]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
ggml : generalize `quantize_fns` for simpler FP16 handling (#1237)
authorStephan Walter <redacted>
Wed, 5 Jul 2023 16:13:06 +0000 (16:13 +0000)
committerGitHub <redacted>
Wed, 5 Jul 2023 16:13:06 +0000 (19:13 +0300)
commit1b107b8550dced48dc5f41184640061354226b96
treea09a4c33c865828cd753c19af71c580f98735be5
parent8567c76b5326e862be0755a8dc1dd988223fcae3
ggml : generalize `quantize_fns` for simpler FP16 handling (#1237)

* Generalize quantize_fns for simpler FP16 handling

* Remove call to ggml_cuda_mul_mat_get_wsize

* ci : disable FMA for mac os actions

---------

Co-authored-by: Georgi Gerganov <redacted>
.github/workflows/build.yml
examples/quantize-stats/quantize-stats.cpp
ggml.c
ggml.h
llama.cpp
pocs/vdot/q8dot.cpp
pocs/vdot/vdot.cpp
tests/test-quantize-fns.cpp
tests/test-quantize-perf.cpp