Problem
Research questions and friction points this paper is trying to address.
catastrophic forgetting
specialized tasks
long agent trajectories
Agents unlock new capabilities through Switching LoRA Adapters as a Tool (SLAaaT)
🤖 AI Summary
通过切换LoRA适配器工具解决代理在不同任务间能力遗忘问题，测试表明该方法有效提升了模型性能并减少了资源消耗。
📝 Abstract
Post-training can unlock new capabilities and improve performance on specialized tasks, but sometimes at the cost of catastrophic forgetting in other domains. This poses a problem in long agent trajectories that compose different capabilities. We reject this tradeoff by giving an agent a tool to switch between specialized LoRA adapters mid-trace. To test its effectiveness, we compose two synthetic coding tasks that are logically simple but require specialization. We find that this allows the model to solve problems it previously could not, that the model is able to switch autonomously (and find a new strategy that beats our human heuristic baseline on one task), and that this incurs an up to an 18x reduction in capability tax compared to an agent using only one specialized adapter. Our approach also substantially outperforms spawning subagents in both task capabilities and token usage.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Switching LoRA Adapters
post-training
specialized tasks
catastrophic forgetting
dynamic switching
🔎 Similar Papers
