Lanerra/reasoning-bank-slm

An experiment that applies Google Research's `ReasoningBank` technique to Small Language Models. This experiment hopes to show that the same gains from the ReasoningBank paper also applies to much smaller, less capable models.