nomic-ai/gpt4all
1
1
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2024-11-05 04:12:29 +03:00
Code Issues Packages Projects Releases Wiki Activity
c73f0e5c8c
gpt4all/gpt4all-chat/build_and_run.md
Jared Van Bortel d2a99d9bc6
support the llama.cpp CUDA backend (#2310) 
* rebase onto llama.cpp commit ggerganov/llama.cpp@d46dbc76f
* support for CUDA backend (enabled by default)
* partial support for Occam's Vulkan backend (disabled by default)
* partial support for HIP/ROCm backend (disabled by default)
* sync llama.cpp.cmake with upstream llama.cpp CMakeLists.txt
* changes to GPT4All backend, bindings, and chat UI to handle choice of llama.cpp backend (Kompute or CUDA)
* ship CUDA runtime with installed version
* make device selection in the UI on macOS actually do something
* model whitelist: remove dbrx, mamba, persimmon, plamo; add internlm and starcoder2

Signed-off-by: Jared Van Bortel <jared@nomic.ai>
2024-05-15 15:27:50 -04:00

5.1 KiB
Raw Blame History

Building gpt4all-chat from source

Depending upon your operating system, there are many ways that Qt is distributed. Here is the recommended method for getting the Qt dependency installed to setup and build gpt4all-chat from source.

Prerequisites

You will need a compiler. On Windows, you should install Visual Studio with the C++ Development components. On macOS, you will need the full version of Xcode—Xcode Command Line Tools lacks certain required tools. On Linux, you will need a GCC or Clang toolchain with C++ support.

On Windows and Linux, building GPT4All with full GPU support requires the Vulkan SDK and the latest CUDA Toolkit.

Note for Linux users

Linux users may install Qt via their distro's official packages instead of using the Qt installer. You need at least Qt 6.5, with support for QPdf and the Qt HTTP Server. It should be straightforward to build with just cmake and make, but you may continue to follow these instructions to build with Qt Creator.

On Arch Linux, this looks like:

sudo pacman -S --needed base-devel qt6-base qt6-declarative qt6-wayland qt6-svg qt6-httpserver qt6-webengine qt6-5compat qt6-shadertools qtcreator cmake ninja

On Ubuntu 23.04, this looks like:

sudo apt install build-essential qt6-base-dev qt6-declarative-dev qt6-wayland-dev qt6-svg-dev qt6-httpserver-dev qt6-webengine-dev libqt6core5compat6 qml6-module-qt5compat-graphicaleffects libqt6shadertools6 qtcreator cmake ninja-build

On Fedora 39, this looks like:

sudo dnf install make gcc gcc-c++ qt6-qtbase-devel qt6-qtdeclarative-devel qt6-qtwayland-devel qt6-qtsvg-devel qt6-qthttpserver-devel qt6-qtwebengine-devel qt6-qt5compat qt5-qtgraphicaleffects qt6-qtshadertools qt-creator cmake ninja-build

Download Qt

  • Go to https://login.qt.io/register to create a free Qt account.
  • Download the Qt Online Installer for your OS from here: https://www.qt.io/download-qt-installer-oss
  • Sign into the installer.
  • Agree to the terms of the (L)GPL 3 license.
  • Select whether you would like to send anonymous usage statistics to Qt.
  • On the Installation Folder page, leave the default installation path, and select "Custom Installation".

Customize the installation

image

Under "Qt", find the latest Qt 6.x release.

Under this release (e.g. Qt 6.5.0), select the target platform:

  • On macOS, it is just called "macOS".
  • On Windows, it is called "MSVC 2019 64-bit" (for 64-bit x86 CPUs). MinGW has not been tested.

Under this release, select the following additional components:

  • Qt Quick 3D
  • Qt Wayland Compositor (for Linux only)
  • Qt 5 Compatibility Module
  • Qt Shader Tools
  • Additional Libraries:
    • Qt HTTP Server
    • Qt PDF
  • Qt Debug information Files

Under Developer and Designer Tools, select the following components:

  • Qt Creator
  • Qt Creator CDB Debugger Support (for Windows only)
  • Debugging Tools for Windows (for Windows only)
  • CMake
  • Ninja

Agree to the license and complete the installation.

Download the source code

You must use git to download the source code for gpt4all:

git clone --recurse-submodules https://github.com/nomic-ai/gpt4all

Note the use of --recurse-submodules, which makes sure the necessary dependencies are downloaded inside the repo. This is why you cannot simply download a zip archive.

Windows users: To install git for Windows, see https://git-scm.com/downloads. Once it is installed, you should be able to shift-right click in any folder, "Open PowerShell window here" (or similar, depending on the version of Windows), and run the above command.

Open gpt4all-chat in Qt Creator

Open Qt Creator. Navigate to File > Open File or Project, find the "gpt4all-chat" folder inside the freshly cloned repository, and select CMakeLists.txt.

image

Configure project

You can now expand the "Details" section next to the build kit. It is best to uncheck all but one build configuration, e.g. "Release", which will produce optimized binaries that are not useful for debugging.

Click "Configure Project", and wait for it to complete.

image

Build project

Now that the project has been configured, click the hammer button on the left sidebar to build the project.

image

Run project

Click the play button on the left sidebar to run the Chat UI.

image

Updating the downloaded source code

You do not need to make a fresh clone of the source code every time. To update it, you may open a terminal/command prompt in the repository, run git pull, and then git submodule update --init --recursive.