b9081
Minor but crucial fix for tagged parsers improves structured output reliability
Deep Dive
llama.cpp released b9081 (08 May) with a commit: "common: do not wrap raw strings in schema parser for tagged parsers (#22827)". Pre-built binaries are available for macOS (Apple Silicon, Intel, iOS XCFramework), Linux (x64/arm64/s390x CPU, Vulkan, ROCm 7.2, OpenVINO, SYCL), Windows (x64/arm64 CPU, CUDA 12 & 13, Vulkan, SYCL, HIP), Android (arm64 CPU), and openEuler (x86/aarch64 with ACL Graph).
Key Points
- Fixes raw string wrapping in schema parser for tagged parsers (#22827)
- Supports 20+ platform+backend variants including macOS, Windows, Linux, Android, iOS
- Project has 109k stars and 18k forks on GitHub
Why It Matters
Ensures correct JSON schema generation for tagged parsers, critical for reliable structured output in local LLM workflows.