CORTEXA
← Browse

Yi Jing

2 papers indexed

arxivcs.LG2026-07-06

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

Yujiang Li, Zhenyu Hou, Yi Jing, Jie Tang, Yuxiao Dong

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…

View free PDFSource page