Lila Sciences' Taylor W. Killian and academic Prithviraj Ammanabrolu introduce Behavior Cues to steer LLM reasoning and prevent speculative collapse

We're introducing a mechanism by which to enforce some control in a LLMs reasoning process. We developed Behavior Cues to steer models, avoiding both overthinking and speculative collapse. @ccui9 provides a great overview of the work in this thread 👇