]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: 305eb5a)
ggml : adjust mul_mat_f16 work memory (#1226)
authorGeorgi Gerganov <redacted>
Sat, 29 Apr 2023 15:43:28 +0000 (18:43 +0300)
committerGitHub <redacted>
Sat, 29 Apr 2023 15:43:28 +0000 (18:43 +0300)
commit214b6a35702a489e3738acd81fad6d46182d3036
treedac39b6d4bb7eaf958735a0dfb5ccabcbbb0821ctree
parent305eb5afd51325e3142c01c17431febb7c67de87commit | diff
ggml : adjust mul_mat_f16 work memory (#1226)

* llama : minor - remove explicity int64_t cast

* ggml : reduce memory buffer for F16 mul_mat when not using cuBLAS

* ggml : add asserts to guard for incorrect wsize
Makefile diff | blob | history
ggml.c diff | blob | history
llama.cpp diff | blob | history
Packaging of ggml-org/llama.cpp