The Abstraction and Reasoning Corpus

Rigourous evaluation of LLM-synthesized code - NeurIPS 2023 & COLM 2024

Using artificial intelligence and genetic algorithms to automatically write programs. Tutorial: http://www.primaryobjects.com/cms/article149

A Dataset of Python Challenges for AI Research

Microsoft Program Synthesis using Examples SDK is a framework of technologies for the automatic generation of programs from input-output examples. This repo includes samples and sample data for the Microsoft Program Synthesis using Example SDK.

Direct Manipulation Programming for HTML/SVG

Performs string manipulation tasks by learning from the provided example(s), instead of having to program them out explicitly.

Code for Parsel 🐍 - generate complex programs with language models

APPS: Automated Programming Progress Standard (NeurIPS 2021)

A relation-aware semantic parsing model from English to SQL

Neural Symbolic Machines is a framework to integrate neural networks and symbolic representations using reinforcement learning, with applications in program synthesis and semantic parsing.

discover elisp functions that do what you want

BigCodeBench: Benchmarking Code Generation Towards AGI

The collection synthesizer

A Python framework for program synthesis with a focus on Automated Machine Learning.

Code and Data artifact for NeurIPS 2023 paper - "Monitor-Guided Decoding of Code LMs with Static Analysis of Repository Context". `multispy` is a lsp client library in Python intended to be used to build applications around language servers.

An inductive logic programming system

Metagol - an inductive logic programming system

🌌️Fuzz4All: Universal Fuzzing with Large Language Models

Programmatic modification and evaluation of software