Mini CLI search engine for your docs, knowledge bases, meeting notes, whatever. Uses current SOTA approaches while being all local.
qmd bundles node-llama-cpp which supports GPU acceleration for embeddings.
CUDA is supported on x86_64-linux. Enable it with the cudaSupport override:
qmd.override { cudaSupport = true; }Vulkan support is enabled by default on Linux for AMD/Intel GPUs. To disable it:
qmd.override { vulkanSupport = false; }