Repository navigation
swift: pin mlx-swift-lm 3.31.13 and mlx-audio-swift 0.1.10 for Recursive Flint - #1150
Siddhesh2377 wants to merge 2 commits into
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Walkthrough
Merge Risk: ⚪ Minimal · up to
|
sanchitmonga22
left a comment
There was a problem hiding this comment.
Limited current-diff review of 9769f64: the two Swift manifests and five lockfiles consistently move mlx-swift-lm to 3.31.13 and mlx-audio-swift to 0.1.10, and the lockfile revisions match those repository tags. The llama.cpp pin and Swift/Kotlin/Flutter constants consistently move to runanywhere-b10453.9. No obvious blocking defect found in these scoped changes. The desktop-kit build and published 0.20.42 assets are verified; that release is this head plus the kit-version-only commit. This approval is not a claim of all-green CI or independently validated Flint generation: swift-spm and native_win_x64_and_package were still running, and MLX/GGUF runtime statements remain author evidence. Existing reviewer and branch requirements still apply.
|
Closing as superseded by #1153, which moves past every pin here (llama.cpp runanywhere-b10453.10, mlx-swift-lm 3.31.14, mlx-audio-swift 0.1.11). The DevTools/example/decision-cli — labubu for Sanchit |
Moves the MLX pins so the SDK can run Recursive Flint (
RecursiveSI/flint-int4).The manifests and every
Package.resolvedare updated by hand to the new tags and revisions. I did not build or re-resolve, so the lockfiles still carry their oldoriginHash; SwiftPM refreshes it on the next resolve. No runtime code changes: Flint loads through the generic MLX LLM path.It also moves llama.cpp from
runanywhere-b10453.8torunanywhere-b10453.9(core/VERSIONSand the version constants in the Swift, Kotlin and Flutter bindings), so the GGUF path can load Flint too..9adds theflintarch (upstream ggml-org/llama.cpp#30928, cherry-picked before it merged; RunanywhereAI/llama.cpp#6) and the pplx-decider v1.2 converter. It changes no public llama.cpp headers, so the engine code is untouched. I builtrac_backend_llamacppwith thecpp-desktop-macos-arm64preset against the new tag (Sherpa off), andlibllama.acarries the Flint model. The unslothflint-preview-UD-Q4_K_XLGGUF was tested separately on the fork build. iOS, Android and Windows were not built.Summary by cubic
Pins
mlx-swift-lm3.31.13 andmlx-audio-swift0.1.10, and bumps the llama.cpp pin torunanywhere-b10453.9, so the SDK can run Recursive Flint (RecursiveSI/flint-int4).mlx-swift-lm3.31.11 → 3.31.13: 3.31.12 adds Flint support and 3.31.13 declares its GLM tool-call format.mlx-audio-swift0.1.9 → 0.1.10 only re-pinsmlx-swift-lmto 3.31.13; since it pins exactly, both move together.runanywhere-b10453.8→ .9 adds theflintarch and the pplx-decider v1.2 converter; the binding version constants are updated to match.Package.resolvedfiles are updated by hand; lockfiles keep their oldoriginHashuntil the next SwiftPM resolve. No runtime code changes — Flint loads through the generic MLX LLM path.Written for commit 9769f64. Summary will update on new commits.
Summary by CodeRabbit