Learning Rate Scheduling

• Learning Rate Decay tricks
• Gradient Descent
• Increasing the batch size, reduces noise in thearchitecture so a larger learning rate is okay
• Linear Learning Rate Scaling
• Learning Rate Warmup