# Bundled llama.cpp sidecar binaries are build artifacts, never committed.
# Produced by scripts/build-llama.sh into lu-llama-server-<target-triple>[.exe]
# plus the ggml/llama/mtmd/omp runtime DLLs it links against on Windows.
# The old unprefixed name stays listed so a checkout from before GitHub #120
# does not suddenly offer a 3 GB binary for commit.
lu-llama-server-*
lu-llama-quantize-*
llama-server-*
*.dll
!.gitignore
