gemma.cpp

gemma.cpp copied to clipboard

WSL support: "Killed", once executed on WSL

4

**WSL subsystem** anything seems normal, once executed, only "**Killed**" print-out. ``` $ ./gemma --tokenizer /home/home/F/GPT/Gemma/2b-it-cpp/tokenizer.spm \ > --compressed_weights /home/home/F/GPT/Gemma/2b-it-cpp/2b-it.sbs \ > --model 2b-it $ Killed ``` **ENV:** clang-10.0 ubuntu18.04 &&...

Code-keys

The application does not use Nvidia GPU in WSL

1

While running the Gemma C++ app using the 7B model in my WSL, the Nvidia GPU usage is at 0% though I have CUDA toolkit installed and also the WSL...

dannyvpm

add.class.ba

1

NegusBas

num_threads doesn't seem to have any effect

5

./gemma \ --tokenizer tokenizer.spm \ --compressed_weights 2b-it-sfp.sbs \ --model 2b-it \ --verbosity 2 \ --num_threads 2 Increasing num_threads like this doesn't improve speed. Is this expected?

windmaple

Restore reverted changes. Sync to https://github.com/google/gemma.cpp/commit/84444c93a44f484442fda2523dde7e77dbd3a53c.

copybara-service[bot]

Add information for the resetting conversation ``` *Usage* Enter an instruction and press enter (%C reset conversation, %Q quits). ```

kishida

I've generated Native C++ project on Android Studio in Windows to use gemma.cpp as library. and fill CMakeLists.txt as below. ``` cmake_minimum_required(VERSION 3.22.1) project("gemmacpp") add_library(${CMAKE_PROJECT_NAME} SHARED # List C/C++ source...

urim85

Zeenat30

Fixes #37, lambda issue: missing HWY_ATTR, and cannot capture SVE in/out vectors.

copybara-service[bot]

This PR makes the libgemma target generate position-independent code by default (aka adds `-fPIC `).

kewde