Autodiff

• Introduction
• Concepts and memory model
• Worked implementation flow
• Operators
• Optimizers
• Training loop
• Fine-tuning an output layer after Ethos-U
• Converting PyTorch modules