b7965

The developers of the llama.cpp project, a key tool for running AI models locally, have released a fix for a synchronization bug affecting Apple's Metal framework. The issue, related to asynchronous tensor copying, could cause instability on macOS and iOS devices. This update ensures smoother and more reliable performance for users running models like Llama on Apple Silicon Macs, Intel Macs, and iPhones, patching a problem identified in the code.