mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-05-07 01:24:24 +00:00
bd562fe4f7
* use fast copy when src and dst are contiguous and same shape * use int64_t ne and ignore shape