vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

ggml : move AMX to the CPU backend (#10570)

Browse source 
* ggml : move AMX to the CPU backend

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Diego Devesa 2024-11-29 21:54:58 +01:00 • committed by GitHub
parent b782e5c7d4
commit 7cc2d2c889
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
64 changed files with 514 additions and 801 deletions
Download patch file Download diff file Expand all files Collapse all files

2
tests/test-sampling.cpp
View file

@ -284,7 +284,7 @@ static void test_perf() {
data.reserve(n_vocab);
for (int i = 0; i < n_vocab; i++) {
const float logit = 2.0f*((float)(rand())/RAND_MAX - 0.5f);
const float logit = 2.0f*((double)(rand())/RAND_MAX - 0.5);
data.emplace_back(llama_token_data{i, logit, 0.0f});
}