llama.cpp b9658 Now Includes Full Unparsed Prompt in Debug on Chat Parse Errors
English summary
The llama.cpp project released build b9658. A key change improves chat debugging: on parse errors, the debug output now includes the full unparsed prompt. The release also provides pre-built binaries for many platforms, including macOS (Apple Silicon, Intel), Linux (CPU, Vulkan, ROCm, OpenVINO, SYCL), Android (arm64 CPU), and Windows (CPU, CUDA, Vulkan, SYCL, HIP). The KleidiAI-enabled macOS Apple Silicon build is currently disabled in this release.
Chinese summary
llama.cpp 项目发布了 build b9658。一项重要改进是增强了聊天调试功能：当发生解析错误时，调试信息现在会包含完整的未解析提示。该版本仍提供面向多个平台的预编译二进制文件，包括 macOS（Apple Silicon 和 Intel）、Linux（CPU、Vulkan、ROCm、OpenVINO、SYCL）、Android（arm64 CPU）以及 Windows（CPU、CUDA、Vulkan、SYCL、HIP）。目前，启用 KleidiAI 的 macOS Apple Silicon 版本被暂时禁用。
Key points
Chat debug now includes the full unparsed prompt on parse errors (PR #24650).
聊天解析错误时，调试输出现在包含完整的未解析提示（PR #24650）。
Pre-built binaries released for macOS, Linux, Android, and Windows across many backends (CPU, CUDA, Vulkan, ROCm, OpenVINO, SYCL, HIP).
发布了面向 macOS、Linux、Android 和 Windows 多个后端（CPU、CUDA、Vulkan、ROCm、OpenVINO、SYCL、HIP）的预编译二进制文件。
The KleidiAI-enabled macOS Apple Silicon package is disabled in this build.
本次构建中，启用 KleidiAI 的 macOS Apple Silicon 包被禁用。