Mixtures of Experts Unlock Parameter Scaling for Deep RL
This work addresses the problem of parameter scaling in deep reinforcement learning for researchers and practitioners, offering a method to enable more scalable models, though it is incremental as it builds on existing MoE techniques.
The paper tackles the lack of scaling laws in reinforcement learning by showing that incorporating Mixture-of-Expert modules, specifically Soft MoEs, into value-based networks improves performance with increased model size, achieving substantial gains across various training regimes.
The recent rapid progress in (self) supervised learning models is in large part predicted by empirical scaling laws: a model's performance scales proportionally to its size. Analogous scaling laws remain elusive for reinforcement learning domains, however, where increasing the parameter count of a model often hurts its final performance. In this paper, we demonstrate that incorporating Mixture-of-Expert (MoE) modules, and in particular Soft MoEs (Puigcerver et al., 2023), into value-based networks results in more parameter-scalable models, evidenced by substantial performance increases across a variety of training regimes and model sizes. This work thus provides strong empirical evidence towards developing scaling laws for reinforcement learning.