Model Swarms: Collaborative Search to Adapt LLM Experts via Swarm Intelligence
This addresses the challenge of tuning-free model adaptation for LLMs in low-data regimes, offering a novel approach to enhance performance across single and multi-task domains, though it is incremental in the context of existing model composition methods.
The paper tackles the problem of adapting large language models (LLMs) to various tasks by introducing Model Swarms, a collaborative search algorithm based on swarm intelligence, which improves over 12 baselines by up to 21.0% across tasks and works with as few as 200 examples.
We propose Model Swarms, a collaborative search algorithm to adapt LLMs via swarm intelligence, the collective behavior guiding individual systems. Specifically, Model Swarms starts with a pool of LLM experts and a utility function. Guided by the best-found checkpoints across models, diverse LLM experts collaboratively move in the weight space and optimize a utility function representing model adaptation objectives. Compared to existing model composition approaches, Model Swarms offers tuning-free model adaptation, works in low-data regimes with as few as 200 examples, and does not require assumptions about specific experts in the swarm or how they should be composed. Extensive experiments demonstrate that Model Swarms could flexibly adapt LLM experts to a single task, multi-task domains, reward models, as well as diverse human interests, improving over 12 model composition baselines by up to 21.0% across tasks and contexts. Further analysis reveals that LLM experts discover previously unseen capabilities in initial checkpoints and that Model Swarms enable the weak-to-strong transition of experts through the collaborative search process.