Communication Dans Un Congrès Année : 2025

MAGELLAN: Metacognitive predictions of learning progress guide autotelic LLM agents in large goal spaces

Résumé

Open-ended learning agents must efficiently prioritize goals in vast possibility spaces, focusing on those that maximize learning progress (LP). When such autotelic exploration is achieved by LLM agents trained with online RL in high-dimensional and evolving goal spaces, a key challenge for LP prediction is modeling one's own competence, a form of metacognitive monitoring. Traditional approaches either require extensive sampling or rely on brittle expert-defined goal groupings. We introduce MAGELLAN, a metacognitive framework that lets LLM agents learn to predict their competence and LP online. By capturing semantic relationships between goals, MAGELLAN enables sample-efficient LP estimation and dynamic adaptation to evolving goal spaces through generalization. In an interactive learning environment, we show that MAGELLAN improves LP prediction efficiency and goal prioritization, being the only method allowing the agent to fully master a large and evolving goal space. These results demonstrate how augmenting LLM agents with a metacognitive ability for LP predictions can effectively scale curriculum learning to open-ended goal spaces.

Fichier principal
Vignette du fichier
2502.07709v3.pdf (9.22 Mo) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte
Licence

Dates et versions

hal-05302437 , version 1 (08-10-2025)

Licence

Identifiants

Citer

Loris Gaven, Thomas Carta, Clément Romac, Cédric Colas, Sylvain Lamprier, et al.. MAGELLAN: Metacognitive predictions of learning progress guide autotelic LLM agents in large goal spaces. ICML 2025 - 42nd International Conference on Machine Learning, Jul 2025, Vancouver (BC), Canada. ⟨hal-05302437⟩
362 Consultations
487 Téléchargements

Altmetric

Partager

  • More