LATIDIA · Investigación

Eficiencia de implementación en el aprendizaje de refuerzo para el razonamiento de modelos de lenguaje grandes: una taxonomía y direcciones futuras

arXiv: 2609.25463v1Tipo de anuncio: nuevo Resumen: El aprendizaje de refuerzo orientado al razonamiento permite que los modelos de lenguaje grandes resuelvan tareas matemáticas, de codificación y otras tareas de varios pasos, pero cambia una parte sustancial de la t

La noticia

arXiv:2609.25463v1 Announce Type: new Abstract: Reasoning-oriented reinforcement learning enables large language models to solve mathematical, coding, and other multi-step tasks, but shifts a substantial portion of the training cost to rollout, where trajectories are generated for policy updates. Efficient rollout mechanisms are therefore essential to reduce this cost while maintaining the freshness, consistency, and statistical validity of training data. This survey provides a systematic taxonomy of recent research on rollout efficiency for reasoning-oriented reinforcement learning, classifying existing approaches from both mechanism and bottleneck perspectives. Based on this taxonomy, we analyze how different technique families address distinct sources of rollout inefficiency, examine opportunities and potential conflicts for combining them, identify gaps