Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pourcel, Julien, Colas, Cédric, Oudeyer, Pierre-Yves
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914394335608832
author Pourcel, Julien
Colas, Cédric
Oudeyer, Pierre-Yves
author_facet Pourcel, Julien
Colas, Cédric
Oudeyer, Pierre-Yves
contents Many program synthesis tasks prove too challenging for even state-of-the-art language models to solve in single attempts. Search-based evolutionary methods offer a promising alternative by exploring solution spaces iteratively, but their effectiveness remain limited by the fixed capabilities of the underlying generative model. We propose SOAR, a method that learns program synthesis by integrating language models into a self-improving evolutionary loop. SOAR alternates between (1) an evolutionary search that uses an LLM to sample and refine candidate solutions, and (2) a hindsight learning phase that converts search attempts into valid problem-solution pairs used to fine-tune the LLM's sampling and refinement capabilities\, -- \,enabling increasingly effective search in subsequent iterations. On the challenging ARC-AGI benchmark, SOAR achieves significant performance gains across model scales and iterations, leveraging positive transfer between the sampling and refinement finetuning tasks. These improvements carry over to test-time adaptation, enabling SOAR to solve 52\% of the public test set. Our code is open-sourced at: https://github.com/flowersteam/SOAR
format Preprint
id arxiv_https___arxiv_org_abs_2507_14172
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI
Pourcel, Julien
Colas, Cédric
Oudeyer, Pierre-Yves
Machine Learning
Artificial Intelligence
Neural and Evolutionary Computing
Many program synthesis tasks prove too challenging for even state-of-the-art language models to solve in single attempts. Search-based evolutionary methods offer a promising alternative by exploring solution spaces iteratively, but their effectiveness remain limited by the fixed capabilities of the underlying generative model. We propose SOAR, a method that learns program synthesis by integrating language models into a self-improving evolutionary loop. SOAR alternates between (1) an evolutionary search that uses an LLM to sample and refine candidate solutions, and (2) a hindsight learning phase that converts search attempts into valid problem-solution pairs used to fine-tune the LLM's sampling and refinement capabilities\, -- \,enabling increasingly effective search in subsequent iterations. On the challenging ARC-AGI benchmark, SOAR achieves significant performance gains across model scales and iterations, leveraging positive transfer between the sampling and refinement finetuning tasks. These improvements carry over to test-time adaptation, enabling SOAR to solve 52\% of the public test set. Our code is open-sourced at: https://github.com/flowersteam/SOAR
title Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI
topic Machine Learning
Artificial Intelligence
Neural and Evolutionary Computing
url https://arxiv.org/abs/2507.14172