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 21 hours
Course Outline
Foundations of Cambricon and MLU Architecture
- Survey of Cambricon’s AI semiconductor lineup.
- MLU architectural design and instruction pipelines.
- Compatible model types and typical applications.
Provisioning the Development Stack
- Installation of BANGPy and the Neuware SDK.
- Configuring Python and C++ workspaces.
- Managing model compatibility and preprocessing.
Model Engineering with BANGPy
- Handling tensor structures and shape management.
- Constructing computation graphs.
- Implementing custom operations in BANGPy.
Deployment via Neuware Runtime
- Model conversion and loading strategies.
- Managing execution and inference control.
- Best practices for edge and data center rollouts.
Performance Enhancement
- Memory mapping and layer-level tuning.
- Utilizing execution tracing and profiling tools.
- Identifying and resolving common performance bottlenecks.
Application Integration
- Leveraging Neuware APIs for seamless app integration.
- Supporting streaming and multi-model operations.
- Designing hybrid CPU-MLU inference paths.
Capstone Project and Practical Application
- Lab: Deploying a computer vision or NLP model.
- Edge inference workflows using BANGPy.
- Benchmarking accuracy and throughput.
Conclusions and Future Roadmap
Requirements
- Proficiency with machine learning model architectures.
- Development experience in Python and/or C++.
- Knowledge of model deployment and acceleration principles.
Target Audience
- Embedded AI software engineers.
- ML specialists focusing on edge or data center deployments.
- Developers engaged with Chinese AI hardware ecosystems.
Testimonials (1)
That we can cover advance topic and work with real-life example
