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
- Insight into Cambricon’s range of AI chips
- Understanding the MLU architecture and instruction flow
- Identifying supported model types and applicable use cases
Configuring the Development Toolchain
- Installation procedures for BANGPy and the Neuware SDK
- Setting up Python and C++ development environments
- Ensuring model compatibility and preprocessing requirements
Building Models with BANGPy
- Managing tensor structures and shapes
- Constructing computational graphs
- Leveraging custom operation support within BANGPy
Deployment Strategies with Neuware Runtime
- Techniques for converting and loading models
- Controlling execution and inference processes
- Best practices for deploying to edge and data center infrastructures
Optimizing Performance
- Tuning memory mapping and individual layers
- Utilizing execution tracing and profiling tools
- Identifying common bottlenecks and applying effective solutions
Integrating MLUs into Software Applications
- Employing Neuware APIs for seamless application integration
- Supporting streaming and multi-model architectures
- Handling hybrid CPU-MLU inference scenarios
End-to-End Project Implementation
- Practical lab: Deploying a vision or NLP model
- Executing edge inference through BANGPy integration
- Evaluating model accuracy and throughput metrics
Conclusion and Future Directions
Requirements
- A foundational grasp of machine learning model architectures
- Proficiency in Python and/or C++
- Basic familiarity with concepts related to model deployment and acceleration
Target Audience
- Developers specializing in embedded AI solutions
- ML engineers responsible for deployment in edge or data center contexts
- Professionals working within Chinese AI infrastructure ecosystems
Testimonials (1)
That we can cover advance topic and work with real-life example
