开始使用免费开始使用

首页
课程
Production Machine Learning Systems

Why distributed training is needed

Create Your Free Account

or

Email Address

Password

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.

本练习是课程的一部分

Production Machine Learning Systems

SkillTag.level.intermediateSkillTag.label

5.0+

免费开始课程

This module previews the topics covered in the course and how to use Qwiklabs to complete each of your labs using Google Cloud.

Exercise 1: Advanced Machine Learning on Google Cloud Exercise 2: Welcome

This module explores what else a production ML system needs to do and how to meet those needs. You review how to make important, high-level, design decisions around training and model serving need to make in order to get the right performance profile for your model.

Exercise 1: Architecting ML systems Exercise 2: Data extraction, analysis, and preparation Exercise 3: Model training, evaluation, and validation Exercise 4: Trained model, prediction service, and performance monitoring Exercise 5: Training design decisions Exercise 6: Serving design decisions Exercise 7: Designing from scratch Exercise 8: Using Vertex AI Exercise 9: Lab introduction: Structured data prediction Exercise 10: Architecting Production ML Systems Question 1 Exercise 11: Architecting Production ML Systems Question 2 Exercise 12: Architecting Production ML Systems Question 3 Exercise 13: Architecting Production ML Systems Question 4 Exercise 14: Architecting Production ML Systems Question 5 Exercise 15: Architecting Production ML Systems Question 6 Exercise 16: Architecting Production ML Systems Question 7 Exercise 17: Architecting Production ML Systems Question 8 Exercise 18: Readings: Architecting production ML systems

In this module, you learn how to recognize the ways that our model is dependent on our data, make cost-conscious engineering decisions, know when to roll back our models to earlier versions, debug the causes of observed model behavior and implement a pipeline that is immune to one type of dependency.

Exercise 1: Introduction Exercise 2: Adapting to data Exercise 3: Changing distributions Exercise 4: Lab: Adapting to data Exercise 5: Right and wrong decisions Exercise 6: System failure Exercise 7: Concept drift Exercise 8: Actions to mitigate concept drift Exercise 9: TensorFlow data validation Exercise 10: Components of TensorFlow data validation Exercise 11: Lab Introduction: Introduction to TensorFlow Data Validation Exercise 12: Introduction to TensorFlow Data Validation Exercise 13: Lab Introduction: Advanced Visualizations with TensorFlow Data Validation Exercise 14: Advanced Visualizations with TensorFlow Data Validation Exercise 15: Mitigating training-serving skew through design Exercise 16: Agent Platform: Training and Serving a Custom Model Exercise 17: Diagnosing a production model Exercise 18: Designing Adaptable ML Systems Question 1 Exercise 19: Designing Adaptable ML Systems Question 2 Exercise 20: Designing Adaptable ML Systems Question 3 Exercise 21: Designing Adaptable ML Systems Question 4 Exercise 22: Designing Adaptable ML Systems Question 5 Exercise 23: Designing Adaptable ML Systems Question 6 Exercise 24: Designing Adaptable ML Systems Question 7 Exercise 25: Readings: Designing adaptable ML systems

In this module, you identify performance considerations for machine learning models.Machine learning models are not all identical. For some models, you focus on improving I/O performance, and on others, you focus on squeezing out more computational speed.

Exercise 1: Introduction Exercise 2: Training Exercise 3: Predictions Exercise 4: Why distributed training is needed

当前练习

Exercise 5: Distributed training architectures Exercise 6: TensorFlow distributed training strategies Exercise 7: Mirrored strategy Exercise 8: Multi-worker mirrored strategy Exercise 9: TPU strategy Exercise 10: Parameter server strategy Exercise 11: Lab Introduction: Distributed Training with Keras Exercise 12: Distributed Training with Keras Exercise 13: Training on large datasets with tf.data API Exercise 14: Lab Introduction: TPU-speed Data Pipelines Exercise 15: Inference Exercise 16: Designing High-Performance ML Systems Question 1 Exercise 17: Designing High-Performance ML Systems Question 2 Exercise 18: Designing High-Performance ML Systems Question 3 Exercise 19: Designing High-Performance ML Systems Question 4 Exercise 20: Readings: Designing high-performance ML systems

Understand the tools and systems available and when to leverage hybrid machine learning models.

Exercise 1: Introduction Exercise 2: Machine Learning on Hybrid Cloud Exercise 3: Kubeflow Exercise 4: Lab Introduction: Kubeflow Pipelines with AI Platform Exercise 5: Running Pipelines on Agent Platform 2.5 Exercise 6: TensorFlow Lite Exercise 7: Optimizing TensorFlow for mobile Exercise 8: Summary Exercise 9: Building Hybrid ML Systems Question 1 Exercise 10: Building Hybrid ML Systems Question 2 Exercise 11: Building Hybrid ML Systems Question 3 Exercise 12: Building Hybrid ML Systems Question 4 Exercise 13: Readings: Hybrid ML systems

PDF links to all modules

Exercise 1: Course summary Exercise 2: Production Machine learning systems - readings Exercise 3: All quiz questions and answers

PDF links to all modules

Exercise 1: Architecting Production ML Systems Course Resources