llama.cpp

ggml-cpu: repack: Fix chunks being too small with small matrix shapes in REPACK forward_mul_mat

#17526

Merged