Change8

b9870

📦 llama-cppView on GitHub →
🐛 1 fixes🔧 1 symbols

Summary

This release addresses an issue causing long reasoning loops by implementing message trimming specifically for the StepFun parser. It also includes updated pre-compiled binaries for macOS, Linux, Android, and Windows supporting various hardware accelerators.

🐛 Bug Fixes

  • Trimmed messages sent to the StepFun parser to fix long reasoning loops by applying trimming earlier, before rendering, to content_parts text, string content, and reasoning_content.

Affected Symbols