jundot/omlx v0.2.19

on GitHub

Download the DMG that matches your macOS version (sequoia or tahoe).
If you're on an M5 Mac, you must use the macos26-tahoe DMG for M5 Neural Accelerator.

Highlights

• Fix sustained GPU spike on idle — removed the keepalive warmup loop that caused unnecessary GPU usage (#292)
• Fix Metal buffer cache race condition — GPU sync before clearing Metal buffer cache (#300)
• Per-model OCR generation defaults — OCR models now use official recommended params to prevent repetition loops (#279)

Improvements

• temperature and repetition_penalty for OCR models can now be customized from the admin dashboard
• marked.parse() output in admin dashboard sanitized with DOMPurify to prevent XSS

Bug Fixes

• fix sustained GPU spike when server is idle due to keepalive warmup loop (#292)
• fix intermittent crash from Metal buffer cache race condition (#300)
• fix OCR model repetition loop from missing repetition_penalty and max_tokens cap (#279)
• fix server crash on startup when configured model directory is inaccessible
• fix update notifications showing pre-release (dev/rc) versions