.oooooo..o ooooo ooo ooooo
d8P' `Y8 `888' `8' `888'
Y88bo. oooo ooo ooo. .oo. .oooo. oo.ooooo. .oooo.o .ooooo. 888 8 888
`"Y8888o. `88. .8' `888P"Y88b `P )88b 888' `88b d88( "8 d88' `88b 888 8 888
`"Y88b `88..8' 888 888 .oP"888 888 888 `"Y88b. 888ooo888 888 8 888
oo .d8P `888' 888 888 d8( 888 888 888 o. )88b 888 .o `88. .8' 888
8""88888P' .8' o888o o888o `Y888""8o 888bod8P' 8""888P' `Y8bod8P' `YbodP' o888o
.o..P' 888
`Y8P' o888o
This project is still in the early stages of development, but it already contains a significant amount of code and foundational features. Many parts are still being refined or expanded as development continues.
For an introduction, check out
This project is designed to run comfortably in both Linux environments and Windows setups using WSL. Either option works without issues. We also provide ready-to-use Docker files, so most of the environment setup is handled automatically, and you can start working with minimal configuration.
Before starting the AI services, make sure to choose your GPU or CPU setup as described below.
After that, the ggml's llama.cpp and Ollama containers need to be started manually by specifying their service names.
Docker commands. You must write the commands in the root directory and have Docker Engine installed.
# App up
docker compose up -d
# For llama.cpp (llama-swap)
docker compose up svc.synui_llama-swap -d
# For Ollama
docker compose up svc.synui_ollama -dThere are two .env files used in this project, each serving a different purpose:
-
The
.envfile in the project root contains Docker-related settings. The options you’ll adjust most often here are the GPU/CPU selection and the AI adapter you want to use. -
The
src/.envfile holds the application’s own configuration. This is where you define database settings and, again, choose the adapter the application should work with.
Both files include comments that explain each option, so following the instructions inside them is usually enough to get everything set up correctly.
Some folders need writable permissions for the application to work properly.
# Create required directories
mkdir -p logs
mkdir -p src/internal/template_cache
mkdir -p src/internal/logs
# Set ownership to web server user
sudo chown -R www-data:www-data logs src/internal/template_cache src/internal/logs
# Set directory permissions to 755 for general directories
sudo find logs src/internal/template_cache src/internal/logs -type d -exec chmod 755 {} \;
# Set file permissions to 644 for any files inside
sudo find logs src/internal/template_cache src/internal/logs -type f -exec chmod 644 {} \;
# Make writable directories (cache/logs) writable by owner and web server
sudo chmod -R 770 logs src/internal/template_cache src/internal/logsCurrently fully support for ollama and llama.cpp(llama-swap). You have to define it in the src/.env file in the APP_LLM_ADAPTER variable and choose which one Docker will install.
But you should know that this will only apply to Docker.
We currently recommend llama.cpp. However, this requires providing a model input when preloading llama.cpp. To overcome this, we used the llama-swap proxy server. This allows us to use any model we want, just like in ollama, and achieve our project goals.
Because llama.cpp feels faster and much more customizable now. Ollama is all set, but we still need a "sufficient" amount of customization. Ollama is like the end user, however in this project we want to enable the end user to use the chain model.
First you must change the UTILIZATION cpu(or cuda) to NVIDIA(nvgpu) or AMD ROCm(amdgpu) in Docker .env file.
Both llama.cpp and Ollama support GPU utilization.
The steps below are for Ubuntu and Debian. However, I recommend installing according to the vendor's wiki page.
UTILIZATION=cudafor both support CPU/GPU in llama.cpp (llama-swap) (recommended)UTILIZATION=nvgpufor GPU acceleration in OllamaUTILIZATION=cpufor CPU in Ollama (llama-swap)
If you're on the llama.cpp platform, you can use cuda directly. This allows you to use both the CPU and the NVIDIA GPU simultaneously.
But you need to be careful, if you want to force it to GPU or CPU, you need to define the model like gemma3:1b_cpu(or cuda) in functions like chat() or generate().
More info for: https://github.com/par274/synapseui/tree/main/.docker/llama-swap/swap-config.yaml
If you are going to use NVIDIA GPU, you must install NVIDIA Container Toolkit.
For Ubuntu, Debian (example. Please follow official NVIDIA steps): https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
# Install the prerequisites for the instructions below:
sudo apt-get update && sudo apt-get install -y --no-install-recommends \
curl \
gnupg2
# Configure the production repository:
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
&& curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | \
sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
# Optionally, configure the repository to use experimental packages:
sudo sed -i -e '/experimental/ s/^#//g' /etc/apt/sources.list.d/nvidia-container-toolkit.list
# Update the packages list from the repository:
sudo apt-get update
# Install the NVIDIA Container Toolkit packages:
export NVIDIA_CONTAINER_TOOLKIT_VERSION=1.18.0-1
sudo apt-get install -y \
nvidia-container-toolkit=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
nvidia-container-toolkit-base=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
libnvidia-container-tools=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
libnvidia-container1=${NVIDIA_CONTAINER_TOOLKIT_VERSION}# Configure the container runtime by using the nvidia-ctk command:
sudo nvidia-ctk runtime configure --runtime=docker
# Restart the Docker daemon: (this is not necessary if you're in WSL)
sudo systemctl restart docker
# For testing
sudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smiUTILIZATION=amdgpu
If you use AMD GPU and it has ROCm support, you can use it too. First, you need to install amdgpu-dkms. If you are already install ROCm, you have it. So you can skip this step.
For Ubuntu, Debian (example. Please follow official AMD steps): https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/quick-start.html#rocm-install-quick
wget https://repo.radeon.com/amdgpu-install/7.1.1/ubuntu/noble/amdgpu-install_7.1.1.70101-1_all.deb
sudo apt install ./amdgpu-install_7.1.1.70101-1_all.deb
sudo apt update
sudo apt install python3-setuptools python3-wheel
sudo usermod -a -G render,video $LOGNAME # Add the current user to the render and video groups
sudo apt install rocmAnd reboot your system.
Runs on a custom solution called the Synaptic Framework (Par3).
This application framework is designed around a modular architecture and uses a custom template engine whose syntax is inspired by XenForo 2 and vBulletin, while its template architecture follows the design principles and evolution of Twig. It separates core functionality into independent services managed by a container system, enabling clean organization, extensibility, and the creation of efficient native-style applications. It also aims to integrate React and PHP in a hybrid structure that unifies modern reactive interfaces with traditional server-side execution.
The software in this repository is released under the GNU General Public License v2 (GPL-2.0) or later from the Free Software Foundation. You can also read the text of the license.