AI engineering glossary · Inference serving optimisation
What is calibration dataset?
Calibration dataset: A small, representative sample of inputs passed through a model to observe the range of its activations, so quantisation scales can be chosen. Poorly matched calibration data leads to bad scales and worse accuracy on real traffic.
Calibration dataset: an example
An Indian customer-support bot calibrates on 256 real, anonymised Hinglish chats rather than on English Wikipedia text.
Related terms
More Inference serving optimisation terms
- Chunked prefill
- Cold start
- Compute-bound
- Continuous batching
- Cost per token
- CPU offloading
- CUDA graphs
- Draft model
Where you learn it
Calibration dataset is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.