Lessons Learned: A Multi-Agent Framework for Code LLMs to Learn and Improve
This addresses the challenge of improving code optimization performance for developers by enabling efficient collaboration among specialized LLMs, though it is incremental as it builds on existing multi-agent concepts.
The paper tackles the problem of leveraging multiple code LLMs with complementary strengths by proposing a lesson-based collaboration framework, where agents learn from each other's successes and failures, and demonstrates that a team of small LLMs with this approach outperforms a larger LLM and other multi-LLM methods.
Recent studies show that LLMs possess different skills and specialize in different tasks. In fact, we observe that their varied performance occur in several levels of granularity. For example, in the code optimization task, code LLMs excel at different optimization categories and no one dominates others. This observation prompts the question of how one leverages multiple LLM agents to solve a coding problem without knowing their complementary strengths a priori. We argue that a team of agents can learn from each other's successes and failures so as to improve their own performance. Thus, a lesson is the knowledge produced by an agent and passed on to other agents in the collective solution process. We propose a lesson-based collaboration framework, design the lesson solicitation--banking--selection mechanism, and demonstrate that a team of small LLMs with lessons learned can outperform a much larger LLM and other multi-LLM collaboration methods.