August 1783 and als...

Efficient Memory Management for Deep Neural Net Inference

While deep neural net inference was considered a task for servers only, latest advances in technology allow the task of inference to be moved to mobile and embedded devices, desired for various reasons ranging from latency to privacy. These devices a...

Efficient Object Localization Using Convolutional Networks

Recent state-of-the-art performance on human-body pose estimation has been achieved with Deep Convolutional Networks (ConvNets). Traditional ConvNet architectures include pooling and sub-sampling layers which reduce computational requirements, introd...

Efficient Simulation of the von Mises Distribution

Efficient softmax approximation for GPUs

We propose an approximate strategy to efficiently train neural network based language models over very large vocabularies. Our approach, called adaptive softmax, circumvents the linear dependency on the vocabulary size by exploiting the unbalanced wo...

Empirical Evaluation of Rectified Activations in Convolutional Network

In this paper we investigate the performance of different types of rectified activation functions in convolutional neural network: standard rectified linear unit (ReLU), leaky rectified linear unit (Leaky ReLU), parametric rectified linear unit (PReL...

Enabling Fast Differentially Private SGD via Just-in-Time Compilation and Vectorization

A common pain point in differentially private machine learning is the significant runtime overhead incurred when executing Differentially Private Stochastic Gradient Descent (DPSGD), which may be as large as two orders of magnitude. We thoroughly dem...

Estimating the mean and variance of the target probability distribution

Introduces a method that estimates the mean and the variance of the probability distribution of the target as a function of the input, given an assumed target error-distribution model. Through the activation of an auxiliary output unit, this method p...

Exact solutions to the nonlinear dynamics of learning in deep linear neural networks.

We investigate the use of large state inventories and the softplus nonlinearity for on-device neural network based mobile speech recognition. Large state inventories are achieved by less aggressive context-dependent state tying, and made possible by ...

Fast Algorithms for Convolutional Neural Networks

Deep convolutional neural networks take GPU days of compute time to train on large data sets. Pedestrian detection for self driving cars requires very low latency. Image recognition for mobile phones is constrained by limited processing resources. Th...

Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

We introduce the "exponential linear unit" (ELU) which speeds up learning in deep neural networks and leads to higher classification accuracies. Like rectified linear units (ReLUs), leaky ReLUs (LReLUs) and parametrized ReLUs (PReLUs), ELUs alleviate...

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN have reduced the running time of these detection networks, exposing region proposal computation as a ...

Fast Image Scanning with Deep Max-Pooling Convolutional Neural Networks

Deep Neural Networks now excel at image classification, detection and segmentation. When used to scan images by means of a sliding window, however, their high computational complexity can bring even the most powerful hardware to its knees. We show ho...

Fast R-CNN

This paper proposes a Fast Region-based Convolutional Network method (Fast R-CNN) for object detection. Fast R-CNN builds on previous work to efficiently classify object proposals using deep convolutional networks. Compared to previous work, Fast R-C...

Fast Sparse ConvNets

Historically, the pursuit of efficient inference has been one of the driving forces behind research into new deep learning architectures and building blocks. Some recent examples include: the squeeze-and-excitation module, depthwise separable convolu...

Fast Transformer Decoding: One Write-Head is All You Need

Multi-head attention layers, as used in the Transformer neural sequence model, are a powerful alternative to RNNs for moving information across and between sequences. While training these layers is generally fast and simple, due to parallelizability ...

Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Acceleration

Previous works utilized ''smaller-norm-less-important'' criterion to prune filters with smaller norm values in a convolutional neural network. In this paper, we analyze this norm-based criterion and point out that its effectiveness depends on two req...

Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions

Low-rank matrix approximations, such as the truncated singular value decomposition and the rank-revealing QR decomposition, play a central role in data analysis and scientific computing. This work surveys and extends recent research which demonstrate...

FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning

Scaling Transformers to longer sequence lengths has been a major problem in the last several years, promising to improve performance in language modeling and high-resolution image understanding, as well as to unlock new applications in code, audio, a...

FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

Transformers are slow and memory-hungry on long sequences, since the time and memory complexity of self-attention are quadratic in sequence length. Approximate attention methods have attempted to address this problem by trading off model quality to r...

