Qiang Wang

Bo Qin, Aixin Jia, Qiang Wang et al.

This paper describes the submission of the RoyalFlush neural machine translation system for the WMT 2022 translation efficiency task. Unlike the commonly used autoregressive translation system, we adopted a two-stage translation paradigm called Hybrid Regression Translation (HRT) to combine the advantages of autoregressive and non-autoregressive translation. Specifically, HRT first autoregressively generates a discontinuous sequence (e.g., make a prediction every $k$ tokens, $k>1$) and then fills in all previously skipped tokens at once in a non-autoregressive manner. Thus, we can easily trade off the translation quality and speed by adjusting $k$. In addition, by integrating other modeling techniques (e.g., sequence-level knowledge distillation and deep-encoder-shallow-decoder layer allocation strategy) and a mass of engineering efforts, HRT improves 80\% inference speed and achieves equivalent translation performance with the same-capacity AT counterpart. Our fastest system reaches 6k+ words/second on the GPU latency setting, estimated to be about 3.1x faster than the last year's winner.

Qiang Wang, Shaohuai Shi, Shizhen Zheng et al.

Deep neural networks (DNNs) have achieved great success in the area of computer vision. The disparity estimation problem tends to be addressed by DNNs which achieve much better prediction accuracy in stereo matching than traditional hand-crafted feature based methods. On one hand, however, the designed DNNs require significant memory and computation resources to accurately predict the disparity, especially for those 3D convolution based networks, which makes it difficult for deployment in real-time applications. On the other hand, existing computation-efficient networks lack expression capability in large-scale datasets so that they cannot make an accurate prediction in many scenarios. To this end, we propose an efficient and accurate deep network for disparity estimation named FADNet with three main features: 1) It exploits efficient 2D based correlation layers with stacked blocks to preserve fast computation; 2) It combines the residual structures to make the deeper model easier to learn; 3) It contains multi-scale predictions so as to exploit a multi-scale weight scheduling training technique to improve the accuracy. We conduct experiments to demonstrate the effectiveness of FADNet on two popular datasets, Scene Flow and KITTI 2015. Experimental results show that FADNet achieves state-of-the-art prediction accuracy, and runs at a significant order of magnitude faster speed than existing 3D models. The codes of FADNet are available at https://github.com/HKBU-HPML/FADNet.

Zhenheng Tang, Yuxin Wang, Xin He et al.

The rapid growth of memory and computation requirements of large language models (LLMs) has outpaced the development of hardware, hindering people who lack large-scale high-end GPUs from training or deploying LLMs. However, consumer-level GPUs, which constitute a larger market share, are typically overlooked in LLM due to their weaker computing performance, smaller storage capacity, and lower communication bandwidth. Additionally, users may have privacy concerns when interacting with remote LLMs. In this paper, we envision a decentralized system unlocking the potential vast untapped consumer-level GPUs in pre-training, inference and fine-tuning of LLMs with privacy protection. However, this system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity. To address these challenges, our system design incorporates: 1) a broker with backup pool to implement dynamic join and quit of computing providers; 2) task scheduling with hardware performance to improve system efficiency; 3) abstracting ML procedures into directed acyclic graphs (DAGs) to achieve model and task universality; 4) abstracting intermediate represention and execution planes to ensure compatibility of various devices and deep learning (DL) frameworks. Our performance analysis demonstrates that 50 RTX 3080 GPUs can achieve throughputs comparable to those of 4 H100 GPUs, which are significantly more expensive.

Qiang Wang, Changliang Li, Yue Zhang et al.

Traditional neural machine translation is limited to the topmost encoder layer's context representation and cannot directly perceive the lower encoder layers. Existing solutions usually rely on the adjustment of network architecture, making the calculation more complicated or introducing additional structural restrictions. In this work, we propose layer-wise multi-view learning to solve this problem, circumventing the necessity to change the model structure. We regard each encoder layer's off-the-shelf output, a by-product in layer-by-layer encoding, as the redundant view for the input sentence. In this way, in addition to the topmost encoder layer (referred to as the primary view), we also incorporate an intermediate encoder layer as the auxiliary view. We feed the two views to a partially shared decoder to maintain independent prediction. Consistency regularization based on KL divergence is used to encourage the two views to learn from each other. Extensive experimental results on five translation tasks show that our approach yields stable improvements over multiple strong baselines. As another bonus, our method is agnostic to network architectures and can maintain the same inference speed as the original model.

Shaohuai Shi, Zhenheng Tang, Qiang Wang et al.

