Open Models Almanac
Methodsinference & servinginference quantization

single mention · inference & serving

Native FP8 KV cache with offloading

A serving configuration combining native FP8 KV-cache storage with offloading.

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

Also called native FP8 KV cache and offloading.

Filed under inference & serving › inference quantization