LATIDIA · Investigación

MOSCOPT: Optimización Colectiva de la Mezcla de Habilidades para Agentes LLM

arXiv: 2609.14399v1Tipo de anuncio: nuevo Resumen: Las indicaciones y habilidades del lenguaje natural sirven como la columna vertebral estratégica de los agentes basados en LLM. Los avances recientes en la optimización rápida y de habilidades han logrado avances notables, sin embargo, un

La noticia

arXiv:2609.14399v1 Announce Type: new Abstract: Natural language prompts and skills serve as the strategic backbone of LLM-based agents. Recent advances in prompt and skill optimization have achieved notable gains, yet all existing methods optimize a \emph{single} text template---missing the synergy among multiple complementary strategies. We propose MOSCOPT, a text-native, parameter-free algorithm that jointly optimizes a pool of $N$ skills and a gating skill $G$ that dynamically selects $K$ skills per step. To effectively optimize the skills, we build the EditAdam with internally maintained dual states. Through the three-phase interleaved updates with EditAdam, the system monotonically improves without gradient or parameter tuning. Extensive experiments and detailed ablations across 5 benchmarks and