mirror of
https://github.com/BigMarc/grammargem.git
synced 2026-10-13 17:01:55 +02:00
8a9ffb55af
Replace the MLXEngine stub with real Qwen2.5-3B-4bit inference via mlx-swift-examples (2.25.9): loads the local model snapshot on disk (no network), chat-templated generation per AIAction, and tone classification. Wired behind the existing AIEngine seam; AppState resolves the model dir through ModelManager. Build notes: - arm64-only now (MLX is Metal / Apple-Silicon only; matches the spec). - swift build (CLI) cannot compile mlx-swift's .metal kernels, so the metallib is generated once via scripts/build-metallib.sh (xcodebuild) and vendored at AppSupport/mlx.metallib; scripts/build.sh ships it as Contents/Resources/mlx-swift_Cmlx.bundle/Contents/Resources/default.metallib (the path mlx resolves, and it codesigns cleanly). - A hidden --mlx-selftest entrypoint runs one headless generation to verify the runtime end-to-end. Verified: the signed app's self-test produces a real rewrite; full test suite green (the MLX integration test is gated behind GRAMMAGEM_MLX_TEST). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
14 lines
316 B
Plaintext
14 lines
316 B
Plaintext
.build/
|
|
dist/
|
|
*.xcodeproj/xcuserdata/
|
|
.DS_Store
|
|
.swiftpm/
|
|
|
|
# Rust / Harper FFI build artifacts (regenerate with harper-ffi/build.sh).
|
|
# Cargo.lock IS committed (binary crate) for reproducible builds.
|
|
harper-ffi/target/
|
|
harper-ffi/lib/
|
|
|
|
# xcodebuild DerivedData (used only to generate the MLX metallib)
|
|
.xcode-build/
|