Reduce allocation overhead in quantized sdpa #20059
build-presets.yml
on: pull_request
Matrix: apple
Matrix: linux
Matrix: windows
Matrix: zephyr