All Tools
L
LLMFreeOpen Source
LLGUIDANCE
Super-fast structured outputs for LLMs
MIT
ABOUT
Asking a model for JSON still yields broken objects, extra prose, and retries. llguidance compiles a grammar and masks tokens at decode time so the completion stays inside JSON, regex, or a CFG.
INSTALL
pip install llguidanceINTEGRATION GUIDE
1. Force JSON Schema-valid tool calls from an OpenAI-compatible server
2. Constrain a local model to a regex or CFG during sampling
3. Speed up structured decoding versus interpreter-based guidance
4. Embed grammar-constrained generation in a custom Rust or Python runtime
TAGS
pythonruststructured-outputjsongrammarconstrained-decoding