Get in Touch

Course Outline

Getting Started with Huawei CloudMatrix

  • The CloudMatrix ecosystem and its deployment architecture
  • Compatible models, file formats, and deployment strategies
  • Common use cases and supported hardware chipsets

Readying Models for Deployment

  • Exporting models from training environments (MindSpore, TensorFlow, PyTorch)
  • Leveraging ATC (Ascend Tensor Compiler) for format transformation
  • Differentiating between static and dynamic shape models

Deployment on CloudMatrix

  • Creating services and registering models
  • Launching inference services via the interface or command line
  • Managing routing, authentication, and access permissions

Handling Inference Requests

  • Distinguishing between batch and real-time inference processes
  • Implementing data preprocessing and postprocessing workflows
  • Interacting with CloudMatrix services from external applications

Monitoring and Performance Adjustment

  • Reviewing deployment logs and tracking requests
  • Implementing resource scaling and load balancing
  • Optimizing latency and throughput

Enterprise Tool Integration

  • Linking CloudMatrix with OBS and ModelArts
  • Employing workflows and managing model versions
  • Implementing CI/CD for model deployment and rollback capabilities

Complete Inference Pipeline

  • Deploying a full image classification pipeline
  • Benchmarking performance and verifying accuracy
  • Testing failover mechanisms and system alerts

Recap and Future Actions

Requirements

  • Familiarity with AI model training processes
  • Practical experience with Python-based machine learning frameworks
  • Fundamental knowledge of cloud deployment principles

Target Audience

  • AI operations teams
  • Machine learning engineers
  • Cloud deployment specialists operating within Huawei’s infrastructure
 21 Hours

Number of participants


Price per participant

Testimonials (2)

Upcoming Courses

Related Categories