ReAct Meets ActRe: Autonomous Annotations of Agent Trajectories for Contrastive Self-Training

arXiv:2403.14589v1 Announce Type: cross
Abstract: Language agents have demonstrated autonomous decision-making abilities by reasoning with foundation models. Recently, efforts have been made to train language agents for performance improvement, with multi-step reasoning and action trajectories as the training data. However, collecting such trajectories still requires considerable human effort, by either artificial annotations or implementations of diverse prompting frameworks. In this work, we propose A$^3$T, a framework that enables the Autonomous Annotation of Agent Trajectories in the style of ReAct. The …

abstract agent agents annotations arxiv autonomous cs.ai cs.cl cs.lg data decision foundation however human improvement language making performance react reasoning self-training train training training data type