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.
Course Outline
Introduction to CANN and Ascend AI Processors
- Defining CANN and its position within Huawei’s AI compute ecosystem
- Examination of Ascend processor architectures (including models 310, 910, and others)
- Overview of supported AI frameworks and the associated toolchain
Model Conversion and Compilation Processes
- Utilizing the ATC tool to convert models from TensorFlow, PyTorch, and ONNX
- Generation and validation of OM model files
- Addressing unsupported operators and resolving frequent conversion challenges
Deployment via MindSpore and Alternative Frameworks
- Executing model deployments using MindSpore Lite
- Integrating OM models with Python APIs or C++ SDKs
- Operating with the Ascend Model Manager
Performance Tuning and Profiling Techniques
- Exploring AI Core, memory management, and tiling optimization strategies
- Profiling model execution through CANN diagnostic tools
- Applying best practices to enhance inference speed and reduce resource consumption
Error Management and Debugging Protocols
- Identifying common deployment errors and implementing effective resolutions
- Interpreting logs and utilizing error diagnosis utilities
- Conducting unit tests and functional validation of deployed models
Edge and Cloud Deployment Use Cases
- Deploying solutions to Ascend 310 for edge-focused applications
- Integrating with cloud-based APIs and microservice architectures
- Analysis of real-world case studies in computer vision and NLP
Concluding Summary and Future Directions
Requirements
- Proficiency with Python-based deep learning frameworks, including TensorFlow or PyTorch
- A solid understanding of neural network architectures and model training workflows
- Foundational knowledge of the Linux CLI and scripting
Target Audience
- AI engineers focused on model deployment strategies
- Machine learning professionals aiming to leverage hardware acceleration
- Deep learning developers constructing inference solutions
14 Hours