Flipout: Efficient Pseudo-Independent Weight Perturbations on Mini-Batches

Stochastic neural net weights are used in a variety of contexts, including regularization, Bayesian neural nets, exploration in reinforcement learning, and evolution strategies. Unfortunately, due to the large number of weights, all the examples in a...

Forcing Generative Models to Degenerate Ones: The Power of Data Poisoning Attacks

Growing applications of large language models (LLMs) trained by a third party raise serious concerns on the security vulnerability of LLMs.It has been demonstrated that malicious actors can covertly exploit these vulnerabilities in LLMs through poiso...

FP8 Formats for Deep Learning

FP8 is a natural progression for accelerating deep learning training inference beyond the 16-bit formats common in modern processors. In this paper we propose an 8-bit floating point (FP8) binary interchange format consisting of two encodings - E4M3 ...

Fractional Max-Pooling

Convolutional networks almost always incorporate some form of spatial pooling, and very often it is alpha times alpha max-pooling with alpha=2. Max-pooling act on the hidden layers of the network, reducing their size by an integer multiplicative fact...

Gaussian Error Linear Units (GELUs)

We propose the Gaussian Error Linear Unit (GELU), a high-performing neural network activation function. The GELU activation function is $x\Phi(x)$, where $\Phi(x)$ the standard Gaussian cumulative distribution function. The GELU nonlinearity weights ...

Generating Sequences With Recurrent Neural Networks

This paper shows how Long Short-term Memory recurrent neural networks can be used to generate complex sequences with long-range structure, simply by predicting one data point at a time. The approach is demonstrated for text (where the data are discre...

GossipGraD: Scalable Deep Learning using Gossip Communication based Asynchronous Gradient Descent

In this paper, we present GossipGraD - a gossip communication protocol based Stochastic Gradient Descent (SGD) algorithm for scaling Deep Learning (DL) algorithms on large-scale systems. The salient features of GossipGraD are: 1) reduction in overall...

GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints

Multi-query attention (MQA), which only uses a single key-value head, drastically speeds up decoder inference. However, MQA can lead to quality degradation, and moreover it may not be desirable to train a separate model just for faster inference. We ...

Gradient Estimation Using Stochastic Computation Graphs

In a variety of problems originating in supervised, unsupervised, and reinforcement learning, the loss function is defined by an expectation over a collection of random variables, which might be part of a probabilistic model or the external world. Es...

Group Normalization

Batch Normalization (BN) is a milestone technique in the development of deep learning, enabling various networks to train. However, normalizing along the batch dimension introduces problems --- BN's error increases rapidly when the batch size becomes...

GSPMD: General and Scalable Parallelization for ML Computation Graphs

We present GSPMD, an automatic, compiler-based parallelization system for common machine learning computations. It allows users to write programs in the same way as for a single device, then give hints through a few annotations on how to distribute t...

Implementing Neural Turing Machines

Neural Turing Machines (NTMs) are an instance of Memory Augmented Neural Networks, a new class of recurrent neural networks which decouple computation from memory by introducing an external memory unit. NTMs have demonstrated superior performance ove...

Implicit Reparameterization Gradients.

By providing a simple and efficient way of computing low-variance gradients of continuous random variables, the reparameterization trick has become the technique of choice for training a variety of latent variable models. However, it is not applicabl...

Improving neural networks by preventing co-adaptation of feature detectors

When a large feedforward neural network is trained on a small training set, it typically performs poorly on held-out test data. This "overfitting" is greatly reduced by randomly omitting half of the feature detectors on each training case. This preve...

Improving the efficiency of forward-backward algorithm using batched computation in TensorFlow.

Sequence-level losses are commonly used to train deep neural network acoustic models for automatic speech recognition. The forward-backward algorithm is used to efficiently compute the gradients of the sequence loss with respect to the model paramete...

Incorporating Nesterov Momentum into

Cognition and behavior exhibit biases consistent with future expectations, and some of these biases result in momentum-like effects and have been linked with the idea of momentum. These momentum-like effects include representational momentum, operati...

IndyLSTMs: Independently Recurrent LSTMs

We introduce Independently Recurrent Long Short-term Memory cells: IndyLSTMs. These differ from regular LSTM cells in that the recurrent weights are not modeled as a full matrix, but as a diagonal matrix, i.e.\ the output and state of each LSTM cell ...

