LATIDIA · Investigación

Escaleras del pensamiento: un plan de estudios autoevolutivo de rastros de razonamiento progresivamente simplificados

arXiv: 2609.25643v1Tipo de anuncio: nuevo Resumen: Los modelos de lenguaje grandes (LLM) sobresalen en el razonamiento cuando se escalan a cientos de miles de millones de parámetros, pero los modelos de escala pequeña y mediana siguen siendo razonadores frágiles incluso con conocimientos

La noticia

arXiv:2609.25643v1 Announce Type: new Abstract: Large language models (LLMs) excel at reasoning when scaled to hundreds of billions of parameters, but small- and mid-scale models remain brittle reasoners even with knowledge distillation (KD). We present Ladders-of-Thought (LoT), a framework that improves reasoning by combining progressive question rewrites with a self-evolving curriculum. LoT automatically generates semantically faithful but easier variants of reasoning problems, organizes them into difficulty buckets using step-based measures, and employs a self-evolving bandit scheduler to allocate training adaptively. Evaluated on two reasoning domains, math and multi-hop reasoning, across 1-8B models from different families, LoT consistently improves over KD. It delivers large gains on arithmetic tasks (e.g., +32 percentage