LATIDIA · Investigación

Rastreo de los pensamientos de un agente de programación que juega a ARC-AGI-3: Lecciones para el aprendizaje continuo

arXiv:2610.11450v1 Anuncio Tipo: nuevo Resumen: Estudiamos cómo un agente de codificación aprende a través de una secuencia de tareas de razonamiento abstracto. El agente se ejecuta en un modelo base congelado dentro de un arnés fijo y actúa escribiendo y

La noticia

arXiv:2610.11450v1 Announce Type: new Abstract: We study how a coding agent learns across a sequence of abstract reasoning tasks. The agent runs on a frozen foundation model inside a fixed harness and acts by writing and running Python and shell scripts. It retains no state across turns other than its written artifacts, so every thought it forms, carries, corrects or abandons leaves a trace, where a thought is any belief, rule or plan committed to a file. We let the agent play ARC-AGI-3, a set of interactive reasoning games that provide no instructions. Each game is a sequence of levels, and a strategy that clears one level can fail on