اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدChimera: Efficiently Training Large-Scale Neural Networks with Bidirectional Pipelines
87
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Training large deep learning models at scale is very challenging. This paper proposes Chimera, a novel pipeline parallelism scheme which combines bidirectional pipelines for efficiently training large-scale models. Chimera is a synchronous approach and therefore no loss of accuracy, which is more convergence-friendly than asynchronous approaches. Compared with the latest synchronous pipeline approach, Chimera reduces the number of bubbles by up to 50%; benefiting from the sophisticated scheduling of bidirectional pipelines, Chimera has a more balanced activation memory consumption. Evaluations are conducted on Transformer based language models. For a GPT-2 model with 1.3 billion parameters running on 2,048 GPU nodes of the Piz Daint supercomputer, Chimera improves the training throughput by 1.16x-2.34x over the state-of-the-art synchronous and asynchronous pipeline approaches.
قيم البحث
اقرأ أيضاً
Recurrent Neural Network (RNN) applications form a major class of AI-powered, low-latency data center workloads. Most execution models for RNN acceleration break computation graphs into BLAS kernels, which lead to significant inter-kernel data moveme
nt and resource underutilization. We show that by supporting more general loop constructs that capture design parameters in accelerators, it is possible to improve resource utilization using cross-kernel optimization without sacrificing programmability. Such abstraction level enables a design space search that can lead to efficient usage of on-chip resources on a spatial architecture across a range of problem sizes. We evaluate our optimization strategy on such abstraction with DeepBench using a configurable spatial accelerator. We demonstrate that this implementation provides a geometric speedup of 30x in performance, 1.6x in area, and 2x in power efficiency compared to a Tesla V100 GPU, and a geometric speedup of 2x compared to Microsoft Brainwave implementation on a Stratix 10 FPGA.
Scale of data and scale of computation infrastructures together enable the current deep learning renaissance. However, training large-scale deep architectures demands both algorithmic improvement and careful system configuration. In this paper, we fo
cus on employing the system approach to speed up large-scale training. Via lessons learned from our routine benchmarking effort, we first identify bottlenecks and overheads that hinter data parallelism. We then devise guidelines that help practitioners to configure an effective system and fine-tune parameters to achieve desired speedup. Specifically, we develop a procedure for setting minibatch size and choosing computation algorithms. We also derive lemmas for determining the quantity of key components such as the number of GPUs and parameter servers. Experiments and examples show that these guidelines help effectively speed up large-scale deep learning training.
The recent Natural Language Processing techniques have been refreshing the state-of-the-art performance at an incredible speed. Training huge language models is therefore an imperative demand in both industry and academy. However, huge language model
s impose challenges to both hardware and software. Graphical processing units (GPUs) are iterated frequently to meet the exploding demand, and a variety of ASICs like TPUs are spawned. However, there is still a tension between the fast growth of the extremely huge models and the fact that Moores law is approaching the end. To this end, many model parallelism techniques are proposed to distribute the model parameters to multiple devices, so as to alleviate the tension on both memory and computation. Our work is the first to introduce a 3-dimensional model parallelism for expediting huge language models. By reaching a perfect load balance, our approach presents smaller memory and communication cost than existing state-of-the-art 1-D and 2-D model parallelism. Our experiments on 64 TACCs V100 GPUs show that our 3-D parallelism outperforms the 1-D and 2-D parallelism with 2.32x and 1.57x speedup, respectively.
Faced with continuously increasing scale of data, original back-propagation neural network based machine learning algorithm presents two non-trivial challenges: huge amount of data makes it difficult to maintain both efficiency and accuracy; redundan
t data aggravates the system workload. This project is mainly focused on the solution to the issues above, combining deep learning algorithm with cloud computing platform to deal with large-scale data. A MapReduce-based handwriting character recognizer will be designed in this project to verify the efficiency improvement this mechanism will achieve on training and practical large-scale data. Careful discussion and experiment will be developed to illustrate how deep learning algorithm works to train handwritten digits data, how MapReduce is implemented on deep learning neural network, and why this combination accelerates computation. Besides performance, the scalability and robustness will be mentioned in this report as well. Our system comes with two demonstration software that visually illustrates our handwritten digit recognition/encoding application.
Supercomputer FCFS-based scheduling policies result in many transient idle nodes, a phenomenon that is only partially alleviated by backfill scheduling methods that promote small jobs to run before large jobs. Here we describe how to realize a novel
use for these otherwise wasted resources, namely, deep neural network (DNN) training. This important workload is easily organized as many small fragments that can be configured dynamically to fit essentially any node*time hole in a supercomputers schedule. We describe how the task of rescaling suitable DNN training tasks to fit dynamically changing holes can be formulated as a deterministic mixed integer linear programming (MILP)-based resource allocation algorithm, and show that this MILP problem can be solved efficiently at run time. We show further how this MILP problem can be adapted to optimize for administrator- or user-defined metrics. We validate our method with supercomputer scheduler logs and different DNN training scenarios, and demonstrate efficiencies of up to 93% compared with running the same training tasks on dedicated nodes. Our method thus enables substantial supercomputer resources to be allocated to DNN training with no impact on other applications.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
