mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-07-27 21:21:02 -05:00
bd562fe4f7
* use fast copy when src and dst are contiguous and same shape * use int64_t ne and ignore shape