Instance Normalization: The Missing Ingredient for Fast Stylization

It this paper we revisit the fast stylization method introduced in Ulyanov et. al. (2016). We show how a small change in the stylization architecture results in a significant qualitative improvement in the generated images. The change is limited to s...

Language Modeling with Gated Convolutional Networks

The pre-dominant approach to language modeling to date is based on recurrent neural networks. Their success on this task is often linked to their ability to capture unbounded context. In this paper we develop a finite context approach through stacked...

Layer Normalization

Training state-of-the-art, deep neural networks is computationally expensive. One way to reduce the training time is to normalize the activities of the neurons. A recently introduced technique called batch normalization uses the distribution of the s...

Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation.

Kyunghyun Cho, Bart van Merriënboer, Caglar Gulcehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, Yoshua Bengio. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2014.

Learning Precise Timing with LSTM Recurrent Networks.

In response to Rodriguez's recent article (2001), we compare the performance of simple recurrent nets and long short-term memory recurrent nets on context-free and context-sensitive languages.

Learning to forget: continual prediction with LSTM

Long short-term memory (LSTM) can solve many tasks not solvable by previous learning algorithms for recurrent neural networks (RNNs). We identify a weakness of LSTM networks processing continual input streams without explicitly marked sequence ends. ...

Leveraging the bfloat16 Artificial Intelligence Datatype For Higher-Precision Computations

In recent years fused-multiply-add (FMA) units with lower-precision multiplications and higher-precision accumulation have proven useful in machine learning/artificial intelligence applications, most notably in training deep neural networks due to th...

Long Short-Term Memory.

Learning to store information over extended time intervals by recurrent backpropagation takes a very long time, mostly because of insufficient, decaying error backflow. We briefly review Hochreiter's (1991) analysis of this problem, then address it b...

Long Short-Term Memory Based Recurrent Neural Network Architectures for Large Vocabulary Speech Recognition

Long Short-Term Memory (LSTM) is a recurrent neural network (RNN) architecture that has been designed to address the vanishing and exploding gradient problems of conventional RNNs. Unlike feedforward neural networks, RNNs have cyclic connections maki...

Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism

Recent work in language modeling demonstrates that training large transformer models advances the state of the art in Natural Language Processing applications. However, very large models can be quite difficult to train due to memory constraints. In t...

Mish: A Self Regularized Non-Monotonic Activation Function

We propose $\textit{Mish}$, a novel self-regularized non-monotonic activation function which can be mathematically defined as: $f(x)=x\tanh(softplus(x))$. As activation functions play a crucial role in the performance and training dynamics in neural ...

MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices

Natural Language Processing (NLP) has recently achieved great success by using huge pre-trained models with hundreds of millions of parameters. However, these models suffer from heavy model sizes and high latency such that they cannot be deployed to ...

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

We present a class of efficient models called MobileNets for mobile and embedded vision applications. MobileNets are based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks. We introd...

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and re...

MoViNets: Mobile Video Networks for Efficient Video Recognition

We present Mobile Video Networks (MoViNets), a family of computation and memory efficient video networks that can operate on streaming video for online inference. 3D convolutional neural networks (CNNs) are accurate at video recognition but require l...

Multi-Scale Context Aggregation by Dilated Convolutions

State-of-the-art models for semantic segmentation are based on adaptations of convolutional networks that had originally been designed for image classification. However, dense prediction and image classification are structurally different. In this wo...

Muon is Scalable for LLM Training

Recently, the Muon optimizer based on matrix orthogonalization has demonstrated strong results in training small-scale language models, but the scalability to larger models has not been proven. We identify two crucial techniques for scaling up Muon: ...

NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications

This work proposes an algorithm, called NetAdapt, that automatically adapts a pre-trained deep neural network to a mobile platform given a resource budget. While many existing algorithms simplify networks based on the number of MACs or weights, optim...

Neural Optimizer Search with Reinforcement Learning.

We present an approach to automate the process of discovering optimization methods, with a focus on deep learning architectures. We train a Recurrent Neural Network controller to generate a string in a domain specific language that describes a mathem...

Noise-contrastive estimation: A new estimation principle for unnormalized statistical models.

We address the problem of articulated 2D human pose estimation in natural images. A well-known person detector - the Implicit Shape Model (ISM) approach introduced by Leibe et al. - is shown not only to be well suited to detect persons, but can also ...

