ATLAS : Adaptive Self-Evolutionary Research Agent with Task-Distributed Multi-LLM Supporters

Recent multi-LLM agent systems perform well in prompt optimization and automated problem-solving, but many either keep the solver frozen after fine-tuning or become inefficient due to the increasing size of context windows. Our proposed ATLAS framework introduces a novel self-evolutionary research agent guided by a task-distributed multi-LLM supporter system. This architecture allows for dynamic adaptation and improvement of research strategies, enabling more robust and scalable automated scientific discovery.