https://www.chenshaowen.com/en/tags/training/Recent content in Training on Shaowen Chen's WebsiteHugo -- gohugo.ioen©2016 - {year}, All Rights Reserved.Sun, 09 Feb 2025 10:00:00 +0000weeklyhttps://www.chenshaowen.com/en/blog/what-is-ray.htmlSun, 09 Feb 2025 10:00:00 +0000Sun, 09 Feb 2025 10:00:00 +0000https://www.chenshaowen.com/en/blog/what-is-ray.html1. What Ray Is In 2016, UC Berkeley’s RISELab released a new distributed computing framework called Ray. In 2017, after the Ray paper was published, it drew broad attention across the industry; in China it was mainly Ant Group that adopted and contributed to Ray. In 2020, Ray released version微信公众号Distributed SystemsComputing FrameworkRayTrainingInferencehttps://www.chenshaowen.com/en/blog/using-tensorboard-to-visualize-pytorch-training-process.htmlSun, 17 Nov 2024 00:00:00 +0000Sun, 17 Nov 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/using-tensorboard-to-visualize-pytorch-training-process.html1. What TensorBoard Is TensorBoard is mainly used to monitor how a model’s various metrics change — accuracy, loss, the weight distributions of each layer, and so on. TensorBoard is a visualization tool from TensorFlow that supports visualizing scalar, text, image, audio, video, and embedding data, among other types. But PyTorch can use TensorBoard too.微信公众号AIPyTorchTensorBoardTrainingR&Dhttps://www.chenshaowen.com/en/blog/using-pytorch-to-train-model-on-mnist-dataset.htmlSat, 16 Nov 2024 00:00:00 +0000Sat, 16 Nov 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/using-pytorch-to-train-model-on-mnist-dataset.html1. Create the Training Script Create a training script mnist.py with the following contents: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36微信公众号AIPyTorchTrainingContainershttps://www.chenshaowen.com/en/blog/kubernetes-dlrover-workflow-analysis.htmlTue, 27 Aug 2024 00:00:00 +0000Tue, 27 Aug 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/kubernetes-dlrover-workflow-analysis.htmlThe DLRover version used in this article is 0.3.7 1. DLRover Operator 1.1 Starting the ElasticJob and ScalePlan Controllers Implementation code: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 // 创建 ElasticJob 的控制器 if err = controllers.NewElasticJobReconciler(mgr, masterImage).SetupWithManager(mgr); err微信公众号DLRoverKubernetesAITrainingFault Self-HealingOperationshttps://www.chenshaowen.com/en/blog/data-parallel-architecture-in-distributed-train.htmlWed, 21 Aug 2024 00:00:00 +0000Wed, 21 Aug 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/data-parallel-architecture-in-distributed-train.html1. Parameter Server Architecture In the Parameter Server architecture, the nodes in the cluster are divided into two categories: parameter server nodes (Parameter Server) and worker server nodes (Worker). 1.1 Parameter Server The Parameter Server is used to store the model’s parameters. Each parameter server node is responsible for managing and updating one part of the model’s parameters, while each worker node only processes the subset of data corresponding to it.微信公众号Distributed SystemsTrainingData ParallelismAIOperationshttps://www.chenshaowen.com/en/blog/use-dlrover-to-manage-training-job.htmlSat, 17 Aug 2024 00:00:00 +0000Sat, 17 Aug 2024 00:00:00 +0000https://www.chenshaowen.com/en/blog/use-dlrover-to-manage-training-job.html1. Problems Facing Distributed Training Estimating training resources is difficult and cannot be automated How much compute, how much time, how much bandwidth, how many CPUs, how much memory — without enough accumulated experience it is hard to estimate accurately. The result is over-requesting and over-allocation, causing enormous resource waste.微信公众号DLRoverAITrainingKubernetesElastic TrainingFault-Tolerant TrainingOperationshttps://www.chenshaowen.com/en/blog/interviewer-training-about-how-to-choose-right-people.htmlTue, 18 Jan 2022 00:00:00 +0000Tue, 18 Jan 2022 00:00:00 +0000https://www.chenshaowen.com/en/blog/interviewer-training-about-how-to-choose-right-people.html1. Interview Process Establish hiring criteria Align on hiring criteria Resume analysis Opening small talk Competency assessment Values assessment Verifying authenticity Motivation assessment Attracting strong candidates Closing the interview Filling out the interview evaluation form Assisting with onboarding follow-up 2. Before the Interview 2.1 Establish Hiring Criteria why Evaluate candidates with a unified assessment standard微信公众号InterviewTrainingManagementTeamsRecruitingTalentLearningHuman Resources