Production-grade GBNF grammar-constrained decoding for LLMs. Zero dependencies, no unsafe code, pure Rust.
-
Updated
Aug 26, 2026 - Rust
#
grammar-constrained-decoding
Here are 7 public repositories matching this topic...
Local-first agentic inference engine in C/C++. Multi-tier model routing, grammar-constrained output, MCP tool servers. Embeddable via C ABI.
cpp mcp cuda inference-engine cpp20 privacy-first embedded-ai edge-ai on-device-ai llm llama-cpp local-llm gguf gbnf agentic-framework agentic-ai tool-calling grammar-constrained-decoding
-
Updated
Oct 17, 2026 - C++
Alert Radar — customizable market alert system (RSI, price, %-change, MA cross) with email delivery via Resend
-
Updated
Sep 1, 2026 - Python
A highly-decoupled, low-latency style dictionary API engine designed to deprecate hardcoded string arrays using grammar-constrained 1.5B SLM infrastructure with multi-tenant semantic isolation.
middleware decoupling api-design software-architecture json-mode small-language-models structured-outputs grammar-constrained-decoding compiler-engineering
-
Updated
Sep 6, 2026
Hybrid neurosymbolic stack for modeling semiconductor process-flow sequences (Infineon / Zero One Hack_01). Symbolic rule validator + grammar-constrained decoder + k-NN retrieval + compositionally-tokenized multi-task transformer with validity & rule-attribution heads. Trained from scratch on EuroHPC Leonardo.
machine-learning hpc slurm pytorch transformer manufacturing semiconductor anomaly-detection multi-task-learning out-of-distribution sequence-modeling neurosymbolic-ai xlstm grammar-constrained-decoding
-
Updated
May 31, 2026 - Python
Grammar-conditioned sparse LM-head projection for exact structured LLM decoding; Triton runtime, raw evidence, and reproducible Qwen3.5-2B benchmarks.
benchmarking triton json-decoding inference-optimization research-software ml-systems constrained-decoding llm-inference structured-generation xgrammar grammar-constrained-decoding sparse-projection
-
Updated
Sep 29, 2026 - Python
Twenty Raspberry Pi projects that pair a small local language model with a two-line control rule and publish the agreement table between them. Honest tokens-per-second figures measured on the boards, llama.cpp pinned at one tag, C, C++, Rust or Python chosen per project. A project where the model adds nothing is a valid result.
c rust raspberry-pi benchmarking yocto embedded-linux cortex-a72 local-inference llama-cpp gguf gbnf grammar-constrained-decoding
-
Updated
Oct 19, 2026 - C
Add this topic to your repo
To associate your repository with the grammar-constrained-decoding topic, visit your repo's landing page and select "manage topics."