Instant, controllable, local pre-trained AI models in Rust
awesome-LLM-controlled-constrained-generation
Bridging Items and Language: A Transition Paradigm for Large Language Model-Based Recommendation (KDD'24)
Experiments in Latin dactylic hexameter generation with transformers: A hybrid post hoc feedback framework
Structured text generation, information extraction, and even more.
lift-sys is a modular, verifiable AI-native software development environment, supporting next-generation constrained code generation and reverse-mode formalization, aiming to democratize high-quality software creation while keeping users in control.
Graph-first intermediate language for auditable LLM-generated graph queries and updates.
A constrained generation library for llama.cpp that prevents malformed outputs. Features include XML/JSON tag completion, multi-step reasoning support, and an ultra-simple API. Built on llama.cpp's sampler architecture for reliable token filtering.
Runtime semantic constraints and targeted repair for structured LLM output — declare which fields your code owns and which the model may invent.
A rule-constrained AI recommendation and production workflow for personalized blockchain services requiring whitelists, human review, traceability, and feedback.
Quantum-assisted verification of LLM reasoning. Translates each step to first-order CNF, grounds against a finite universe, then checks consistency with Grover's search on PennyLane simulator or IBM Heron r2 hardware.
A constrained generation filter for local LLMs that makes them quote properly from a source document
Resume interrupted structured LLM output with automatically derived continuation schemas, immutable reconstruction, and reproducible benchmarks.
Educational repository demonstrating how BNF grammars can constrain LLM generation to produce guaranteed valid structured outputs (JSON, code, etc.)
To associate your repository with the constrained-generation topic, visit your repo's landing page and select "manage topics."