Query language for blending SQL and local language models across structured + unstructured data, with type constraints.

Reproduction Package for the paper "Type-Constrained Code Generation with Language Models" [PLDI 2025]

For our ICRA 2025 paper 🏆 "SELP: Generating Safe and Efficient Task Plans for Robot Agents with Large Language Models" by Yi Wu, Zikang Xiong, Yiran Hu, Shreyash Iyengar, Nan Jiang, Aniket Bera, Lin Tan, and Suresh Jagannathan. (🏆 Best Paper Award Finalist!)

Code for paper "Extract, Denoise and Enforce: Evaluating and Improving Concept Preservation for Text-to-Text Generation" EMNLP 2021 and "Constrained Abstractive Summarization: Preserving Factual Consistency with Constrained Generation" arXiv 2020

[Pytorch] Efficient tokenization for recommendations and generative retrieval. Inspired by STATIC decoding from "Vectorizing the Trie"

Speculative grammar backtracking algorithm for LLM decoding conforming to some lark context-free grammar (CFG)

Context-Free Grammar-guided Generation of FHIR Resources Using Large Language Models

VibeDrift - Run any LLM on your own hardware. Bypass the VRAM wall with CPU/RAM inference, MOE expert offloading, and 4-bit quantization. No Cloud, no Subscription.

Introduction to function calling in LLMs

ServiceNow → GBNF grammar + iii worker generator | Open-source deterministic AI connector

TexLM: Synthesizing Reliable Latex Matrix from Natural Language Input

Grammar-aided Constrained Decondig for self-aligned LLM with SFT and GRPO

Constrained decoding from scratch on Qwen3-0.6B: token-level logit masking for 100% valid function-call JSON

MAZE is an adaptive constraint-based code generation system that combines Large Language Models (LLMs) with formal constraint enforcement to produce more accurate and contextually appropriate code.

Research pilot (dormant): does one simple repair decision space work across math & output-constraint domains?

My 'Call Me Maybe' project from 42KL core

Translate natural language prompts into structured function calls using constrained decoding with a local LLM.

Constrained function calling with Qwen3-0.6B

Constrained-decoding function-calling tool — trie + per-argument type masking on Qwen3-0.6B for 100% schema-valid JSON output. 42 School project.

a function-calling tool that uses constrained decoding to generate valid from LLM outputs.