Search papers, labs, and topics across Lattice.
This paper introduces OpenLoopEvolve (OLE), a self-evolution framework for Loop Policies that enables agents to adaptively manage their control experience in long-horizon complex tasks. By treating policies as governable assets with versions and lineages, OLE allows for both online and offline evolution modes, enhancing agents' ability to learn from continuous operation and historical traces. Experimental results on the YC-Bench benchmark demonstrate significant improvements in task performance, success rates, and risk metrics compared to static policies, highlighting the framework's potential for effective policy management in dynamic environments.
Treating Loop Policies as governable assets can significantly enhance agent performance in complex tasks, leading to improved adaptability and efficiency.
Long-horizon complex tasks require agents to repeatedly observe states, formulate plans, invoke tools, verify results, and recover from failures in continuously changing environments. However, such control experience often remains confined to a single context or a fixed prompt, and is difficult to accumulate and reuse across historical traces. This paper presents OpenLoopEvolve (OLE), a self-evolution framework centered on the Loop Policy. OLE represents an agent's observation, planning, memory, action, verification, recovery, stopping, and budget control as portable policy assets with versions and lineages, and provides online and offline evolution modes that can be selected according to practical needs: the online mode triggers candidate generation based on feedback from continuous operation, whereas the offline mode searches for candidate policies from archived traces and failure evidence. Both modes share an evolution mechanism consisting of autonomous proposals by a large language model, Champion--Challenger paired evaluation, and robust release. Policies released online are activated at a subsequent task boundary, monitored using subsequent feedback, and rolled back to their parent versions when degradation conditions are met. On the simulated business benchmark YC-Bench, both modes improve aggregate task performance, task success rate, and risk metrics relative to a fixed initial Loop Policy. The results indicate that treating the Loop Policy as a governable asset can support the accumulation, comparison, release, and reuse of control experience and improve agent performance on long-horizon complex tasks.
