llama.cpp

295f8565 - allocators wip

allocators wip renamed ggml_backend functions changed ggml_buffer and ggml_backend to always be used as pointers rename ggml_tensor::params -> op_params