Nonmetric Multidimensional Scaling: A Numerical Method

We describe the numerical methods required in our approach to multi-dimensional scaling. The rationale of this approach has appeared previously.

On-Device Neural Net Inference with Mobile GPUs

On-device inference of machine learning models for mobile phones is desirable due to its lower latency and increased privacy. Running such a compute-intensive task solely on the mobile CPU, however, can be difficult due to limited computing power, th...

On Layer Normalization in the Transformer Architecture

The Transformer is widely used in natural language processing tasks. To train a Transformer however, one usually needs a carefully designed learning rate warm-up stage, which is shown to be crucial to the final performance but will slow down the opti...

On Multiplicative Integration with Recurrent Neural Networks

We introduce a general and simple structural design called Multiplicative Integration (MI) to improve recurrent neural networks (RNNs). MI changes the way in which information from difference sources flows and is integrated in the computational build...

On the Computation of Complex-valued Gradients with Application to Statistically Optimum Beamforming

This report describes the computation of gradients by algorithmic differentiation for statistically optimum beamforming operations. Especially the derivation of complex-valued functions is a key component of this approach. Therefore the real-valued a...

On the Convergence of Adam and Beyond.

On the difficulty of training Recurrent Neural Networks

There are two widely known issues with properly training Recurrent Neural Networks, the vanishing and the exploding gradient problems detailed in Bengio et al. (1994). In this paper we attempt to improve the understanding of the underlying issues by ...

On the importance of initialization and momentum in deep learning.

On the Variance of the Adaptive Learning Rate and Beyond

The learning rate warmup heuristic achieves remarkable success in stabilizing training, accelerating convergence and improving generalization for adaptive stochastic optimization algorithms like RMSprop and Adam. Here, we study its mechanism in detai...

On Using Very Large Target Vocabulary for Neural Machine Translation

Neural machine translation, a recently proposed approach to machine translation based purely on neural networks, has shown promising results compared to the existing approaches such as phrase-based statistical machine translation. Despite its recent ...

Optimization of Collective Communication Operations in MPICH.

We describe our work on improving the performance of collective communication operations in MPICH for clusters connected by switched networks. For each collective operation, we use multiple algorithms depending on the message size, with the goal of m...

OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks.

We propose a deep convolutional neural network architecture codenamed "Inception", which was responsible for setting the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC 2014)...

Parallel WaveNet: Fast High-Fidelity Speech Synthesis

The recently-developed WaveNet architecture is the current state of the art in realistic speech synthesis, consistently rated as more natural sounding for many different languages than any previous system. However, because WaveNet relies on sequentia...

PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model

We present a box-free bottom-up approach for the tasks of pose estimation and instance segmentation of people in multi-person images using an efficient single-shot model. The proposed PersonLab model tackles both semantic-level reasoning and object-p...

PipeDream: Fast and Efficient Pipeline Parallel DNN Training

PipeDream is a Deep Neural Network(DNN) training system for GPUs that parallelizes computation by pipelining execution across multiple machines. Its pipeline parallel computing model avoids the slowdowns faced by data-parallel training when large mod...

PowerSGD: Practical Low-Rank Gradient Compression for Distributed Optimization

We study gradient compression methods to alleviate the communication bottleneck in data-parallel distributed optimization. Despite the significant attention received, current compression schemes either do not scale well or fail to achieve the target ...

ProjectionNet: Learning Efficient On-Device Deep Networks Using Neural Projections

Deep neural networks have become ubiquitous for applications related to visual recognition and language understanding tasks. However, it is often prohibitive to use typical neural networks on devices like mobile phones or smart watches since the mode...

Proximal Stochastic Dual Coordinate Ascent

We introduce a proximal version of dual coordinate ascent method. We demonstrate how the derived algorithmic framework can be used for numerous regularized loss minimization problems, including $\ell_1$ regularization and structured output SVM. The c...

QR and LQ Decomposition Matrix Backpropagation Algorithms for Square, Wide, and Deep -- Real or Complex -- Matrices and Their Software Implementation

This article presents matrix backpropagation algorithms for the QR decomposition of matrices $A_{m, n}$, that are either square (m = n), wide (m < n), or deep (m > n), with rank $k = min(m, n)$. Furthermore, we derive novel matrix backpropagation res...

Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference

The rising popularity of intelligent mobile devices and the daunting computational cost of deep learning-based models call for efficient and accurate on-device inference schemes. We propose a quantization scheme that allows inference to be carried ou...

Quantizing deep convolutional networks for efficient inference: A whitepaper

We present an overview of techniques for quantizing convolutional neural networks for inference with integer weights and activations. Per-channel quantization of weights and per-layer quantization of activations to 8-bits of precision post-training p...

Random Walk Initialization for Training Very Deep Feedforward Networks

Training very deep networks is an important open problem in machine learning. One of many difficulties is that the norm of the back-propagated error gradient can grow or decay exponentially. Here we show that training very deep feed-forward networks ...

Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network

Recently, several models based on deep neural networks have achieved great success in terms of both reconstruction accuracy and computational performance for single image super-resolution. In these methods, the low resolution (LR) input image is upsc...

Reconstruction filters in computer-graphics

Problems of signal processing arise in image synthesis because of transformations between continuous and discrete representations of 2D images. Aliasing introduced by sampling has received much attention in graphics, but reconstruction of samples int...

Rectifier Nonlinearities Improve Neural Network Acoustic Models

YouTube is a highly visited video sharing website where over one billion people watch six billion hours of video every month. Improving accessibility to these videos for the hearing impaired and for search and indexing purposes is an excellent applic...

Recurrent Neural Network Regularization

We present a simple regularization technique for Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units. Dropout, the most successful technique for regularizing neural networks, does not work well with RNNs and LSTMs. In this paper...

Reducing Activation Recomputation in Large Transformer Models

Training large transformer models is one of the most important computational challenges of modern AI. In this paper, we show how to significantly accelerate training of large transformer models by reducing activation recomputation. Activation recompu...

Reliable and fast DWARF-based stack unwinding

Debug information, usually encoded in the DWARF format, is a hidden and obscure component of our computing infrastructure. Debug information is obviously used by debuggers, but it also plays a key role in program analysis tools, and, most surprisingl...

Rethinking the Inception Architecture for Computer Vision

Convolutional networks are at the core of most state-of-the-art computer vision solutions for a wide variety of tasks. Since 2014 very deep convolutional networks started to become mainstream, yielding substantial gains in various benchmarks. Althoug...

Revisiting ResNets: Improved Training and Scaling Strategies

Novel computer vision architectures monopolize the spotlight, but the impact of the model architecture is often conflated with simultaneous changes to training methodology and scaling strategies. Our work revisits the canonical ResNet (He et al., 201...

RoFormer: Enhanced Transformer with Rotary Position Embedding

Position encoding recently has shown effective in the transformer architecture. It enables valuable supervision for dependency modeling between elements at different positions of the sequence. In this paper, we first investigate various methods to in...

Root Mean Square Layer Normalization

Layer normalization (LayerNorm) has been successfully applied to various deep neural networks to help stabilize training and boost model convergence because of its capability in handling re-centering and re-scaling of both inputs and weight matrix. H...

Scalable Object Detection using Deep Neural Networks

Deep convolutional neural networks have recently achieved state-of-the-art performance on a number of image recognition benchmarks, including the ImageNet Large-Scale Visual Recognition Challenge (ILSVRC-2012). The winning model on the localization s...

Searching for Activation Functions

The choice of activation functions in deep networks has a significant effect on the training dynamics and task performance. Currently, the most successful and widely-used activation function is the Rectified Linear Unit (ReLU). Although various hand-...

Searching for MobileNetV3

We present the next generation of MobileNets based on a combination of complementary search techniques as well as a novel architecture design. MobileNetV3 is tuned to mobile phone CPUs through a combination of hardware-aware network architecture sear...

Self-Normalizing Neural Networks.

The Internet of Things (IoT) gains momentum. Developments regarding smart grids, intelligent transportation systems, and low-power networks for smart cities constitute significant drivers in the evolution of network industries. IoT creates an array o...

Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs

Deep Convolutional Neural Networks (DCNNs) have recently shown state of the art performance in high level vision tasks, such as image classification and object detection. This work brings together methods from DCNNs and probabilistic graphical models...

SGDR: Stochastic Gradient Descent with Warm Restarts.

