b9761
📦 llama-cppView on GitHub →
✨ 2 features🐛 3 fixes🔧 1 symbols
Summary
The server component has been refactored to move model downloading into a dedicated child process, enabling real-time progress tracking via SSE. This release also includes various platform-specific binary updates and minor fixes.
✨ New Features
- Server now moves model downloading to a dedicated child process.
- Real-time model load progress tracking is now available via the /models/sse endpoint.
🐛 Bug Fixes
- Fixed test case issues.
- Prevented detaching the thread.
- Shortened MODEL_DOWNLOAD_TIMEOUT in tests.