llama.cpp

ggml : reuse ggml_get_n_tasks() in ggml_graph_plan()

#4308

Merged