In this paper, we describe a phenomenon, which we named "super-convergence",\nwhere neural networks can be trained an order of magnitude faster than with\nstandard training methods. The existence of super-convergence is relevant to\nunderstanding why...

Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learning

In recent years, neural networks have enjoyed a renaissance as function approximators in reinforcement learning. Two decades after Tesauro's TD-Gammon achieved near top-level human performance in backgammon, the deep reinforcement learning algorithm ...

Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity. If an AI system learned such a deceptive strategy,...

Soft-NMS -- Improving Object Detection With One Line of Code

Non-maximum suppression is an integral part of the object detection pipeline. First, it sorts all detection boxes on the basis of their scores. The detection box M with the maximum score is selected and all other detection boxes with a significant ov...

Some windows with very good sidelobe behavior

Some of the windows presented by Harris [1] are not correct in terms of their reported peak sidelobes and optimal behavior. We present corrected plots of Harris' windows and also derive additional windows with very good sidelobes and optimal behavior...

Sparse matrix multiplication package (SMMP)

Spatial Transformer Networks

Convolutional Neural Networks define an exceptionally powerful class of models, but are still limited by the lack of ability to be spatially invariant to the input data in a computationally and parameter efficient manner. In this work we introduce a ...

Spectral Normalization for Generative Adversarial Networks

One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. O...

SSD: Single Shot MultiBox Detector

We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map locati...

Stochastic Dual Coordinate Ascent Methods for Regularized Loss Minimization

Stochastic Gradient Descent (SGD) has become popular for solving large scale supervised machine learning optimization problems such as SVM, due to their strong theoretical guarantees. While the closely related Dual Coordinate Ascent (DCA) method has ...

Stochastic Dual Coordinate Ascent with Adaptive Probabilities

This paper introduces AdaSDCA: an adaptive variant of stochastic dual coordinate ascent (SDCA) for solving the regularized empirical risk minimization problems. Our modification consists in allowing the method adaptively change the probability distri...

Stochastic Weight Averaging in Parallel: Large-Batch Training that Generalizes Well

We propose Stochastic Weight Averaging in Parallel (SWAP), an algorithm to accelerate DNN training. Our algorithm uses large mini-batches to compute an approximate solution quickly and then refines it by averaging the weights of multiple models compu...

Super-Convergence: Very Fast Training of Neural Networks Using Large Learning Rates

In this paper, we describe a phenomenon, which we named "super-convergence", where neural networks can be trained an order of magnitude faster than with standard training methods. The existence of super-convergence is relevant to understanding why de...

SWALP : Stochastic Weight Averaging in Low-Precision Training

Low precision operations can provide scalability, memory savings, portability, and energy efficiency. This paper proposes SWALP, an approach to low precision training that averages low-precision SGD iterates with a modified learning rate schedule. SW...

TensorFlow

TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries, and community resources that lets researchers push the state-of-the-art in ML and developers easily build and deplo...

TensorFlow Distributions

The TensorFlow Distributions library implements a vision of probability theory adapted to the modern deep-learning paradigm of end-to-end differentiable computation. Building on two basic abstractions, it offers flexible building blocks for probabili...

The Complex Gradient Operator and the CR-Calculus

A thorough discussion and development of the calculus of real-valued functions of complex-valued vectors is given using the framework of the Wirtinger Calculus. The presented material is suitable for exposition in an introductory Electrical Engineeri...

The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables

The reparameterization trick enables optimizing large scale stochastic computation graphs via gradient descent. The essence of the trick is to refactor each stochastic node into a differentiable function of its parameters and a random variable with f...

The continuous Bernoulli: fixing a pervasive error in variational autoencoders

Variational autoencoders (VAE) have quickly become a central tool in machine learning, applicable to a broad range of data types and latent variable models. By far the most common first step, taken by seminal papers and by core software libraries ali...

The differentiation of pseudo-inverses and non-linear least squares problems whose variables separate

For given data $(t_i ,y_i ),i = 1, \cdots ,m$, we consider the least squares fit of nonlinear models of the form \[ \eta ({\bf a},{\boldsymbol \alpha} ;t) = \sum _{j = 1}^n {a_j \varphi _j ({\boldsymbol \alpha} ;t),\qquad {\bf a} \in \mathcal{R}^n ,\...

