llama.cpp

0f1e9d14 - docs: update CPU backend ops to mark POOL_1D as supported (#20304)