Advanced AI Coaching System Requirements

Macro Trainer features a state-of-the-art Advanced AI Coaching system. To guarantee complete user privacy, the coaching engine runs entirely locally on your device. No conversations, stats, or logs ever leave your machine or get sent to third-party APIs.

While mobile platforms (iOS and Android) support this out-of-the-box using integrated LiteRT runtimes, desktop clients (macOS and Linux) run the model using an external inference CLI utility.

Supported Platforms

• iOS & Android: No external programs required. The application runs natively on-device.
• macOS & Linux: Requires the installation of llama.cpp (specifically the llama-cli or llama-completion executable).

Desktop Requirements

To enable Advanced AI Coaching on macOS or Linux desktop builds, the application must be able to find a compiled executable named llama-cli or llama-completion in your system PATH or at one of the following directories:

• /usr/local/bin/ (Recommended path)
• /usr/bin/
• /usr/bin/llama.cpp/
• /opt/llama.cpp/bin/

Installation Guides

macOS Setup

The easiest way to install the necessary tools on macOS is using Homebrew:

# Install llama.cpp
brew install llama.cpp

Macro Trainer automatically scans Homebrew’s default path (/opt/homebrew/bin/llama-cli) and will detect the tool immediately after installation.

Linux Setup (Debian / Ubuntu / Linux Mint)

For Linux distributions, building from source is recommended to ensure the compilation targets your CPU’s hardware instructions (like AVX2/AVX512) and GPU drivers.

# 1. Install compiler tools and Vulkan headers (for GPU acceleration)
sudo apt update
sudo apt install build-essential cmake git libvulkan-dev glslc vulkan-tools spirv-headers glslang-tools

# 2. Clone the official llama.cpp repository
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp

# 3. Configure CMake with Vulkan GPU support enabled
cmake -B build -DBUILD_SHARED_LIBS=OFF -DGGML_VULKAN=ON
cmake --build build --config Release --target llama-cli

# 4. Copy the compiled CLI to your local bin directory
sudo cp build/bin/llama-cli /usr/local/bin/

Hardware & System Requirements

• Model File: The application automatically downloads the Gemma-4-2B-IT (Q4_K_M) model in GGUF format.
• Free Disk Space: Minimum of 4.0 GB of free disk space is required to download and save the 3.1 GB model file.
• System Memory (RAM):
  • GPU offloading (recommended): A graphics card with at least 4 GB VRAM and installed drivers (Apple Silicon Unified Memory, NVIDIA CUDA, or Vulkan).
  • CPU execution: Minimum of 8 GB RAM.