]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
all : be more strict about converting float to double (#458)
authorStephan Walter <redacted>
Tue, 28 Mar 2023 16:48:20 +0000 (16:48 +0000)
committerGitHub <redacted>
Tue, 28 Mar 2023 16:48:20 +0000 (19:48 +0300)
commit436e56193199a1625f8c561069f702e8840a9e08
tree9e7f39e1736ccff5728bb6194f160dfa94cf552d
parent20e1e84884376b3fb44ffbfd48d478b2934b0b5e
all : be more strict about converting float to double (#458)

* Be more strict about converting float to double

* Test equivalence of round, SILU implementations

Test module is commented out in CMakeLists.txt because the tests may
take a long time, depending on how much the compiler optimizes.

* Fix softmax in perplexity.cpp

* all : prefer float over double where appropriate

* perplexity : add <cmath>

---------

Co-authored-by: Georgi Gerganov <redacted>
CMakeLists.txt
Makefile
examples/common.cpp
examples/main/main.cpp
examples/perplexity/perplexity.cpp
examples/quantize/quantize.cpp
ggml.c
llama.cpp
llama.h
tests/CMakeLists.txt
tests/test-double-float.c [new file with mode: 0644]