Tree-Based Training Method Improves AI Agents That Follow Instructions

Original: TSBPO: Tree-Structured Branching for Policy Optimization in Interactive Language Agents

Source: arxiv.org ↗