Open Models Almanac
Methodsinference & servingKV cache management

single mention · inference & serving

Disabling prefix caching for benchmarking

Turns off prefix caching to obtain a clean benchmark measurement.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See KV cache management for the methods filed beside it.

Also called Disable prefix caching for benchmarking, disabling prefix caching.

Filed under inference & serving › KV cache management