Start Here: The Plan
The goal, units and vocabulary (FLOPs, bytes, bandwidth), a prerequisite check and your first measurements.
0 of 3 complete
- ▶ Up next10 min+50 XPWhat you will learn, and the planThe goal of Inference Engineering (predict, explain and improve inference performance with evidence), the route…
- 20 min+50 XPUnits and vocabulary: FLOPs, bytes, bandwidth and the balance pointThe numbers every inference calculation uses: parameters to bytes, FLOPs per token, memory bandwidth, compute…
- 15 min+50 XPPrerequisite check, and how to measure from day oneA short self-check of what the course assumes, the two measurements to take on your own machine before the first…