To reduce the long training time of large deep neural network (DNN) models, distributed synchronous stochastic gradient descent (S-SGD) is commonly used on a cluster of workers. However, the speedup brought by multiple workers is limited by the communication overhead. Two approaches, namely pipelining and gradient sparsification, have been separately proposed to alleviate the impact of communication overheads. Yet, the gradient sparsification methods can only initiate the communication after the backpropagation, and hence miss the pipelining opportunity. In this paper, we propose a new distributed optimization method named LAGS-SGD, which combines S-SGD with a novel layer-wise adaptive gradient sparsification (LAGS) scheme. In LAGS-SGD, every worker selects a small set of "significant" gradients from each layer independently whose size can be adaptive to the communication-to-computation ratio of that layer. The layer-wise nature of LAGS-SGD opens the opportunity of overlapping communications with computations, while the adaptive nature of LAGS-SGD makes it flexible to control the communication time. We prove that LAGS-SGD has convergence guarantees and it has the same order of convergence rate as vanilla S-SGD under a weak analytical assumption. Extensive experiments are conducted to verify the analytical assumption and the convergence performance of LAGS-SGD. Experimental results on a 16-GPU cluster show that LAGS-SGD outperforms the original S-SGD and existing sparsified S-SGD without losing obvious model accuracy.

Yuxin Wang, Qiang Wang, Shaohuai Shi et al.

Deep learning has become widely used in complex AI applications. Yet, training a deep neural network (DNNs) model requires a considerable amount of calculations, long running time, and much energy. Nowadays, many-core AI accelerators (e.g., GPUs and TPUs) are designed to improve the performance of AI training. However, processors from different vendors perform dissimilarly in terms of performance and energy consumption. To investigate the differences among several popular off-the-shelf processors (i.e., Intel CPU, NVIDIA GPU, AMD GPU, and Google TPU) in training DNNs, we carry out a comprehensive empirical study on the performance and energy efficiency of these processors by benchmarking a representative set of deep learning workloads, including computation-intensive operations, classical convolutional neural networks (CNNs), recurrent neural networks (LSTM), Deep Speech 2, and Transformer. Different from the existing end-to-end benchmarks which only present the training time, We try to investigate the impact of hardware, vendor's software library, and deep learning framework on the performance and energy consumption of AI training. Our evaluation methods and results not only provide an informative guide for end-users to select proper AI accelerators, but also expose some opportunities for the hardware vendors to improve their software library.

Shaohuai Shi, Qiang Wang, Kaiyong Zhao et al.

Distributed synchronous stochastic gradient descent (S-SGD) has been widely used in training large-scale deep neural networks (DNNs), but it typically requires very high communication bandwidth between computational workers (e.g., GPUs) to exchange gradients iteratively. Recently, Top-$k$ sparsification techniques have been proposed to reduce the volume of data to be exchanged among workers. Top-$k$ sparsification can zero-out a significant portion of gradients without impacting the model convergence. However, the sparse gradients should be transferred with their irregular indices, which makes the sparse gradients aggregation difficult. Current methods that use AllGather to accumulate the sparse gradients have a communication complexity of $O(kP)$, where $P$ is the number of workers, which is inefficient on low bandwidth networks with a large number of workers. We observe that not all top-$k$ gradients from $P$ workers are needed for the model update, and therefore we propose a novel global Top-$k$ (gTop-$k$) sparsification mechanism to address the problem. Specifically, we choose global top-$k$ largest absolute values of gradients from $P$ workers, instead of accumulating all local top-$k$ gradients to update the model in each iteration. The gradient aggregation method based on gTop-$k$ sparsification reduces the communication complexity from $O(kP)$ to $O(k\log P)$. Through extensive experiments on different DNNs, we verify that gTop-$k$ S-SGD has nearly consistent convergence performance with S-SGD, and it has only slight degradations on generalization performance. In terms of scaling efficiency, we evaluate gTop-$k$ on a cluster with 32 GPU machines which are interconnected with 1 Gbps Ethernet. The experimental results show that our method achieves $2.7-12\times$ higher scaling efficiency than S-SGD and $1.1-1.7\times$ improvement than the existing Top-$k$ S-SGD.

Ke Yang, Dongsheng Li, Yong Dou et al.

Object detection is an import task of computer vision.A variety of methods have been proposed,but methods using the weak labels still do not have a satisfactory result.In this paper,we propose a new framework that using the weakly supervised method's output as the pseudo-strong labels to train a strongly supervised model.One weakly supervised method is treated as black-box to generate class-specific bounding boxes on train dataset.A de-noise method is then applied to the noisy bounding boxes.Then the de-noised pseudo-strong labels are used to train a strongly object detection network.The whole framework is still weakly supervised because the entire process only uses the image-level labels.The experiment results on PASCAL VOC 2007 prove the validity of our framework, and we get result 43.4% on mean average precision compared to 39.5% of the previous best result and 34.5% of the initial method,respectively.And this frame work is simple and distinct,and is promising to be applied to other method easily.