Fix: just setup omits mlx-audio/mlx-lm install on Apple Silicon#681
Fix: just setup omits mlx-audio/mlx-lm install on Apple Silicon#681august-hill wants to merge 1 commit into
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughThe ChangesMLX Package Installation for Apple Silicon
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Problem
A fresh
just setupon Apple Silicon produces a venv missing themlx_audiomodule. Any MLX model load — e.g. downloading Whisper for dictation — then fails with:In the app this surfaces as a model download that appears to stall on "Connecting to download…": the background download task raises immediately, but the progress UI is never moved off its initial state, so it looks like a hung/slow download rather than an error.
Root cause
backend/requirements-mlx.txtintentionally excludesmlx-audio— from 0.3.1 it declarestransformers>=5.x, which conflicts with thetransformers<=4.57.6cap inrequirements.txt. That file's own comment says to install it separately withpip install --no-deps mlx-audio==0.4.1, pointing to.github/workflows/release.yml.release.ymldoes exactly that. Butsetup-pythonin thejustfile— the path every contributor and local builder uses — installsrequirements-mlx.txtand stops. Release builds getmlx-audio; local dev/build setups silently do not.Fix
Add the two
--no-depsinstalls to the Apple Silicon branch ofsetup-python, mirroringrelease.yml:Windows
setup-pythonis unaffected (no MLX path). One file changed (justfile).Testing
On an M-series Mac, after applying the fix:
python -c "import mlx_audio, mlx_lm"succeeds.whisper-turbomodel (openai/whisper-large-v3-turbo, 1.62 GB) downloads and loads:Note on the PR checklist
CONTRIBUTING.mdasks contributors to updateCHANGELOG.md, butCHANGELOG.md's header states it is compiled automatically and must not be edited manually. I left it untouched accordingly — flagging the doc inconsistency in case the checklist wording should be revisited.Summary by CodeRabbit