Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 7 hours
Course Outline
Overview of Huawei’s AI Ecosystem
- Ascend AI hardware: including 310, 910, and 910B chips.
- MindSpore, CANN, and their associated supporting tools.
- The end-to-end AI development workflow, from training through to deployment.
Deep Dive into the CANN Toolkit
- Defining CANN and its significance in AI computing.
- An overview of essential components, including ATC, AscendCL, and operator libraries.
- The specific role CANN plays within AI inference pipelines.
Initial Setup with MindSpore and CANN
- Configuring the development environment (MindSpore + CANN + Python).
- Training a basic model using MindSpore.
- Exporting and converting the model via ATC.
Executing Inference on Ascend Devices
- Utilizing the OM model through AscendCL or Python APIs.
- Handling basic input and output preprocessing tasks.
- Validating the accuracy and performance of model outputs.
Integrating with Alternative Frameworks
- A summary of support for TensorFlow, PyTorch, and ONNX.
- Details on supported operators and existing limitations.
- A practical demo of model conversion (e.g., transforming ONNX to OM format).
Navigating the CANN and MindSpore Developer Community
- Key resources such as documentation, GitHub repositories, and sample code.
- An introduction to MindSpore Hub and the model zoo.
- Accessing community forums, events, and support channels.
Recap and Future Directions
Requirements
- A foundational understanding of machine learning and deep learning principles.
- Familiarity with Python programming basics.
- No previous experience with CANN or Ascend hardware is necessary.
Target Audience
- Machine learning developers seeking to understand deployment workflows.
- Students and researchers new to Huawei’s AI ecosystem.
- AI framework contributors and enthusiasts interested in model acceleration techniques.