There Are Many Consistent Explanations of Unlabeled Data: Why You Should Average

Presently the most successful approaches to semi-supervised learning are based on consistency regularization, whereby a model is trained to be robust to small perturbations of its inputs and parameters. To understand consistency regularization, we co...

The relationship between Precision-Recall and ROC curves.

Receiver Operator Characteristic (ROC) curves are commonly used to present results for binary decision problems in machine learning. However, when dealing with highly skewed datasets, Precision-Recall (PR) curves give a more informative picture of an...

Toeplitz And Circulant Matrices: A Review (Foundations and Trends(R) in Communications and Information Theory)

To prune, or not to prune: exploring the efficacy of pruning for model compression

Model pruning seeks to induce sparsity in a deep neural network's various connection matrices, thereby reducing the number of nonzero-valued parameters in the model. Recent reports (Han et al., 2015; Narang et al., 2017) prune deep networks at the co...

Toward the Optimal Preconditioned Eigensolver: Locally Optimal Block Preconditioned Conjugate Gradient Method.

We describe new algorithms of the locally optimal block preconditioned conjugate gradient (LOBPCG) method for symmetric eigenvalue problems, based on a local optimization of a three-term recurrence, and suggest several other new methods. To be able t...

TPU-KNN: K Nearest Neighbor Search at Peak FLOP/s

This paper presents a novel nearest neighbor search algorithm achieving TPU (Google Tensor Processing Unit) peak performance, outperforming state-of-the-art GPU algorithms with similar level of recall. The design of the proposed algorithm is motivate...

Trained Quantization Thresholds for Accurate and Efficient Fixed-Point Inference of Deep Neural Networks

We propose a method of training quantization thresholds (TQT) for uniform symmetric quantizers using standard backpropagation and gradient descent. Contrary to prior work, we show that a careful analysis of the straight-through estimator for threshol...

Training Deep Networks with Structured Layers by Matrix Backpropagation

Deep neural network architectures have recently produced excellent results in a variety of areas in artificial intelligence and visual recognition, well surpassing traditional shallow architectures trained using hand-designed features. The power of d...

Trivializations for Gradient-Based Optimization on Manifolds

We introduce a framework to study the transformation of problems with manifold constraints into unconstrained problems through parametrizations in terms of a Euclidean space. We call these parametrizations "trivializations". We prove conditions under...

Understanding and Improving Convolutional Neural Networks via Concatenated Rectified Linear Units

Recently, convolutional neural networks (CNNs) have been used as a powerful tool to solve many problems of machine learning and computer vision. In this paper, we aim to provide insight on the property of convolutional neural networks, as well as a g...

Understanding the difficulty of training deep feedforward neural networks.

Cellular Neural Networks (CNN) [1] main assets are quoted to be their capacity for parallel hardware implementation and their universality. On top, the possibility to add the information of a local sensor on every cell, provides a unique system for m...

Up or Down? Adaptive Rounding for Post-Training Quantization

When quantizing neural networks, assigning each floating-point weight to its nearest fixed-point value is the predominant approach. We find that, perhaps surprisingly, this is not the best we can do. In this paper, we propose AdaRound, a better weigh...

Wav2Letter: an End-to-End ConvNet-based Speech Recognition System

This paper presents a simple end-to-end model for speech recognition, combining a convolutional network based acoustic model and a graph decoding. It is trained to output letters, with transcribed speech, without the need for force alignment of phone...

WaveNet: A Generative Model for Raw Audio

This paper introduces WaveNet, a deep neural network for generating raw audio waveforms. The model is fully probabilistic and autoregressive, with the predictive distribution for each audio sample conditioned on all previous ones; nonetheless we show...

Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks

We present weight normalization: a reparameterization of the weight vectors in a neural network that decouples the length of those weight vectors from their direction. By reparameterizing the weights in this way we improve the conditioning of the opt...

Zero Bubble Pipeline Parallelism

Pipeline parallelism is one of the key components for large-scale distributed training, yet its efficiency suffers from pipeline bubbles which were deemed inevitable. In this work, we introduce a scheduling strategy that, to our knowledge, is the fir...

ZeRO: Memory Optimizations Toward Training Trillion Parameter Models

Large deep learning models offer significant accuracy gains, but training billions to trillions of parameters is challenging. Existing solutions such as data and model parallelisms exhibit fundamental limitations to fit these models into limited devi...