Deployment & Scaling · Go deeper
Model Optimization
Make inference faster and cheaper: quantization, KV caching, batching, and compilation for production serving.
- 45 min lesson
- advanced level
- Narrated presentation
What you’ll learn
- Quantization for inference
- KV cache and PagedAttention
- Continuous batching
- Speculative decoding
- Kernel fusion / FlashAttention
- Latency vs. throughput benchmarking
Inside this lesson
A narrated, slide-by-slide presentation with 1 prediction exercise, 1 knowledge check, 2 videos. You answer as you go, and the platform checks your work.