CompactionRL: Reinforcement Learning with Context Compaction for Long-Horizon Agents

By Yujiang Li · Paper · cs.LG

Long-horizon agentic LLMs are increasingly limited by finite context windows, as extended interaction trajectories can exceed the maximum context length before a task is completed. Context compaction offers a natural solution by summarizing previous interaction states and continu

AI Agents · Cs.lg

View original

HomeResourceLoading…