Sundong Kim

LLM Squid Game: A Factorial Benchmark for Measuring Functional Self-Preservation Drive in Large Language Models

Neural MCTS with LLM Guidance for Effective Program Synthesis on Abstraction and Reasoning Corpus

This work introduces a neural MCTS algorithm customized for ARC-AGI, which amplifies exploration by directing the search toward promising solutions via learned neural guidance, yielding robust performance on demanding ARC tasks.

Semantic Mutation via LLMs: A Hybrid Approach to Evolutionary Program Synthesis

This work proposes a hybrid framework that integrates Large Language Models into GP in two complementary roles: as a semantic mutation operator that proposes non-local program rewrites during evolution, and as a post-evolution repair that iteratively refines single failed programs after search terminates.