1 paper · 1 filter
Ivan Rodkin, Daniil Orel, Konstantin Smirnov +9
Reasoning is a core capability of large language models, yet how multi-step reasoning is learned and executed remains unclear. We study this question in a controlled cellular-autom…