DOCSEARCHbyAlgolia
Submit Docs

LM Studio Configuration and Extension Packs

repository·main·Indexed 42 days ago

https://github.com/lmstudio-ai/configs

Reference for the LM Studio JSON configuration file format, including schema definitions and examples. Provides installation instructions for extension packs, including AMD ROCm (0.2.31), OpenCL (0.2.25), CUDA AVX-Only, and CPU AVX-Only for Windows and Linux.

Tokens
879
Snippets
5
Records
7
Agent score
3%

What's inside LM Studio

  1. Understand the LM Studio JSON configuration format

    main
    The configs repository provides the specification for the LM Studio JSON configuration file format and a collection of example configuration files. Developers can use these examples to understand how to structure configuration files for LM Studio applications.

  2. Install AMD ROCm extension pack on Windows (0.2.31)

    main

    To install the AMD ROCm extension pack on Windows, run the provided PowerShell command before starting the LM Studio application.

    After installation, you must manually activate the new backend:

    1. Navigate to Settings -> Advanced Configuration -> GPU Settings.
    2. Locate GPU Backend Type.
    3. Right-click and select the new ROCm backend to make it the active LM-Runtime.
    Invoke-Expression ([System.Text.Encoding]::UTF8.GetString((Invoke-WebRequest -Uri https://files.lmstudio.ai/windows/extension-pack-install-scripts/win-rocm-0.2.31-ext-install.ps1 -UseBasicParsing).Content))

  3. Install AMD ROCm extension pack on Linux (0.2.31)

    main

    To install the AMD ROCm extension pack on Linux, run the following command in your terminal before starting the LM Studio application.

    After installation, activate the backend via the UI:

    1. Navigate to Settings -> Advanced Configuration -> GPU Settings.
    2. Locate GPU Backend Type.
    3. Right-click and select the ROCm backend.
    curl -fsSL https://files.lmstudio.ai/linux/0.2.31/beta/extension-pack-install-scripts/lin-rocm-0.2.31-ext-install.sh | sh

  4. Clean up existing extension backends

    main

    If you have previously downloaded extension packs, you must clear the existing backend cache to ensure new extension packs install correctly. This is a required one-time step for users migrating or updating extensions.

    Windows: Delete all contents in %USERPROFILE%\ .cache\lm-studio\extensions\backends Linux: Delete all contents in ~/.cache/lm-studio/extensions/backends

  5. Install CPU AVX-Only extension pack on Linux (0.2.31)

    main

    To install the CPU AVX-Only extension pack on Linux, run the following command in your terminal before starting the LM Studio application.

    curl -fsSL https://files.lmstudio.ai/linux/0.2.31/beta/extension-pack-install-scripts/lin-cpu-avx-0.2.31-ext-install.sh | sh

  6. Install CUDA AVX-Only extension pack on Linux (0.2.31)

    main

    To install the CUDA AVX-Only extension pack on Linux, run the following command in your terminal before starting the LM Studio application.

    curl -fsSL https://files.lmstudio.ai/linux/0.2.31/beta/extension-pack-install-scripts/lin-cuda-avx-0.2.31-ext-install.sh | sh

  7. Install OpenCL extension pack on Windows (0.2.25)

    main

    To install the OpenCL extension pack on Windows, run the following PowerShell command before starting the application.

    Warning: This version (v0.2.25) is deprecated and does not support Gemma 2 or Llama 3.1 models.

    Invoke-Expression ([System.Text.Encoding]::UTF8.GetString((Invoke-WebRequest -Uri https://files.lmstudio.ai/win-opencl-ext-install.ps1 -UseBasicParsing).Content))