mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-10-27 08:21:30 +00:00
bd562fe4f7
* use fast copy when src and dst are contiguous and same shape * use int64_t ne and ignore shape