#text-to-sql

Researchers target sparse rewards in agentic RL

New arXiv papers propose self-distillation methods for denser credit assignment in LLM agents.